rpms/glusterfs
7
0
Fork 0
Code Issues Pull Requests Releases Activity

import glusterfs-6.0-56.4.el8

Browse Source
This commit is contained in:
CentOS Sources 2021-10-06 10:56:09 -04:00 committed by Stepan Oksanichenko
parent aff8de6339
commit f6faf97671
163 changed files with 31168 additions and 126 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
SOURCES/0384-Update-rfc.sh-to-rhgs-3.5.3.patch Normal file
View File

@ -0,0 +1,26 @@
From 27dc773af276e33fcca10788fae17d131c8d9bce Mon Sep 17 00:00:00 2001
From: Rinku Kothiya <rkothiya@redhat.com>
Date: Sun, 31 May 2020 15:46:24 -0400
Subject: [PATCH 384/449] Update rfc.sh to rhgs-3.5.3
Signed-off-by: Rinku Kothiya <rkothiya@redhat.com>
---
rfc.sh | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/rfc.sh b/rfc.sh
index 37d551f..1dca29f 100755
--- a/rfc.sh
+++ b/rfc.sh
@@ -18,7 +18,7 @@ done
shift $((OPTIND-1))
-branch="rhgs-3.5.2";
+branch="rhgs-3.5.3";
set_hooks_commit_msg()
{
--
1.8.3.1

50
SOURCES/0385-glusterd-start-glusterd-automatically-on-abnormal-sh.patch Normal file
View File

@ -0,0 +1,50 @@
From 143f85f55ded7a9075408e97d05abd9568d56e7b Mon Sep 17 00:00:00 2001
From: Sanju Rakonde <srakonde@redhat.com>
Date: Mon, 25 Nov 2019 16:35:42 +0530
Subject: [PATCH 385/449] glusterd: start glusterd automatically on abnormal
shutdown
If glusterd crashes or goes down abnormally, systemd should
automatically bring the glusterd up.
With this change, systemd brings glusterd up for atmost 3 times
within time period of 1 hour. If the limit exceeds, we have to
start the glusterd manually and reset the failure count using
systemctl reset-failed.
credits: John Strunk <jstrunk@redhat.com>
> upstream patch link: https://review.gluster.org/#/c/glusterfs/+/23751/
> fixes: bz#1776264
> Change-Id: I312d243652fb13ba028814a2ea615b67e3b10b6a
> Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
BUG: 1663557
Change-Id: I312d243652fb13ba028814a2ea615b67e3b10b6a
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202251
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
extras/systemd/glusterd.service.in | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/extras/systemd/glusterd.service.in b/extras/systemd/glusterd.service.in
index f604160..b944762 100644
--- a/extras/systemd/glusterd.service.in
+++ b/extras/systemd/glusterd.service.in
@@ -15,6 +15,11 @@ ExecStart=@prefix@/sbin/glusterd -p @localstatedir@/run/glusterd.pid --log-leve
KillMode=process
TimeoutSec=300
SuccessExitStatus=15
+Restart=on-abnormal
+RestartSec=60
+StartLimitBurst=3
+StartLimitIntervalSec=3600
+StartLimitInterval=3600
[Install]
WantedBy=multi-user.target
--
1.8.3.1

39
SOURCES/0386-glusterd-increase-the-StartLimitBurst.patch Normal file
View File

@ -0,0 +1,39 @@
From 02e7afdfb740db7cfa1a2f0f79933172d172ff27 Mon Sep 17 00:00:00 2001
From: Sanju Rakonde <srakonde@redhat.com>
Date: Tue, 7 Jan 2020 15:32:13 +0530
Subject: [PATCH 386/449] glusterd: increase the StartLimitBurst
Based on https://bugzilla.redhat.com/show_bug.cgi?id=1782200#c6
increasing the limit.
> upstream patch link: https://review.gluster.org/#/c/glusterfs/+/23970/
> fixes: bz#1782200
> Change-Id: Ia885c7bdb2a90f0946c5268da894f6a4da5a69b7
> Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
BUG: 1663557
Change-Id: Ia885c7bdb2a90f0946c5268da894f6a4da5a69b7
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202252
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
extras/systemd/glusterd.service.in | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/extras/systemd/glusterd.service.in b/extras/systemd/glusterd.service.in
index b944762..699aea3 100644
--- a/extras/systemd/glusterd.service.in
+++ b/extras/systemd/glusterd.service.in
@@ -17,7 +17,7 @@ TimeoutSec=300
SuccessExitStatus=15
Restart=on-abnormal
RestartSec=60
-StartLimitBurst=3
+StartLimitBurst=6
StartLimitIntervalSec=3600
StartLimitInterval=3600
--
1.8.3.1

47
SOURCES/0387-To-fix-readdir-ahead-memory-leak.patch Normal file
View File

@ -0,0 +1,47 @@
From d54f087a2484695ff7ac214d39f2750fddcef2d5 Mon Sep 17 00:00:00 2001
From: HuangShujun <549702281@qq.com>
Date: Thu, 5 Dec 2019 10:07:10 +0200
Subject: [PATCH 387/449] To fix readdir-ahead memory leak
Glusterfs client process has memory leak if create several files under
one folder, and delete the folder. According to statedump, the ref
counts of readdir-ahead is bigger than zero in the inode table.
Readdir-ahead get parent inode by inode_parent in rda_mark_inode_dirty
when each rda_writev_cbk,the inode ref count of parent folder will be
increased in inode_parent, but readdir-ahead do not unref it later.
The correction is unref the parent inode at the end of
rda_mark_inode_dirty.
Upstream patch:
> Upstream patch link: https://review.gluster.org/c/glusterfs/+/23815
> Fixes: bz#1779055
> Signed-off-by: HuangShujun <549702281@qq.com>
> Change-Id: Iee68ab1089cbc2fbc4185b93720fb1f66ee89524
BUG: 1781550
Change-Id: Iee68ab1089cbc2fbc4185b93720fb1f66ee89524
Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202312
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/performance/readdir-ahead/src/readdir-ahead.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/xlators/performance/readdir-ahead/src/readdir-ahead.c b/xlators/performance/readdir-ahead/src/readdir-ahead.c
index 7fd4f8d..933941d 100644
--- a/xlators/performance/readdir-ahead/src/readdir-ahead.c
+++ b/xlators/performance/readdir-ahead/src/readdir-ahead.c
@@ -254,6 +254,7 @@ rda_mark_inode_dirty(xlator_t *this, inode_t *inode)
}
}
UNLOCK(&parent->lock);
+ inode_unref(parent);
}
return;
--
1.8.3.1

8
SOURCES/0388-rpc-Cleanup-SSL-specific-data-at-the-time-of-freeing.patch
View File

@ -1,7 +1,7 @@
From be6fafebe1e391e9d9f14d9aed18adbfda8a262b Mon Sep 17 00:00:00 2001
From fbda9baaf7231e3237277348cc7e873f3113fd14 Mon Sep 17 00:00:00 2001
From: l17zhou <cynthia.zhou@nokia-sbell.com.cn>
Date: Mon, 4 Nov 2019 08:45:52 +0200
Subject: [PATCH 388/392] rpc: Cleanup SSL specific data at the time of freeing
Subject: [PATCH 388/449] rpc: Cleanup SSL specific data at the time of freeing
rpc object
Problem: At the time of cleanup rpc object ssl specific data
@ -17,9 +17,9 @@ Solution: To avoid the leak cleanup ssl specific data at the
> (Reviewed on upstream link https://review.gluster.org/#/c/glusterfs/+/23650/)
Change-Id: I37f598673ae2d7a33c75f39eb8843ccc6dffaaf0
BUG: 1848891
BUG: 1786516
Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/203698
Reviewed-on: https://code.engineering.redhat.com/gerrit/202308
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---

297
SOURCES/0389-posix-Avoid-diskpace-error-in-case-of-overwriting-th.patch Normal file
View File

@ -0,0 +1,297 @@
From 50318713486e79d9258cf22e656caff402256dde Mon Sep 17 00:00:00 2001
From: Mohit Agrawal <moagrawal@redhat.com>
Date: Sun, 20 Oct 2019 22:01:01 +0530
Subject: [PATCH 389/449] posix: Avoid diskpace error in case of overwriting
the data
Problem: Sometime fops like posix_writev, posix_fallocate, posix_zerofile
failed and throw error ENOSPC if storage.reserve threshold limit
has reached even fops is overwriting the data
Solution: Retry the fops in case of overwrite if diskspace check
is failed
> Credits: kinsu <vpolakis@gmail.com>
> Change-Id: I987d73bcf47ed1bb27878df40c39751296e95fe8
> Updates: #745
> Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
> (Cherry pick from commit ca3e5905ac02fb9c373ac3de10b44f061d04cd6f)
> (Reviewed on upstream link https://review.gluster.org/#/c/glusterfs/+/23572/)
Change-Id: I987d73bcf47ed1bb27878df40c39751296e95fe8
BUG: 1787331
Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202307
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/bugs/posix/bug-1651445.t | 1 +
xlators/storage/posix/src/posix-entry-ops.c | 1 -
xlators/storage/posix/src/posix-inode-fd-ops.c | 141 ++++++++++++++++++++++---
3 files changed, 126 insertions(+), 17 deletions(-)
diff --git a/tests/bugs/posix/bug-1651445.t b/tests/bugs/posix/bug-1651445.t
index 5248d47..4d08b69 100644
--- a/tests/bugs/posix/bug-1651445.t
+++ b/tests/bugs/posix/bug-1651445.t
@@ -33,6 +33,7 @@ sleep 5
# setup_lvm create lvm partition of 150M and 40M are reserve so after
# consuming more than 110M next dd should fail
TEST ! dd if=/dev/zero of=$M0/c bs=5M count=1
+TEST dd if=/dev/urandom of=$M0/a bs=1022 count=1 oflag=seek_bytes,sync seek=102 conv=notrunc
rm -rf $M0/*
diff --git a/xlators/storage/posix/src/posix-entry-ops.c b/xlators/storage/posix/src/posix-entry-ops.c
index 283b305..bea0bbf 100644
--- a/xlators/storage/posix/src/posix-entry-ops.c
+++ b/xlators/storage/posix/src/posix-entry-ops.c
@@ -1634,7 +1634,6 @@ posix_rename(call_frame_t *frame, xlator_t *this, loc_t *oldloc, loc_t *newloc,
priv = this->private;
VALIDATE_OR_GOTO(priv, out);
- DISK_SPACE_CHECK_AND_GOTO(frame, priv, xdata, op_ret, op_errno, out);
SET_FS_ID(frame->root->uid, frame->root->gid);
MAKE_ENTRY_HANDLE(real_oldpath, par_oldpath, this, oldloc, NULL);
diff --git a/xlators/storage/posix/src/posix-inode-fd-ops.c b/xlators/storage/posix/src/posix-inode-fd-ops.c
index a2a518f..bcce06e 100644
--- a/xlators/storage/posix/src/posix-inode-fd-ops.c
+++ b/xlators/storage/posix/src/posix-inode-fd-ops.c
@@ -692,6 +692,10 @@ posix_do_fallocate(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t flags,
gf_boolean_t locked = _gf_false;
posix_inode_ctx_t *ctx = NULL;
struct posix_private *priv = NULL;
+ gf_boolean_t check_space_error = _gf_false;
+ struct stat statbuf = {
+ 0,
+ };
DECLARE_OLD_FS_ID_VAR;
@@ -711,7 +715,10 @@ posix_do_fallocate(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t flags,
if (priv->disk_reserve)
posix_disk_space_check(this);
- DISK_SPACE_CHECK_AND_GOTO(frame, priv, xdata, ret, ret, out);
+ DISK_SPACE_CHECK_AND_GOTO(frame, priv, xdata, ret, ret, unlock);
+
+overwrite:
+ check_space_error = _gf_true;
ret = posix_fd_ctx_get(fd, this, &pfd, &op_errno);
if (ret < 0) {
@@ -735,7 +742,7 @@ posix_do_fallocate(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t flags,
ret = -errno;
gf_msg(this->name, GF_LOG_ERROR, errno, P_MSG_FSTAT_FAILED,
"fallocate (fstat) failed on fd=%p", fd);
- goto out;
+ goto unlock;
}
if (xdata) {
@@ -745,7 +752,7 @@ posix_do_fallocate(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t flags,
gf_msg(this->name, GF_LOG_ERROR, 0, 0,
"file state check failed, fd %p", fd);
ret = -EIO;
- goto out;
+ goto unlock;
}
}
@@ -756,7 +763,7 @@ posix_do_fallocate(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t flags,
"fallocate failed on %s offset: %jd, "
"len:%zu, flags: %d",
uuid_utoa(fd->inode->gfid), offset, len, flags);
- goto out;
+ goto unlock;
}
ret = posix_fdstat(this, fd->inode, pfd->fd, statpost);
@@ -764,16 +771,47 @@ posix_do_fallocate(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t flags,
ret = -errno;
gf_msg(this->name, GF_LOG_ERROR, errno, P_MSG_FSTAT_FAILED,
"fallocate (fstat) failed on fd=%p", fd);
- goto out;
+ goto unlock;
}
posix_set_ctime(frame, this, NULL, pfd->fd, fd->inode, statpost);
-out:
+unlock:
if (locked) {
pthread_mutex_unlock(&ctx->write_atomic_lock);
locked = _gf_false;
}
+
+ if (op_errno == ENOSPC && priv->disk_space_full && !check_space_error) {
+#ifdef FALLOC_FL_KEEP_SIZE
+ if (flags & FALLOC_FL_KEEP_SIZE) {
+ goto overwrite;
+ }
+#endif
+ ret = posix_fd_ctx_get(fd, this, &pfd, &op_errno);
+ if (ret < 0) {
+ gf_msg(this->name, GF_LOG_WARNING, ret, P_MSG_PFD_NULL,
+ "pfd is NULL from fd=%p", fd);
+ goto out;
+ }
+
+ if (sys_fstat(pfd->fd, &statbuf) < 0) {
+ gf_msg(this->name, GF_LOG_WARNING, op_errno, P_MSG_FILE_OP_FAILED,
+ "%d", pfd->fd);
+ goto out;
+ }
+
+ if (offset + len <= statbuf.st_size) {
+ gf_msg_debug(this->name, 0,
+ "io vector size will not"
+ " change disk size so allow overwrite for"
+ " fd %d",
+ pfd->fd);
+ goto overwrite;
+ }
+ }
+
+out:
SET_TO_OLD_FS_ID();
if (ret == ENOSPC)
ret = -ENOSPC;
@@ -1083,25 +1121,57 @@ posix_zerofill(call_frame_t *frame, xlator_t *this, fd_t *fd, off_t offset,
int op_ret = -1;
int op_errno = EINVAL;
dict_t *rsp_xdata = NULL;
+ gf_boolean_t check_space_error = _gf_false;
+ struct posix_fd *pfd = NULL;
+ struct stat statbuf = {
+ 0,
+ };
- VALIDATE_OR_GOTO(frame, out);
- VALIDATE_OR_GOTO(this, out);
+ VALIDATE_OR_GOTO(frame, unwind);
+ VALIDATE_OR_GOTO(this, unwind);
priv = this->private;
DISK_SPACE_CHECK_AND_GOTO(frame, priv, xdata, op_ret, op_errno, out);
+overwrite:
+ check_space_error = _gf_true;
ret = posix_do_zerofill(frame, this, fd, offset, len, &statpre, &statpost,
xdata, &rsp_xdata);
if (ret < 0) {
op_ret = -1;
op_errno = -ret;
- goto out;
+ goto unwind;
}
STACK_UNWIND_STRICT(zerofill, frame, 0, 0, &statpre, &statpost, rsp_xdata);
return 0;
out:
+ if (op_errno == ENOSPC && priv->disk_space_full && !check_space_error) {
+ ret = posix_fd_ctx_get(fd, this, &pfd, &op_errno);
+ if (ret < 0) {
+ gf_msg(this->name, GF_LOG_WARNING, ret, P_MSG_PFD_NULL,
+ "pfd is NULL from fd=%p", fd);
+ goto out;
+ }
+
+ if (sys_fstat(pfd->fd, &statbuf) < 0) {
+ gf_msg(this->name, GF_LOG_WARNING, op_errno, P_MSG_FILE_OP_FAILED,
+ "%d", pfd->fd);
+ goto out;
+ }
+
+ if (offset + len <= statbuf.st_size) {
+ gf_msg_debug(this->name, 0,
+ "io vector size will not"
+ " change disk size so allow overwrite for"
+ " fd %d",
+ pfd->fd);
+ goto overwrite;
+ }
+ }
+
+unwind:
STACK_UNWIND_STRICT(zerofill, frame, op_ret, op_errno, NULL, NULL,
rsp_xdata);
return 0;
@@ -1857,19 +1927,28 @@ posix_writev(call_frame_t *frame, xlator_t *this, fd_t *fd,
gf_boolean_t write_append = _gf_false;
gf_boolean_t update_atomic = _gf_false;
posix_inode_ctx_t *ctx = NULL;
+ gf_boolean_t check_space_error = _gf_false;
+ struct stat statbuf = {
+ 0,
+ };
+ int totlen = 0;
+ int idx = 0;
- VALIDATE_OR_GOTO(frame, out);
- VALIDATE_OR_GOTO(this, out);
- VALIDATE_OR_GOTO(fd, out);
- VALIDATE_OR_GOTO(fd->inode, out);
- VALIDATE_OR_GOTO(vector, out);
- VALIDATE_OR_GOTO(this->private, out);
+ VALIDATE_OR_GOTO(frame, unwind);
+ VALIDATE_OR_GOTO(this, unwind);
+ VALIDATE_OR_GOTO(fd, unwind);
+ VALIDATE_OR_GOTO(fd->inode, unwind);
+ VALIDATE_OR_GOTO(vector, unwind);
+ VALIDATE_OR_GOTO(this->private, unwind);
priv = this->private;
- VALIDATE_OR_GOTO(priv, out);
+ VALIDATE_OR_GOTO(priv, unwind);
DISK_SPACE_CHECK_AND_GOTO(frame, priv, xdata, op_ret, op_errno, out);
+overwrite:
+
+ check_space_error = _gf_true;
if ((fd->inode->ia_type == IA_IFBLK) || (fd->inode->ia_type == IA_IFCHR)) {
gf_msg(this->name, GF_LOG_ERROR, EINVAL, P_MSG_INVALID_ARGUMENT,
"writev received on a block/char file (%s)",
@@ -2011,6 +2090,36 @@ out:
locked = _gf_false;
}
+ if (op_errno == ENOSPC && priv->disk_space_full && !check_space_error) {
+ ret = posix_fd_ctx_get(fd, this, &pfd, &op_errno);
+ if (ret < 0) {
+ gf_msg(this->name, GF_LOG_WARNING, ret, P_MSG_PFD_NULL,
+ "pfd is NULL from fd=%p", fd);
+ goto unwind;
+ }
+
+ if (sys_fstat(pfd->fd, &statbuf) < 0) {
+ gf_msg(this->name, GF_LOG_WARNING, op_errno, P_MSG_FILE_OP_FAILED,
+ "%d", pfd->fd);
+ goto unwind;
+ }
+
+ for (idx = 0; idx < count; idx++) {
+ totlen = vector[idx].iov_len;
+ }
+
+ if ((offset + totlen <= statbuf.st_size) &&
+ !(statbuf.st_blocks * statbuf.st_blksize < statbuf.st_size)) {
+ gf_msg_debug(this->name, 0,
+ "io vector size will not"
+ " change disk size so allow overwrite for"
+ " fd %d",
+ pfd->fd);
+ goto overwrite;
+ }
+ }
+
+unwind:
STACK_UNWIND_STRICT(writev, frame, op_ret, op_errno, &preop, &postop,
rsp_xdata);
--
1.8.3.1

93
SOURCES/0390-glusterd-deafult-options-after-volume-reset.patch Normal file
View File

@ -0,0 +1,93 @@
From 86df0ced1cac0e3c48f6149bb2f5442f8548f89e Mon Sep 17 00:00:00 2001
From: Sanju Rakonde <srakonde@redhat.com>
Date: Wed, 25 Dec 2019 21:56:32 +0530
Subject: [PATCH 390/449] glusterd: deafult options after volume reset
Problem: default option itransport.address-family is disappered
in volume info output after a volume reset.
Cause: with 3.8.0 onwards volume option transport.address-family
has default value, any volume which is created will have this
option set. So, volume info will show this in its output. But,
with reset volume, this option is not handled.
Solution: In glusterd_enable_default_options(), we should add this
option along with other default options. This function is called
by glusterd_options_reset() with volume reset command.
> upstream patch link: https://review.gluster.org/#/c/glusterfs/+/23921/
> fixes: bz#1786478
> Change-Id: I58f7aa24cf01f308c4efe6cae748cc3bc8b99b1d
> Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
BUG: 1781710
Change-Id: I58f7aa24cf01f308c4efe6cae748cc3bc8b99b1d
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202258
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/bugs/glusterd/optimized-basic-testcases.t | 5 +++++
xlators/mgmt/glusterd/src/glusterd-utils.c | 23 +++++++++++++++++++++++
2 files changed, 28 insertions(+)
diff --git a/tests/bugs/glusterd/optimized-basic-testcases.t b/tests/bugs/glusterd/optimized-basic-testcases.t
index d700b5e..c7e8c32 100644
--- a/tests/bugs/glusterd/optimized-basic-testcases.t
+++ b/tests/bugs/glusterd/optimized-basic-testcases.t
@@ -69,6 +69,11 @@ TEST pidof glusterd;
TEST $CLI volume create $V0 $H0:$B0/${V0}{1,2};
EXPECT 'Created' volinfo_field $V0 'Status';
+#bug-1786478 - default volume option after volume reset
+addr_family=`volinfo_field $V0 'transport.address-family'`
+TEST $CLI volume reset $V0
+EXPECT $addr_family volinfo_field $V0 'transport.address-family'
+
#bug-955588 - uuid validation
uuid=`grep UUID $GLUSTERD_WORKDIR/glusterd.info | cut -f2 -d=`
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index c92cdf3..6654741 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -13032,6 +13032,11 @@ glusterd_enable_default_options(glusterd_volinfo_t *volinfo, char *option)
int ret = 0;
xlator_t *this = NULL;
glusterd_conf_t *conf = NULL;
+#ifdef IPV6_DEFAULT
+ char *addr_family = "inet6";
+#else
+ char *addr_family = "inet";
+#endif
this = THIS;
GF_ASSERT(this);
@@ -13109,6 +13114,24 @@ glusterd_enable_default_options(glusterd_volinfo_t *volinfo, char *option)
}
}
}
+
+ if (conf->op_version >= GD_OP_VERSION_3_9_0) {
+ if (!option || !strcmp("transport.address-family", option)) {
+ if (volinfo->transport_type == GF_TRANSPORT_TCP) {
+ ret = dict_set_dynstr_with_alloc(
+ volinfo->dict, "transport.address-family", addr_family);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, errno,
+ GD_MSG_DICT_SET_FAILED,
+ "failed to set transport."
+ "address-family on %s",
+ volinfo->volname);
+ goto out;
+ }
+ }
+ }
+ }
+
if (conf->op_version >= GD_OP_VERSION_7_0) {
ret = dict_set_dynstr_with_alloc(volinfo->dict,
"storage.fips-mode-rchecksum", "on");
--
1.8.3.1

51
SOURCES/0390-tests-Avoid-ssl-authz.t-failure.patch
View File

@ -1,51 +0,0 @@
From b68fa363c5981441c20fbc78b6dc00437bd698a7 Mon Sep 17 00:00:00 2001
From: Mohit Agrawal <moagrawa@redhat.com>
Date: Mon, 22 Jun 2020 11:35:29 +0530
Subject: [PATCH 390/392] tests: Avoid ssl-authz.t failure
Problem: ssl-authz.t is failing at the time of checking memory
consumption if brick is consuming more than 5M
Solution: Update the check to avoid a failure.
> Change-Id: Iffb031f0695a7da83d5a2f6bac8863dad225317e
> Fixes: bz#1811631
> Signed-off-by: Mohit Agrawal <moagrawal@redhat.com>
> Reviewd on upstream link https://review.gluster.org/#/c/glusterfs/+/24221/)
> (Cherry pick from commit fb20713b380e1df8d7f9e9df96563be2f9144fd6)
BUG: 1848894
Change-Id: I4fc5d2e2597abfafc1e26d908c8c4184ab82afd5
Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/203844
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/features/ssl-authz.t | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/tests/features/ssl-authz.t b/tests/features/ssl-authz.t
index 132b598..ab05c49 100755
--- a/tests/features/ssl-authz.t
+++ b/tests/features/ssl-authz.t
@@ -67,13 +67,15 @@ echo "Memory consumption for glusterfsd process"
for i in $(seq 1 100); do
gluster v heal $V0 info >/dev/null
done
+#Wait to cleanup memory
+sleep 10
end=`pmap -x $glusterfsd_pid | grep total | awk -F " " '{print $4}'`
diff=$((end-start))
-# If memory consumption is more than 5M some leak in SSL code path
+# If memory consumption is more than 15M some leak in SSL code path
-TEST [ $diff -lt 5000 ]
+TEST [ $diff -lt 15000 ]
# Set ssl-allow to a wildcard that includes our identity.
--
1.8.3.1

10
SOURCES/0392-glusterd-unlink-the-file-after-killing-the-process.patch → SOURCES/0391-glusterd-unlink-the-file-after-killing-the-process.patch
View File

@ -1,20 +1,20 @@
From 6a2cef546457e6dc9a2268cc2f5cc11b850b7f5c Mon Sep 17 00:00:00 2001
From d23859d5cbd5823b2587811aa57030436ce9e74c Mon Sep 17 00:00:00 2001
From: Sanju Rakonde <srakonde@redhat.com>
Date: Tue, 17 Dec 2019 15:52:30 +0530
Subject: [PATCH 392/392] glusterd: unlink the file after killing the process
Subject: [PATCH 391/449] glusterd: unlink the file after killing the process
In glusterd_proc_stop(), after killing the pid
we should remove the pidfile.
> upstream patch: https://review.gluster.org/#/c/glusterfs/+/23890/
> upstream patch link: https://review.gluster.org/#/c/glusterfs/+/23890/
> fixes: bz#1784375
> Change-Id: Ib6367aed590932c884b0f6f892fc40542aa19686
> Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
BUG: 1849533
BUG: 1784211
Change-Id: Ib6367aed590932c884b0f6f892fc40542aa19686
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/203871
Reviewed-on: https://code.engineering.redhat.com/gerrit/202257
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---

187
SOURCES/0392-glusterd-Brick-process-fails-to-come-up-with-brickmu.patch Normal file
View File

@ -0,0 +1,187 @@
From a30a5fdef2e252eba9f44a3c671de8f3aa4f17d7 Mon Sep 17 00:00:00 2001
From: Vishal Pandey <vpandey@redhat.com>
Date: Tue, 19 Nov 2019 11:39:22 +0530
Subject: [PATCH 392/449] glusterd: Brick process fails to come up with
brickmux on
Issue:
1- In a cluster of 3 Nodes N1, N2, N3. Create 3 volumes vol1,
vol2, vol3 with 3 bricks (one from each node)
2- Set cluster.brick-multiplex on
3- Start all 3 volumes
4- Check if all bricks on a node are running on same port
5- Kill N1
6- Set performance.readdir-ahead for volumes vol1, vol2, vol3
7- Bring N1 up and check volume status
8- All bricks processes not running on N1.
Root Cause -
Since, There is a diff in volfile versions in N1 as compared
to N2 and N3 therefore glusterd_import_friend_volume() is called.
glusterd_import_friend_volume() copies the new_volinfo and deletes
old_volinfo and then calls glusterd_start_bricks().
glusterd_start_bricks() looks for the volfiles and sends an rpc
request to glusterfs_handle_attach(). Now, since the volinfo
has been deleted by glusterd_delete_stale_volume()
from priv->volumes list before glusterd_start_bricks() and
glusterd_create_volfiles_and_notify_services() and
glusterd_list_add_order is called after glusterd_start_bricks(),
therefore the attach RPC req gets an empty volfile path
and that causes the brick to crash.
Fix- Call glusterd_list_add_order() and
glusterd_create_volfiles_and_notify_services before
glusterd_start_bricks() cal is made in glusterd_import_friend_volume
> upstream patch link: https://review.gluster.org/#/c/glusterfs/+/23724/
> Change-Id: Idfe0e8710f7eb77ca3ddfa1cabeb45b2987f41aa
> Fixes: bz#1773856
> Signed-off-by: Mohammed Rafi KC <rkavunga@redhat.com>
BUG: 1683602
Change-Id: Idfe0e8710f7eb77ca3ddfa1cabeb45b2987f41aa
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202255
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
.../glusterd/brick-mux-validation-in-cluster.t | 61 +++++++++++++++++++++-
xlators/mgmt/glusterd/src/glusterd-utils.c | 28 +++++-----
2 files changed, 75 insertions(+), 14 deletions(-)
diff --git a/tests/bugs/glusterd/brick-mux-validation-in-cluster.t b/tests/bugs/glusterd/brick-mux-validation-in-cluster.t
index 4e57038..f088dbb 100644
--- a/tests/bugs/glusterd/brick-mux-validation-in-cluster.t
+++ b/tests/bugs/glusterd/brick-mux-validation-in-cluster.t
@@ -7,6 +7,20 @@ function count_brick_processes {
pgrep glusterfsd | wc -l
}
+function count_brick_pids {
+ $CLI_1 --xml volume status all | sed -n '/.*<pid>\([^<]*\).*/s//\1/p' \
+ | grep -v "N/A" | sort | uniq | wc -l
+}
+
+function count_N/A_brick_pids {
+ $CLI_1 --xml volume status all | sed -n '/.*<pid>\([^<]*\).*/s//\1/p' \
+ | grep -- '\-1' | sort | uniq | wc -l
+}
+
+function check_peers {
+ $CLI_2 peer status | grep 'Peer in Cluster (Connected)' | wc -l
+}
+
cleanup;
TEST launch_cluster 3
@@ -48,4 +62,49 @@ TEST $CLI_1 volume stop $V1
EXPECT 3 count_brick_processes
-cleanup
+TEST $CLI_1 volume stop $META_VOL
+
+TEST $CLI_1 volume delete $META_VOL
+TEST $CLI_1 volume delete $V0
+TEST $CLI_1 volume delete $V1
+
+#bug-1773856 - Brick process fails to come up with brickmux on
+
+TEST $CLI_1 volume create $V0 $H1:$B1/${V0}1 $H2:$B2/${V0}1 $H3:$B3/${V0}1 force
+TEST $CLI_1 volume start $V0
+
+
+EXPECT 3 count_brick_processes
+
+#create and start a new volume
+TEST $CLI_1 volume create $V1 $H1:$B1/${V1}2 $H2:$B2/${V1}2 $H3:$B3/${V1}2 force
+TEST $CLI_1 volume start $V1
+
+EXPECT 3 count_brick_processes
+
+V2=patchy2
+TEST $CLI_1 volume create $V2 $H1:$B1/${V2}3 $H2:$B2/${V2}3 $H3:$B3/${V2}3 force
+TEST $CLI_1 volume start $V2
+
+EXPECT 3 count_brick_processes
+
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT 3 count_brick_pids
+
+TEST kill_node 1
+
+sleep 10
+
+EXPECT_WITHIN $PROBE_TIMEOUT 1 check_peers;
+
+$CLI_2 volume set $V0 performance.readdir-ahead on
+$CLI_2 volume set $V1 performance.readdir-ahead on
+
+TEST $glusterd_1;
+
+sleep 10
+
+EXPECT 4 count_brick_processes
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT 4 count_brick_pids
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT 0 count_N/A_brick_pids
+
+cleanup;
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index 6654741..1b78812 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -4988,16 +4988,6 @@ glusterd_import_friend_volume(dict_t *peer_data, int count)
glusterd_volinfo_unref(old_volinfo);
}
- if (glusterd_is_volume_started(new_volinfo)) {
- (void)glusterd_start_bricks(new_volinfo);
- if (glusterd_is_snapd_enabled(new_volinfo)) {
- svc = &(new_volinfo->snapd.svc);
- if (svc->manager(svc, new_volinfo, PROC_START_NO_WAIT)) {
- gf_event(EVENT_SVC_MANAGER_FAILED, "svc_name=%s", svc->name);
- }
- }
- }
-
ret = glusterd_store_volinfo(new_volinfo, GLUSTERD_VOLINFO_VER_AC_NONE);
if (ret) {
gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_VOLINFO_STORE_FAIL,
@@ -5007,19 +4997,31 @@ glusterd_import_friend_volume(dict_t *peer_data, int count)
goto out;
}
- ret = glusterd_create_volfiles_and_notify_services(new_volinfo);
+ ret = glusterd_create_volfiles(new_volinfo);
if (ret)
goto out;
+ glusterd_list_add_order(&new_volinfo->vol_list, &priv->volumes,
+ glusterd_compare_volume_name);
+
+ if (glusterd_is_volume_started(new_volinfo)) {
+ (void)glusterd_start_bricks(new_volinfo);
+ if (glusterd_is_snapd_enabled(new_volinfo)) {
+ svc = &(new_volinfo->snapd.svc);
+ if (svc->manager(svc, new_volinfo, PROC_START_NO_WAIT)) {
+ gf_event(EVENT_SVC_MANAGER_FAILED, "svc_name=%s", svc->name);
+ }
+ }
+ }
+
ret = glusterd_import_quota_conf(peer_data, count, new_volinfo, "volume");
if (ret) {
gf_event(EVENT_IMPORT_QUOTA_CONF_FAILED, "volume=%s",
new_volinfo->volname);
goto out;
}
- glusterd_list_add_order(&new_volinfo->vol_list, &priv->volumes,
- glusterd_compare_volume_name);
+ ret = glusterd_fetchspec_notify(this);
out:
gf_msg_debug("glusterd", 0, "Returning with ret: %d", ret);
return ret;
--
1.8.3.1

129
SOURCES/0393-afr-restore-timestamp-of-files-during-metadata-heal.patch Normal file
View File

@ -0,0 +1,129 @@
From b528c21e6fedc9ac841942828b82e0c808da5efb Mon Sep 17 00:00:00 2001
From: Sheetal Pamecha <spamecha@redhat.com>
Date: Thu, 2 Jan 2020 12:05:12 +0530
Subject: [PATCH 393/449] afr: restore timestamp of files during metadata heal
For files: During metadata heal, we restore timestamps
only for non-regular (char, block etc.) files.
Extenting it for regular files as timestamp is updated
via touch command also
> upstream patch link: https://review.gluster.org/#/c/glusterfs/+/23953/
> fixes: bz#1787274
> Change-Id: I26fe4fb6dff679422ba4698a7f828bf62ca7ca18
> Signed-off-by: Sheetal Pamecha <spamecha@redhat.com>
BUG: 1761531
Change-Id: I26fe4fb6dff679422ba4698a7f828bf62ca7ca18
Signed-off-by: Sheetal Pamecha <spamecha@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202332
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
.../bug-1761531-metadata-heal-restore-time.t | 74 ++++++++++++++++++++++
xlators/cluster/afr/src/afr-self-heal-metadata.c | 8 +--
2 files changed, 76 insertions(+), 6 deletions(-)
create mode 100644 tests/bugs/replicate/bug-1761531-metadata-heal-restore-time.t
diff --git a/tests/bugs/replicate/bug-1761531-metadata-heal-restore-time.t b/tests/bugs/replicate/bug-1761531-metadata-heal-restore-time.t
new file mode 100644
index 0000000..7e24eae
--- /dev/null
+++ b/tests/bugs/replicate/bug-1761531-metadata-heal-restore-time.t
@@ -0,0 +1,74 @@
+#!/bin/bash
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../afr.rc
+cleanup
+
+GET_MDATA_PATH=$(dirname $0)/../../utils
+build_tester $GET_MDATA_PATH/get-mdata-xattr.c
+
+TEST glusterd
+TEST pidof glusterd
+TEST $CLI volume create $V0 replica 3 $H0:$B0/brick{0..2}
+TEST $CLI volume start $V0
+TEST $GFS --volfile-id=/$V0 --volfile-server=$H0 $M0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 1
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 2
+
+TEST touch $M0/a
+sleep 1
+TEST kill_brick $V0 $H0 $B0/brick0
+TEST touch $M0/a
+
+EXPECT_WITHIN $HEAL_TIMEOUT "^2$" get_pending_heal_count $V0
+
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status $V0 1
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "Y" glustershd_up_status
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 0
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 1
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 2
+
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" get_pending_heal_count $V0
+
+mtime0=$(get_mtime $B0/brick0/a)
+mtime1=$(get_mtime $B0/brick1/a)
+TEST [ $mtime0 -eq $mtime1 ]
+
+ctime0=$(get_ctime $B0/brick0/a)
+ctime1=$(get_ctime $B0/brick1/a)
+TEST [ $ctime0 -eq $ctime1 ]
+
+###############################################################################
+# Repeat the test with ctime feature disabled.
+TEST $CLI volume set $V0 features.ctime off
+
+TEST touch $M0/b
+sleep 1
+TEST kill_brick $V0 $H0 $B0/brick0
+TEST touch $M0/b
+
+EXPECT_WITHIN $HEAL_TIMEOUT "^2$" get_pending_heal_count $V0
+
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status $V0 1
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "Y" glustershd_up_status
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 0
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 1
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 2
+
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" get_pending_heal_count $V0
+
+mtime2=$(get_mtime $B0/brick0/b)
+mtime3=$(get_mtime $B0/brick1/b)
+TEST [ $mtime2 -eq $mtime3 ]
+
+TEST rm $GET_MDATA_PATH/get-mdata-xattr
+
+TEST force_umount $M0
+TEST $CLI volume stop $V0
+TEST $CLI volume delete $V0
+
+cleanup
diff --git a/xlators/cluster/afr/src/afr-self-heal-metadata.c b/xlators/cluster/afr/src/afr-self-heal-metadata.c
index ecfa791..f4e31b6 100644
--- a/xlators/cluster/afr/src/afr-self-heal-metadata.c
+++ b/xlators/cluster/afr/src/afr-self-heal-metadata.c
@@ -421,12 +421,8 @@ afr_selfheal_metadata(call_frame_t *frame, xlator_t *this, inode_t *inode)
if (ret)
goto unlock;
- /* Restore atime/mtime for files that don't need data heal as
- * restoring timestamps happens only as a part of data-heal.
- */
- if (!IA_ISREG(locked_replies[source].poststat.ia_type))
- afr_selfheal_restore_time(frame, this, inode, source, healed_sinks,
- locked_replies);
+ afr_selfheal_restore_time(frame, this, inode, source, healed_sinks,
+ locked_replies);
ret = afr_selfheal_undo_pending(
frame, this, inode, sources, sinks, healed_sinks, undid_pending,
--
1.8.3.1

38
SOURCES/0394-man-gluster-Add-volume-top-command-to-gluster-man-pa.patch Normal file
View File

@ -0,0 +1,38 @@
From 768a6d9bca86c0a50128b8776c11ef2b6d36388d Mon Sep 17 00:00:00 2001
From: Vishal Pandey <vpandey@redhat.com>
Date: Thu, 21 Nov 2019 12:56:34 +0530
Subject: [PATCH 394/449] man/gluster: Add volume top command to gluster man
page
> Upstream patch link: https://review.gluster.org/#/c/glusterfs/+/23735/
> Change-Id: Ib74607d2b2e5a1a0316221f1176a7dcccea632d4
> Fixes: bz#1774866
> Signed-off-by: Vishal Pandey <vpandey@redhat.com>
BUG: 1754391
Change-Id: Ib74607d2b2e5a1a0316221f1176a7dcccea632d4
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202333
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
doc/gluster.8 | 3 +++
1 file changed, 3 insertions(+)
diff --git a/doc/gluster.8 b/doc/gluster.8
index da6472d..88cbf44 100644
--- a/doc/gluster.8
+++ b/doc/gluster.8
@@ -113,6 +113,9 @@ Rotate the log file for corresponding volume/brick.
\fB\ volume profile <VOLNAME> {start|info [peek|incremental [peek]|cumulative|clear]|stop} [nfs] \fR
Profile operations on the volume. Once started, volume profile <volname> info provides cumulative statistics of the FOPs performed.
.TP
+\fB\ volume top <VOLNAME> {open|read|write|opendir|readdir|clear} [nfs|brick <brick>] [list-cnt <value>] | {read-perf|write-perf} [bs <size> count <count>] [brick <brick>] [list-cnt <value>] \fR
+Generates a profile of a volume representing the performance and bottlenecks/hotspots of each brick.
+.TP
\fB\ volume statedump <VOLNAME> [[nfs|quotad] [all|mem|iobuf|callpool|priv|fd|inode|history]... | [client <hostname:process-id>]] \fR
Dumps the in memory state of the specified process or the bricks of the volume.
.TP
--
1.8.3.1

111
SOURCES/0395-Cli-Removing-old-log-rotate-command.patch Normal file
View File

@ -0,0 +1,111 @@
From 5b3fcc8db86b4dc7af1eb63315ca2ff41c60fdea Mon Sep 17 00:00:00 2001
From: kshithijiyer <kshithij.ki@gmail.com>
Date: Sat, 30 Nov 2019 15:25:11 +0530
Subject: [PATCH 395/449] [Cli] Removing old log rotate command.
The old command for log rotate is still present removing
it completely. Also adding testcase to test the
log rotate command with both the old as well as the new command
and fixing testcase which use the old syntax to use the new
one.
Code to be removed:
1. In cli-cmd-volume.c from struct cli_cmd volume_cmds[]:
{"volume log rotate <VOLNAME> [BRICK]", cli_cmd_log_rotate_cbk,
"rotate the log file for corresponding volume/brick"
" NOTE: This is an old syntax, will be deprecated from next release."},
2. In cli-cmd-volume.c from cli_cmd_log_rotate_cbk():
||(strcmp("rotate", words[2]) == 0)))
3. In cli-cmd-parser.c from cli_cmd_log_rotate_parse()
if (strcmp("rotate", words[2]) == 0)
volname = (char *)words[3];
else
> upstream patch link: https://review.gluster.org/#/c/glusterfs/+/23392/
> fixes: bz#1750387
> Change-Id: I56e4d295044e8d5fd1fc0d848bc87e135e9e32b4
> Signed-off-by: kshithijiyer <kshithij.ki@gmail.com>
BUG: 1784415
Change-Id: I56e4d295044e8d5fd1fc0d848bc87e135e9e32b4
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202334
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
cli/src/cli-cmd-parser.c | 2 --
cli/src/cli-cmd-volume.c | 7 +------
tests/bugs/glusterd/optimized-basic-testcases.t | 3 ++-
tests/bugs/glusterfs-server/bug-852147.t | 2 +-
4 files changed, 4 insertions(+), 10 deletions(-)
diff --git a/cli/src/cli-cmd-parser.c b/cli/src/cli-cmd-parser.c
index 4456a7b..ac0a263 100644
--- a/cli/src/cli-cmd-parser.c
+++ b/cli/src/cli-cmd-parser.c
@@ -2592,8 +2592,6 @@ cli_cmd_log_rotate_parse(const char **words, int wordcount, dict_t **options)
if (strcmp("rotate", words[3]) == 0)
volname = (char *)words[2];
- else if (strcmp("rotate", words[2]) == 0)
- volname = (char *)words[3];
GF_ASSERT(volname);
ret = dict_set_str(dict, "volname", volname);
diff --git a/cli/src/cli-cmd-volume.c b/cli/src/cli-cmd-volume.c
index 754d333..f33fc99 100644
--- a/cli/src/cli-cmd-volume.c
+++ b/cli/src/cli-cmd-volume.c
@@ -2349,8 +2349,7 @@ cli_cmd_log_rotate_cbk(struct cli_state *state, struct cli_cmd_word *word,
goto out;
}
- if (!((strcmp("rotate", words[2]) == 0) ||
- (strcmp("rotate", words[3]) == 0))) {
+ if (!(strcmp("rotate", words[3]) == 0)) {
cli_usage_out(word->pattern);
parse_error = 1;
goto out;
@@ -3401,10 +3400,6 @@ struct cli_cmd volume_cmds[] = {
{"volume log <VOLNAME> rotate [BRICK]", cli_cmd_log_rotate_cbk,
"rotate the log file for corresponding volume/brick"},
- {"volume log rotate <VOLNAME> [BRICK]", cli_cmd_log_rotate_cbk,
- "rotate the log file for corresponding volume/brick"
- " NOTE: This is an old syntax, will be deprecated from next release."},
-
{"volume sync <HOSTNAME> [all|<VOLNAME>]", cli_cmd_sync_volume_cbk,
"sync the volume information from a peer"},
diff --git a/tests/bugs/glusterd/optimized-basic-testcases.t b/tests/bugs/glusterd/optimized-basic-testcases.t
index c7e8c32..862f329 100644
--- a/tests/bugs/glusterd/optimized-basic-testcases.t
+++ b/tests/bugs/glusterd/optimized-basic-testcases.t
@@ -129,7 +129,8 @@ TEST ! $CLI volume set all $V0 cluster.op-version $OP_VERS_NEW
#bug-1022055 - validate log rotate command
-TEST $CLI volume log rotate $V0;
+TEST ! $CLI volume log rotate $V0;
+TEST $CLI volume log $V0 rotate;
#bug-1092841 - validating barrier enable/disable
diff --git a/tests/bugs/glusterfs-server/bug-852147.t b/tests/bugs/glusterfs-server/bug-852147.t
index c644cfa..75db2a2 100755
--- a/tests/bugs/glusterfs-server/bug-852147.t
+++ b/tests/bugs/glusterfs-server/bug-852147.t
@@ -66,7 +66,7 @@ ren_file=$log_file".*"
rm -rf $ren_file
#Initiating log rotate
-TEST $CLI volume log rotate $V0
+TEST $CLI volume log $V0 rotate
#Capturing new log file's size
new_file_size=`file-size $log_file`
--
1.8.3.1

56
SOURCES/0396-Updating-gluster-manual.patch Normal file
View File

@ -0,0 +1,56 @@
From 728aab1c1cfcf352d4ca1fde0b80044dc24bd9fa Mon Sep 17 00:00:00 2001
From: Rishubh Jain <risjain@redhat.com>
Date: Sun, 18 Aug 2019 18:02:57 +0530
Subject: [PATCH 396/449] Updating gluster manual.
Adding disperse-data to gluster manual under
volume create command
> Upstream Patch Link: https://review.gluster.org/#/c/glusterfs/+/23258/
> Change-Id: Ic9eb47c9e71a1d7a11af9394c615c8e90f8d1d69
> Fixes: bz#1668239
> Signed-off-by: Rishubh Jain <risjain@redhat.com>
> Signed-off-by: Sheetal Pamecha <spamecha@redhat.com>
BUG: 1667954
Change-Id: Ic9eb47c9e71a1d7a11af9394c615c8e90f8d1d69
Signed-off-by: Sheetal Pamecha <spamecha@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202342
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
doc/gluster.8 | 2 +-
tests/basic/glusterd/disperse-create.t | 4 ++++
2 files changed, 5 insertions(+), 1 deletion(-)
diff --git a/doc/gluster.8 b/doc/gluster.8
index 88cbf44..66bdb48 100644
--- a/doc/gluster.8
+++ b/doc/gluster.8
@@ -41,7 +41,7 @@ List all volumes in cluster
\fB\ volume status [all | <VOLNAME> [nfs|shd|<BRICK>|quotad|tierd]] [detail|clients|mem|inode|fd|callpool|tasks|client-list] \fR
Display status of all or specified volume(s)/brick
.TP
-\fB\ volume create <NEW-VOLNAME> [stripe <COUNT>] [replica <COUNT>] [disperse [<COUNT>]] [redundancy <COUNT>] [transport <tcp|rdma|tcp,rdma>] <NEW-BRICK> ... \fR
+\fB\ volume create <NEW-VOLNAME> [stripe <COUNT>] [[replica <COUNT> [arbiter <COUNT>]]|[replica 2 thin-arbiter 1]] [disperse [<COUNT>]] [disperse-data <COUNT>] [redundancy <COUNT>] [transport <tcp|rdma|tcp,rdma>] <NEW-BRICK> ... <TA-BRICK> \fR
Create a new volume of the specified type using the specified bricks and transport type (the default transport type is tcp).
To create a volume with both transports (tcp and rdma), give 'transport tcp,rdma' as an option.
.TP
diff --git a/tests/basic/glusterd/disperse-create.t b/tests/basic/glusterd/disperse-create.t
index 384c675..db8a621 100644
--- a/tests/basic/glusterd/disperse-create.t
+++ b/tests/basic/glusterd/disperse-create.t
@@ -20,6 +20,10 @@ TEST $CLI volume create $V0 disperse 3 redundancy 1 $H0:$B0/b7 $H0:$B0/b8 $H0:$B
EXPECT "1 x \(2 \+ 1\) = 3" volinfo_field $V0 "Number of Bricks"
TEST $CLI volume delete $V0
+TEST $CLI volume create $V0 disperse-data 2 $H0:$B0/b10 $H0:$B0/b11 $H0:$B0/b12
+EXPECT "1 x \(2 \+ 1\) = 3" volinfo_field $V0 "Number of Bricks"
+
+TEST $CLI volume delete $V0
TEST $CLI volume create $V0 redundancy 1 $H0:$B0/b10 $H0:$B0/b11 $H0:$B0/b12
EXPECT "1 x \(2 \+ 1\) = 3" volinfo_field $V0 "Number of Bricks"
--
1.8.3.1

52
SOURCES/0397-mgmt-brick-mux-Avoid-sending-two-response-when-attac.patch Normal file
View File

@ -0,0 +1,52 @@
From 73cef29731c0d7b8b4f3b880c032dc232b8fcc31 Mon Sep 17 00:00:00 2001
From: Mohit Agrawal <moagrawa@redhat.com>
Date: Thu, 4 Jun 2020 16:06:44 +0530
Subject: [PATCH 397/449] mgmt/brick-mux: Avoid sending two response when
attach is failed.
We were sending two response back to glusterd when an attach is
failed. One from the handler function glusterfs_handle_attach and
another from rpcsvc_check_and_reply_error. It was causing problems
like ref leaks, transport disconnect etc.
> Change-Id: I3bb5b59959530760b568d52becb519499b3dcd2b
> updates: bz#1785143
> Signed-off-by: Mohammed Rafi KC <rkavunga@redhat.com>
> (Cherry pick from commit 42f484dcecd9942611396d9bd2ad3a39019b0e1f)
> (Reviewed on upstream link https://review.gluster.org/#/c/glusterfs/+/23906/)
Change-Id: I3bb5b59959530760b568d52becb519499b3dcd2b
BUG: 1776901
Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202346
Reviewed-by: Sanju Rakonde <srakonde@redhat.com>
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
glusterfsd/src/glusterfsd-mgmt.c | 10 +++++++++-
1 file changed, 9 insertions(+), 1 deletion(-)
diff --git a/glusterfsd/src/glusterfsd-mgmt.c b/glusterfsd/src/glusterfsd-mgmt.c
index 15acc10..61d1b21 100644
--- a/glusterfsd/src/glusterfsd-mgmt.c
+++ b/glusterfsd/src/glusterfsd-mgmt.c
@@ -954,7 +954,15 @@ glusterfs_handle_attach(rpcsvc_request_t *req)
ret = -1;
}
- glusterfs_translator_info_response_send(req, ret, NULL, NULL);
+ ret = glusterfs_translator_info_response_send(req, ret, NULL, NULL);
+ if (ret) {
+ /* Response sent back to glusterd, req is already destroyed. So
+ * resetting the ret to 0. Otherwise another response will be
+ * send from rpcsvc_check_and_reply_error. Which will lead to
+ * double resource leak.
+ */
+ ret = 0;
+ }
out:
UNLOCK(&ctx->volfile_lock);
--
1.8.3.1

75
SOURCES/0398-ec-change-error-message-for-heal-commands-for-disper.patch Normal file
View File

@ -0,0 +1,75 @@
From 03d2c7b52da5efd6ad660315a0548c8b91e51439 Mon Sep 17 00:00:00 2001
From: Sheetal Pamecha <spamecha@redhat.com>
Date: Sun, 22 Dec 2019 22:52:30 +0530
Subject: [PATCH 398/449] ec: change error message for heal commands for
disperse volume
Currently when we issue a heal statistics or similar commands
for disperse volume, it fails with message "Volume is not of
type replicate." Adding message "this command is supported for
volumes of type replicate" to reflect supportability and better
understanding of heal functionality for disperse volumes.
> Upstream Patch Link: https://review.gluster.org/#/c/glusterfs/+/23916/
> fixes: bz#1785998
> Change-Id: I9688a9fdf427cb6f657cfd5b8db2f76a6c56f6e2
> Signed-off-by: Sheetal Pamecha <spamecha@redhat.com>
BUG: 1487177
Change-Id: I9688a9fdf427cb6f657cfd5b8db2f76a6c56f6e2
Signed-off-by: Sheetal Pamecha <spamecha@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202344
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
heal/src/glfs-heal.c | 15 ++++++++++-----
xlators/mgmt/glusterd/src/glusterd-volume-ops.c | 4 +++-
2 files changed, 13 insertions(+), 6 deletions(-)
diff --git a/heal/src/glfs-heal.c b/heal/src/glfs-heal.c
index 7e37e47..125b12c 100644
--- a/heal/src/glfs-heal.c
+++ b/heal/src/glfs-heal.c
@@ -1726,14 +1726,19 @@ main(int argc, char **argv)
goto out;
}
+ char *var_str = (heal_op == GF_SHD_OP_INDEX_SUMMARY ||
+ heal_op == GF_SHD_OP_HEAL_SUMMARY)
+ ? "replicate/disperse"
+ : "replicate";
+
ret = glfsh_validate_volume(top_subvol, heal_op);
if (ret < 0) {
ret = -EINVAL;
- gf_asprintf(&op_errstr, "Volume %s is not of type %s", volname,
- (heal_op == GF_SHD_OP_INDEX_SUMMARY ||
- heal_op == GF_SHD_OP_HEAL_SUMMARY)
- ? "replicate/disperse"
- : "replicate");
+ gf_asprintf(&op_errstr,
+ "This command is supported "
+ "for only volumes of %s type. Volume %s "
+ "is not of type %s",
+ var_str, volname, var_str);
goto out;
}
rootloc.inode = inode_ref(top_subvol->itable->root);
diff --git a/xlators/mgmt/glusterd/src/glusterd-volume-ops.c b/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
index 076bc80..93042ab 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
@@ -2008,7 +2008,9 @@ glusterd_handle_heal_cmd(xlator_t *this, glusterd_volinfo_t *volinfo,
if (!glusterd_is_volume_replicate(volinfo)) {
ret = -1;
snprintf(msg, sizeof(msg),
- "Volume %s is not of type "
+ "This command is supported "
+ "for only volume of replicated "
+ "type. Volume %s is not of type "
"replicate",
volinfo->volname);
*op_errstr = gf_strdup(msg);
--
1.8.3.1

79
SOURCES/0399-glusterd-coverity-fixes.patch Normal file
View File

@ -0,0 +1,79 @@
From 1ebd2a3227469b1775f19c8f78af7d3d19f749a3 Mon Sep 17 00:00:00 2001
From: Atin Mukherjee <amukherj@redhat.com>
Date: Fri, 26 Apr 2019 08:47:12 +0530
Subject: [PATCH 399/449] glusterd: coverity fixes
1400775 - USE_AFTER_FREE
1400742 - Missing Unlock
1400736 - CHECKED_RETURN
1398470 - Missing Unlock
Missing unlock is the tricky one, we have had annotation added, but
coverity still continued to complaint. Added pthread_mutex_unlock to
clean up the lock before destroying it to see if it makes coverity
happy.
> upstream patch link: https://review.gluster.org/#/c/glusterfs/+/22634/
> Updates: bz#789278
> Change-Id: I1d892612a17f805144d96c1b15004a85a1639414
> Signed-off-by: Atin Mukherjee <amukherj@redhat.com>
BUG: 1787310
Change-Id: I1d892612a17f805144d96c1b15004a85a1639414
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202343
Reviewed-by: Mohit Agrawal <moagrawa@redhat.com>
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/mgmt/glusterd/src/glusterd-peer-utils.c | 1 +
xlators/mgmt/glusterd/src/glusterd-sm.c | 1 -
xlators/mgmt/glusterd/src/glusterd-utils.c | 8 +++++++-
3 files changed, 8 insertions(+), 2 deletions(-)
diff --git a/xlators/mgmt/glusterd/src/glusterd-peer-utils.c b/xlators/mgmt/glusterd/src/glusterd-peer-utils.c
index f24c86e..8c1feeb 100644
--- a/xlators/mgmt/glusterd/src/glusterd-peer-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-peer-utils.c
@@ -48,6 +48,7 @@ glusterd_peerinfo_destroy(struct rcu_head *head)
}
glusterd_sm_tr_log_delete(&peerinfo->sm_log);
+ pthread_mutex_unlock(&peerinfo->delete_lock);
pthread_mutex_destroy(&peerinfo->delete_lock);
GF_FREE(peerinfo);
diff --git a/xlators/mgmt/glusterd/src/glusterd-sm.c b/xlators/mgmt/glusterd/src/glusterd-sm.c
index 54a7bd1..044da3d 100644
--- a/xlators/mgmt/glusterd/src/glusterd-sm.c
+++ b/xlators/mgmt/glusterd/src/glusterd-sm.c
@@ -868,7 +868,6 @@ glusterd_ac_friend_remove(glusterd_friend_sm_event_t *event, void *ctx)
"Cleanup returned: %d", ret);
}
out:
- /* coverity[ LOCK] */
return 0;
}
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index 1b78812..a1299bc 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -5840,7 +5840,13 @@ attach_brick_callback(struct rpc_req *req, struct iovec *iov, int count,
/* PID file is copied once brick has attached
successfully
*/
- glusterd_copy_file(pidfile1, pidfile2);
+ ret = glusterd_copy_file(pidfile1, pidfile2);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, ENOMEM, GD_MSG_NO_MEMORY,
+ "Could not copy file %s to %s", pidfile1, pidfile2);
+ goto out;
+ }
+
brickinfo->status = GF_BRICK_STARTED;
brickinfo->rpc = rpc_clnt_ref(other_brick->rpc);
gf_log(THIS->name, GF_LOG_INFO, "brick %s is attached successfully",
--
1.8.3.1

98
SOURCES/0400-cli-throw-a-warning-if-replica-count-greater-than-3.patch Normal file
View File

@ -0,0 +1,98 @@
From 12ed9226fa24d073ab2b89692194b454a194c379 Mon Sep 17 00:00:00 2001
From: Sanju Rakonde <srakonde@redhat.com>
Date: Thu, 4 Jun 2020 15:14:29 +0530
Subject: [PATCH 400/449] cli: throw a warning if replica count greater than 3
As volumes with replica count greater than 3 are not
supported, a warning message is be thrown to user
while creating the volume with replica count greater
than 3 or while converting a volume to replica > 3
volume by add-brick/remove-brick operations.
Label: DOWNSTREAM ONLY
BUG: 1763129
Change-Id: I5a32a5a2d99b5175fb692dfcab27396089f24b72
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202338
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-by: Ravishankar Narayanankutty <ravishankar@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
cli/src/cli-cmd-parser.c | 45 +++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 45 insertions(+)
diff --git a/cli/src/cli-cmd-parser.c b/cli/src/cli-cmd-parser.c
index ac0a263..5e7ce53 100644
--- a/cli/src/cli-cmd-parser.c
+++ b/cli/src/cli-cmd-parser.c
@@ -619,6 +619,23 @@ cli_cmd_volume_create_parse(struct cli_state *state, const char **words,
}
}
}
+
+ if (replica_count > 3) {
+ if (strcmp(words[wordcount - 1], "force")) {
+ question =
+ "Volumes with replica count greater than 3 are"
+ "not supported. \nDo you still want to continue?\n";
+ answer = cli_cmd_get_confirmation(state, question);
+ if (GF_ANSWER_NO == answer) {
+ gf_log("cli", GF_LOG_ERROR,
+ "Volume create "
+ "cancelled, exiting");
+ ret = -1;
+ goto out;
+ }
+ }
+ }
+
ret = dict_set_int32(dict, "replica-count", replica_count);
if (ret)
goto out;
@@ -1815,6 +1832,20 @@ cli_cmd_volume_add_brick_parse(struct cli_state *state, const char **words,
goto out;
}
}
+ } else if (count > 3) {
+ if (strcmp(words[wordcount - 1], "force")) {
+ question =
+ "Volumes with replica count greater than 3 are"
+ "not supported. \nDo you still want to continue?\n";
+ answer = cli_cmd_get_confirmation(state, question);
+ if (GF_ANSWER_NO == answer) {
+ gf_log("cli", GF_LOG_ERROR,
+ "add-brick "
+ "cancelled, exiting");
+ ret = -1;
+ goto out;
+ }
+ }
}
} else if ((strcmp(w, "stripe")) == 0) {
cli_err("stripe option not supported");
@@ -2082,6 +2113,20 @@ cli_cmd_volume_remove_brick_parse(struct cli_state *state, const char **words,
goto out;
}
}
+ } else if (count > 3) {
+ if (strcmp(words[wordcount - 1], "force")) {
+ ques =
+ "Volumes with replica count greater than 3 are"
+ "not supported. \nDo you still want to continue?\n";
+ answer = cli_cmd_get_confirmation(state, ques);
+ if (GF_ANSWER_NO == answer) {
+ gf_log("cli", GF_LOG_ERROR,
+ "Remove-brick "
+ "cancelled, exiting");
+ ret = -1;
+ goto out;
+ }
+ }
}
ret = dict_set_int32(dict, "replica-count", count);
--
1.8.3.1

70
SOURCES/0401-cli-change-the-warning-message.patch Normal file
View File

@ -0,0 +1,70 @@
From 704bf84d432e1eea1534e35ee27d4116a7273146 Mon Sep 17 00:00:00 2001
From: Sanju Rakonde <srakonde@redhat.com>
Date: Thu, 4 Jun 2020 16:15:35 +0530
Subject: [PATCH 401/449] cli: change the warning message
while creating the replica 2 volume or converting
a volume to replica 2 volume, we issue a warning
saying "replica 2 volumes are prone to split brain".
As the support for replica 2 volumes has been deprecated,
warning message should be changed accordingly to reflect
the same.
Label: DOWNSTREAM ONLY
BUG: 1763124
Change-Id: If55e5412cda2e4a21a6359492d8d704dd702530d
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202348
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-by: Ravishankar Narayanankutty <ravishankar@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
cli/src/cli-cmd-parser.c | 16 ++++++++--------
1 file changed, 8 insertions(+), 8 deletions(-)
diff --git a/cli/src/cli-cmd-parser.c b/cli/src/cli-cmd-parser.c
index 5e7ce53..7446b95 100644
--- a/cli/src/cli-cmd-parser.c
+++ b/cli/src/cli-cmd-parser.c
@@ -603,8 +603,8 @@ cli_cmd_volume_create_parse(struct cli_state *state, const char **words,
if (replica_count == 2) {
if (strcmp(words[wordcount - 1], "force")) {
question =
- "Replica 2 volumes are prone"
- " to split-brain. Use "
+ "Support for replica 2 volumes stands deprecated as "
+ "they are prone to split-brain. Use "
"Arbiter or Replica 3 to "
"avoid this.\n"
"Do you still want to "
@@ -1817,9 +1817,9 @@ cli_cmd_volume_add_brick_parse(struct cli_state *state, const char **words,
if (count == 2) {
if (strcmp(words[wordcount - 1], "force")) {
question =
- "Replica 2 volumes are prone to "
- "split-brain. Use Arbiter or "
- "Replica 3 to avaoid this. See: "
+ "Support for replica 2 volumes stands deprecated as they "
+ "are prone to split-brain. Use Arbiter or "
+ "Replica 3 to avoid this. See: "
"http://docs.gluster.org/en/latest/Administrator%20Guide/"
"Split%20brain%20and%20ways%20to%20deal%20with%20it/."
"\nDo you still want to continue?\n";
@@ -2098,9 +2098,9 @@ cli_cmd_volume_remove_brick_parse(struct cli_state *state, const char **words,
if (count == 2) {
if (strcmp(words[wordcount - 1], "force")) {
ques =
- "Replica 2 volumes are prone to "
- "split-brain. Use Arbiter or Replica 3 "
- "to avaoid this. See: "
+ "Support for replica 2 volumes stands deprecated as they "
+ "are prone to split-brain. Use Arbiter or Replica 3 "
+ "to avoid this. See: "
"http://docs.gluster.org/en/latest/Administrator%20Guide/"
"Split%20brain%20and%20ways%20to%20deal%20with%20it/."
"\nDo you still want to continue?\n";
--
1.8.3.1

198
SOURCES/0402-afr-wake-up-index-healer-threads.patch Normal file
View File

@ -0,0 +1,198 @@
From ecaa0f10820f4b6e803021919ce59a43aedf356b Mon Sep 17 00:00:00 2001
From: Ravishankar N <ravishankar@redhat.com>
Date: Thu, 4 Jun 2020 16:15:35 +0530
Subject: [PATCH 402/449] afr: wake up index healer threads
...whenever shd is re-enabled after disabling or there is a change in
`cluster.heal-timeout`, without needing to restart shd or waiting for the
current `cluster.heal-timeout` seconds to expire.
> Upstream patch link:https://review.gluster.org/#/c/glusterfs/+/23288/
> Change-Id: Ia5ebd7c8e9f5b54cba3199c141fdd1af2f9b9bfe
> fixes: bz#1744548
> Reported-by: Glen Kiessling <glenk1973@hotmail.com>
> Signed-off-by: Ravishankar N <ravishankar@redhat.com>
BUG: 1764091
Change-Id: I42aa0807f09b5a09510fe9efb4a1697dad3410a3
Signed-off-by: Ravishankar N <ravishankar@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202368
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/bugs/replicate/bug-1744548-heal-timeout.t | 42 +++++++++++++++++++++++++
xlators/cluster/afr/src/afr-common.c | 6 ++--
xlators/cluster/afr/src/afr-self-heald.c | 14 ++++++---
xlators/cluster/afr/src/afr-self-heald.h | 3 --
xlators/cluster/afr/src/afr.c | 10 ++++++
xlators/cluster/afr/src/afr.h | 2 ++
6 files changed, 66 insertions(+), 11 deletions(-)
create mode 100644 tests/bugs/replicate/bug-1744548-heal-timeout.t
diff --git a/tests/bugs/replicate/bug-1744548-heal-timeout.t b/tests/bugs/replicate/bug-1744548-heal-timeout.t
new file mode 100644
index 0000000..3cb73bc
--- /dev/null
+++ b/tests/bugs/replicate/bug-1744548-heal-timeout.t
@@ -0,0 +1,42 @@
+#!/bin/bash
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../afr.rc
+
+cleanup;
+
+TEST glusterd;
+TEST pidof glusterd;
+TEST $CLI volume create $V0 replica 3 $H0:$B0/${V0}{0,1,2}
+TEST $CLI volume heal $V0 disable
+TEST $CLI volume start $V0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V0 $H0 $B0/${V0}0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V0 $H0 $B0/${V0}1
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V0 $H0 $B0/${V0}2
+TEST ! $CLI volume heal $V0
+
+# Enable shd and verify that index crawl is triggered immediately.
+TEST $CLI volume profile $V0 start
+TEST $CLI volume profile $V0 info clear
+TEST $CLI volume heal $V0 enable
+TEST $CLI volume heal $V0
+# Each brick does 3 opendirs, corresponding to dirty, xattrop and entry-changes
+COUNT=`$CLI volume profile $V0 info incremental |grep OPENDIR|awk '{print $8}'|tr -d '\n'`
+TEST [ "$COUNT" == "333" ]
+
+# Check that a change in heal-timeout is honoured immediately.
+TEST $CLI volume set $V0 cluster.heal-timeout 5
+sleep 10
+COUNT=`$CLI volume profile $V0 info incremental |grep OPENDIR|awk '{print $8}'|tr -d '\n'`
+# Two crawls must have happened.
+TEST [ "$COUNT" == "666" ]
+
+# shd must not heal if it is disabled and heal-timeout is changed.
+TEST $CLI volume heal $V0 disable
+TEST $CLI volume profile $V0 info clear
+TEST $CLI volume set $V0 cluster.heal-timeout 6
+sleep 6
+COUNT=`$CLI volume profile $V0 info incremental |grep OPENDIR|awk '{print $8}'|tr -d '\n'`
+TEST [ -z $COUNT ]
+cleanup;
diff --git a/xlators/cluster/afr/src/afr-common.c b/xlators/cluster/afr/src/afr-common.c
index 3690b84..eef7fd2 100644
--- a/xlators/cluster/afr/src/afr-common.c
+++ b/xlators/cluster/afr/src/afr-common.c
@@ -5613,10 +5613,8 @@ afr_notify(xlator_t *this, int32_t event, void *data, void *data2)
* b) Already heard from everyone, but we now got a child-up
* event.
*/
- if (have_heard_from_all && priv->shd.iamshd) {
- for (i = 0; i < priv->child_count; i++)
- if (priv->child_up[i])
- afr_selfheal_childup(this, i);
+ if (have_heard_from_all) {
+ afr_selfheal_childup(this, priv);
}
}
out:
diff --git a/xlators/cluster/afr/src/afr-self-heald.c b/xlators/cluster/afr/src/afr-self-heald.c
index 7eb1207..95ac5f2 100644
--- a/xlators/cluster/afr/src/afr-self-heald.c
+++ b/xlators/cluster/afr/src/afr-self-heald.c
@@ -1258,12 +1258,18 @@ out:
return ret;
}
-int
-afr_selfheal_childup(xlator_t *this, int subvol)
+void
+afr_selfheal_childup(xlator_t *this, afr_private_t *priv)
{
- afr_shd_index_healer_spawn(this, subvol);
+ int subvol = 0;
- return 0;
+ if (!priv->shd.iamshd)
+ return;
+ for (subvol = 0; subvol < priv->child_count; subvol++)
+ if (priv->child_up[subvol])
+ afr_shd_index_healer_spawn(this, subvol);
+
+ return;
}
int
diff --git a/xlators/cluster/afr/src/afr-self-heald.h b/xlators/cluster/afr/src/afr-self-heald.h
index 7de7c43..1990539 100644
--- a/xlators/cluster/afr/src/afr-self-heald.h
+++ b/xlators/cluster/afr/src/afr-self-heald.h
@@ -60,9 +60,6 @@ typedef struct {
} afr_self_heald_t;
int
-afr_selfheal_childup(xlator_t *this, int subvol);
-
-int
afr_selfheal_daemon_init(xlator_t *this);
int
diff --git a/xlators/cluster/afr/src/afr.c b/xlators/cluster/afr/src/afr.c
index 33258a0..8f9e71f 100644
--- a/xlators/cluster/afr/src/afr.c
+++ b/xlators/cluster/afr/src/afr.c
@@ -141,6 +141,7 @@ reconfigure(xlator_t *this, dict_t *options)
afr_private_t *priv = NULL;
xlator_t *read_subvol = NULL;
int read_subvol_index = -1;
+ int timeout_old = 0;
int ret = -1;
int index = -1;
char *qtype = NULL;
@@ -150,6 +151,7 @@ reconfigure(xlator_t *this, dict_t *options)
char *locking_scheme = NULL;
gf_boolean_t consistent_io = _gf_false;
gf_boolean_t choose_local_old = _gf_false;
+ gf_boolean_t enabled_old = _gf_false;
priv = this->private;
@@ -255,11 +257,13 @@ reconfigure(xlator_t *this, dict_t *options)
GF_OPTION_RECONF("ensure-durability", priv->ensure_durability, options,
bool, out);
+ enabled_old = priv->shd.enabled;
GF_OPTION_RECONF("self-heal-daemon", priv->shd.enabled, options, bool, out);
GF_OPTION_RECONF("iam-self-heal-daemon", priv->shd.iamshd, options, bool,
out);
+ timeout_old = priv->shd.timeout;
GF_OPTION_RECONF("heal-timeout", priv->shd.timeout, options, int32, out);
GF_OPTION_RECONF("consistent-metadata", priv->consistent_metadata, options,
@@ -283,6 +287,12 @@ reconfigure(xlator_t *this, dict_t *options)
consistent_io = _gf_false;
priv->consistent_io = consistent_io;
+ if (priv->shd.enabled) {
+ if ((priv->shd.enabled != enabled_old) ||
+ (timeout_old != priv->shd.timeout))
+ afr_selfheal_childup(this, priv);
+ }
+
ret = 0;
out:
return ret;
diff --git a/xlators/cluster/afr/src/afr.h b/xlators/cluster/afr/src/afr.h
index e731cfa..18f1a6a 100644
--- a/xlators/cluster/afr/src/afr.h
+++ b/xlators/cluster/afr/src/afr.h
@@ -1332,4 +1332,6 @@ afr_lookup_has_quorum(call_frame_t *frame, xlator_t *this,
void
afr_mark_new_entry_changelog(call_frame_t *frame, xlator_t *this);
+void
+afr_selfheal_childup(xlator_t *this, afr_private_t *priv);
#endif /* __AFR_H__ */
--
1.8.3.1

84
SOURCES/0403-Fix-spurious-failure-in-bug-1744548-heal-timeout.t.patch Normal file
View File

@ -0,0 +1,84 @@
From b311385a3c4bd56d69d1fa7e9bd3d9a2ae5c344e Mon Sep 17 00:00:00 2001
From: Pranith Kumar K <pkarampu@redhat.com>
Date: Mon, 7 Oct 2019 12:27:01 +0530
Subject: [PATCH 403/449] Fix spurious failure in bug-1744548-heal-timeout.t
Script was assuming that the heal would have triggered
by the time test was executed, which may not be the case.
It can lead to following failures when the race happens:
...
18:29:45 not ok 14 [ 85/ 1] < 26> '[ 331 == 333 ]' -> ''
...
18:29:45 not ok 16 [ 10097/ 1] < 33> '[ 668 == 666 ]' -> ''
Heal on 3rd brick didn't start completely first time the command was executed.
So the extra count got added to the next profile info.
Fixed it by depending on cumulative stats and waiting until the count is
satisfied using EXPECT_WITHIN
> Upstream patch link:https://review.gluster.org/23523
>fixes: bz#1759002
>Change-Id: I3b410671c902d6b1458a757fa245613cb29d967d
>Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
BUG: 1764091
Change-Id: Ic4d16b6c8a1bbc35735567d60fd0383456b9f534
Signed-off-by: Ravishankar N <ravishankar@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202369
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/bugs/replicate/bug-1744548-heal-timeout.t | 17 +++++++++++------
1 file changed, 11 insertions(+), 6 deletions(-)
diff --git a/tests/bugs/replicate/bug-1744548-heal-timeout.t b/tests/bugs/replicate/bug-1744548-heal-timeout.t
index 3cb73bc..0aaa3ea 100644
--- a/tests/bugs/replicate/bug-1744548-heal-timeout.t
+++ b/tests/bugs/replicate/bug-1744548-heal-timeout.t
@@ -4,6 +4,11 @@
. $(dirname $0)/../../volume.rc
. $(dirname $0)/../../afr.rc
+function get_cumulative_opendir_count {
+#sed 'n:d' prints odd-numbered lines
+ $CLI volume profile $V0 info |grep OPENDIR|sed 'n;d' | awk '{print $8}'|tr -d '\n'
+}
+
cleanup;
TEST glusterd;
@@ -20,23 +25,23 @@ TEST ! $CLI volume heal $V0
TEST $CLI volume profile $V0 start
TEST $CLI volume profile $V0 info clear
TEST $CLI volume heal $V0 enable
-TEST $CLI volume heal $V0
# Each brick does 3 opendirs, corresponding to dirty, xattrop and entry-changes
-COUNT=`$CLI volume profile $V0 info incremental |grep OPENDIR|awk '{print $8}'|tr -d '\n'`
-TEST [ "$COUNT" == "333" ]
+EXPECT_WITHIN $HEAL_TIMEOUT "^333$" get_cumulative_opendir_count
# Check that a change in heal-timeout is honoured immediately.
TEST $CLI volume set $V0 cluster.heal-timeout 5
sleep 10
-COUNT=`$CLI volume profile $V0 info incremental |grep OPENDIR|awk '{print $8}'|tr -d '\n'`
# Two crawls must have happened.
-TEST [ "$COUNT" == "666" ]
+EXPECT_WITHIN $HEAL_TIMEOUT "^999$" get_cumulative_opendir_count
# shd must not heal if it is disabled and heal-timeout is changed.
TEST $CLI volume heal $V0 disable
+#Wait for configuration update and any opendir fops to complete
+sleep 10
TEST $CLI volume profile $V0 info clear
TEST $CLI volume set $V0 cluster.heal-timeout 6
-sleep 6
+#Better to wait for more than 6 seconds to account for configuration updates
+sleep 10
COUNT=`$CLI volume profile $V0 info incremental |grep OPENDIR|awk '{print $8}'|tr -d '\n'`
TEST [ -z $COUNT ]
cleanup;
--
1.8.3.1

38
SOURCES/0404-tests-Fix-spurious-failure.patch Normal file
View File

@ -0,0 +1,38 @@
From b65ca1045910bc18c601681788eb322dbb8ec2fa Mon Sep 17 00:00:00 2001
From: Pranith Kumar K <pkarampu@redhat.com>
Date: Mon, 14 Oct 2019 10:29:31 +0530
Subject: [PATCH 404/449] tests: Fix spurious failure
> Upstream patch:https://review.gluster.org/23546
> fixes: bz#1759002
> Change-Id: I4d49e1c2ca9b3c1d74b9dd5a30f1c66983a76529
> Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
BUG: 1764091
Change-Id: I8b66f08cce7a87788867c6373aed71d6fc65155f
Signed-off-by: Ravishankar N <ravishankar@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202370
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/bugs/replicate/bug-1744548-heal-timeout.t | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/tests/bugs/replicate/bug-1744548-heal-timeout.t b/tests/bugs/replicate/bug-1744548-heal-timeout.t
index 0aaa3ea..c208112 100644
--- a/tests/bugs/replicate/bug-1744548-heal-timeout.t
+++ b/tests/bugs/replicate/bug-1744548-heal-timeout.t
@@ -5,8 +5,8 @@
. $(dirname $0)/../../afr.rc
function get_cumulative_opendir_count {
-#sed 'n:d' prints odd-numbered lines
- $CLI volume profile $V0 info |grep OPENDIR|sed 'n;d' | awk '{print $8}'|tr -d '\n'
+#sed command prints content between Cumulative and Interval, this keeps content from Cumulative stats
+ $CLI volume profile $V0 info |sed -n '/^Cumulative/,/^Interval/p'|grep OPENDIR| awk '{print $8}'|tr -d '\n'
}
cleanup;
--
1.8.3.1

175
SOURCES/0405-core-fix-return-of-local-in-__nlc_inode_ctx_get.patch Normal file
View File

@ -0,0 +1,175 @@
From 9c5f5b4ffd49e8c8631defb7b6873248bbfdaf9c Mon Sep 17 00:00:00 2001
From: Rinku Kothiya <rkothiya@redhat.com>
Date: Tue, 23 Jul 2019 13:16:04 +0000
Subject: [PATCH 405/449] [core] fix return of local in __nlc_inode_ctx_get
__nlc_inode_ctx_get assigns a value to nlc_pe_p which is never used by
its parent function or any of the predecessor hence remove the
assignment and also that function argument as it is not being used
anywhere.
> fixes: bz#1732496
> Change-Id: I5b950e1e251bd50a646616da872a4efe9d2ff8c9
> (Cherry pick from commit 84a55090123a7e3124100e5564da8c521c3c22ab )
> (Reviewed on upstream link https://review.gluster.org/#/c/glusterfs/+/23093/)
BUG: 1686897
Change-Id: I5b950e1e251bd50a646616da872a4efe9d2ff8c9
Signed-off-by: Rinku Kothiya <rkothiya@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202372
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/performance/nl-cache/src/nl-cache-helper.c | 36 +++++++++-------------
1 file changed, 14 insertions(+), 22 deletions(-)
diff --git a/xlators/performance/nl-cache/src/nl-cache-helper.c b/xlators/performance/nl-cache/src/nl-cache-helper.c
index 009f33a..4314038 100644
--- a/xlators/performance/nl-cache/src/nl-cache-helper.c
+++ b/xlators/performance/nl-cache/src/nl-cache-helper.c
@@ -145,12 +145,10 @@ nlc_disable_cache(xlator_t *this)
}
static int
-__nlc_inode_ctx_get(xlator_t *this, inode_t *inode, nlc_ctx_t **nlc_ctx_p,
- nlc_pe_t **nlc_pe_p)
+__nlc_inode_ctx_get(xlator_t *this, inode_t *inode, nlc_ctx_t **nlc_ctx_p)
{
int ret = 0;
nlc_ctx_t *nlc_ctx = NULL;
- nlc_pe_t *nlc_pe = NULL;
uint64_t nlc_ctx_int = 0;
uint64_t nlc_pe_int = 0;
@@ -159,10 +157,6 @@ __nlc_inode_ctx_get(xlator_t *this, inode_t *inode, nlc_ctx_t **nlc_ctx_p,
nlc_ctx = (void *)(long)(nlc_ctx_int);
*nlc_ctx_p = nlc_ctx;
}
- if (ret == 0 && nlc_pe_p) {
- nlc_pe = (void *)(long)(nlc_pe_int);
- *nlc_pe_p = nlc_pe;
- }
return ret;
}
@@ -186,14 +180,13 @@ nlc_inode_ctx_set(xlator_t *this, inode_t *inode, nlc_ctx_t *nlc_ctx,
}
static void
-nlc_inode_ctx_get(xlator_t *this, inode_t *inode, nlc_ctx_t **nlc_ctx_p,
- nlc_pe_t **nlc_pe_p)
+nlc_inode_ctx_get(xlator_t *this, inode_t *inode, nlc_ctx_t **nlc_ctx_p)
{
int ret = 0;
LOCK(&inode->lock);
{
- ret = __nlc_inode_ctx_get(this, inode, nlc_ctx_p, nlc_pe_p);
+ ret = __nlc_inode_ctx_get(this, inode, nlc_ctx_p);
if (ret < 0)
gf_msg_debug(this->name, 0,
"inode ctx get failed for "
@@ -290,8 +283,7 @@ out:
}
static nlc_ctx_t *
-nlc_inode_ctx_get_set(xlator_t *this, inode_t *inode, nlc_ctx_t **nlc_ctx_p,
- nlc_pe_t **nlc_pe_p)
+nlc_inode_ctx_get_set(xlator_t *this, inode_t *inode, nlc_ctx_t **nlc_ctx_p)
{
int ret = 0;
nlc_ctx_t *nlc_ctx = NULL;
@@ -301,7 +293,7 @@ nlc_inode_ctx_get_set(xlator_t *this, inode_t *inode, nlc_ctx_t **nlc_ctx_p,
LOCK(&inode->lock);
{
- ret = __nlc_inode_ctx_get(this, inode, &nlc_ctx, nlc_pe_p);
+ ret = __nlc_inode_ctx_get(this, inode, &nlc_ctx);
if (nlc_ctx)
goto unlock;
@@ -410,7 +402,7 @@ nlc_set_dir_state(xlator_t *this, inode_t *inode, uint64_t state)
goto out;
}
- nlc_inode_ctx_get_set(this, inode, &nlc_ctx, NULL);
+ nlc_inode_ctx_get_set(this, inode, &nlc_ctx);
if (!nlc_ctx)
goto out;
@@ -430,7 +422,7 @@ nlc_cache_timeout_handler(struct gf_tw_timer_list *timer, void *data,
nlc_timer_data_t *tmp = data;
nlc_ctx_t *nlc_ctx = NULL;
- nlc_inode_ctx_get(tmp->this, tmp->inode, &nlc_ctx, NULL);
+ nlc_inode_ctx_get(tmp->this, tmp->inode, &nlc_ctx);
if (!nlc_ctx)
goto out;
@@ -696,7 +688,7 @@ nlc_inode_clear_cache(xlator_t *this, inode_t *inode, int reason)
{
nlc_ctx_t *nlc_ctx = NULL;
- nlc_inode_ctx_get(this, inode, &nlc_ctx, NULL);
+ nlc_inode_ctx_get(this, inode, &nlc_ctx);
if (!nlc_ctx)
goto out;
@@ -883,7 +875,7 @@ nlc_dir_add_ne(xlator_t *this, inode_t *inode, const char *name)
goto out;
}
- nlc_inode_ctx_get_set(this, inode, &nlc_ctx, NULL);
+ nlc_inode_ctx_get_set(this, inode, &nlc_ctx);
if (!nlc_ctx)
goto out;
@@ -914,7 +906,7 @@ nlc_dir_remove_pe(xlator_t *this, inode_t *parent, inode_t *entry_ino,
goto out;
}
- nlc_inode_ctx_get(this, parent, &nlc_ctx, NULL);
+ nlc_inode_ctx_get(this, parent, &nlc_ctx);
if (!nlc_ctx)
goto out;
@@ -945,7 +937,7 @@ nlc_dir_add_pe(xlator_t *this, inode_t *inode, inode_t *entry_ino,
goto out;
}
- nlc_inode_ctx_get_set(this, inode, &nlc_ctx, NULL);
+ nlc_inode_ctx_get_set(this, inode, &nlc_ctx);
if (!nlc_ctx)
goto out;
@@ -1051,7 +1043,7 @@ nlc_is_negative_lookup(xlator_t *this, loc_t *loc)
goto out;
}
- nlc_inode_ctx_get(this, inode, &nlc_ctx, NULL);
+ nlc_inode_ctx_get(this, inode, &nlc_ctx);
if (!nlc_ctx)
goto out;
@@ -1102,7 +1094,7 @@ nlc_get_real_file_name(xlator_t *this, loc_t *loc, const char *fname,
goto out;
}
- nlc_inode_ctx_get(this, inode, &nlc_ctx, NULL);
+ nlc_inode_ctx_get(this, inode, &nlc_ctx);
if (!nlc_ctx)
goto out;
@@ -1152,7 +1144,7 @@ nlc_dump_inodectx(xlator_t *this, inode_t *inode)
nlc_ne_t *ne = NULL;
nlc_ne_t *tmp1 = NULL;
- nlc_inode_ctx_get(this, inode, &nlc_ctx, NULL);
+ nlc_inode_ctx_get(this, inode, &nlc_ctx);
if (!nlc_ctx)
goto out;
--
1.8.3.1

185
SOURCES/0406-afr-support-split-brain-CLI-for-replica-3.patch Normal file
View File

@ -0,0 +1,185 @@
From a75bb15fbe64f14580c44b8a33314c8bbeffdede Mon Sep 17 00:00:00 2001
From: Ravishankar N <ravishankar@redhat.com>
Date: Thu, 4 Jun 2020 18:54:46 +0530
Subject: [PATCH 406/449] afr: support split-brain CLI for replica 3
Patch in upstream master: https://review.gluster.org/#/c/glusterfs/+/23502/
Ever since we added quorum checks for lookups in afr via commit
bd44d59741bb8c0f5d7a62c5b1094179dd0ce8a4, the split-brain resolution
commands would not work for replica 3 because there would be no
readables for the lookup fop.
The argument was that split-brains do not occur in replica 3 but we do
see (data/metadata) split-brain cases once in a while which indicate that there are
a few bugs/corner cases yet to be discovered and fixed.
Fortunately, commit 8016d51a3bbd410b0b927ed66be50a09574b7982 added
GF_CLIENT_PID_GLFS_HEALD as the pid for all fops made by glfsheal. If we
leverage this and allow lookups in afr when pid is GF_CLIENT_PID_GLFS_HEALD,
split-brain resolution commands will work for replica 3 volumes too.
Likewise, the check is added in shard_lookup as well to permit resolving
split-brains by specifying "/.shard/shard-file.xx" as the file name
(which previously used to fail with EPERM).
BUG: 1759875
Change-Id: I203735b909c7d30fc4faaf3ecd4f5b6b379ab266
Signed-off-by: Ravishankar N <ravishankar@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202375
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
.../replicate/bug-1756938-replica-3-sbrain-cli.t | 111 +++++++++++++++++++++
xlators/cluster/afr/src/afr-common.c | 3 +-
xlators/features/shard/src/shard.c | 3 +-
3 files changed, 115 insertions(+), 2 deletions(-)
create mode 100644 tests/bugs/replicate/bug-1756938-replica-3-sbrain-cli.t
diff --git a/tests/bugs/replicate/bug-1756938-replica-3-sbrain-cli.t b/tests/bugs/replicate/bug-1756938-replica-3-sbrain-cli.t
new file mode 100644
index 0000000..c1bdf34
--- /dev/null
+++ b/tests/bugs/replicate/bug-1756938-replica-3-sbrain-cli.t
@@ -0,0 +1,111 @@
+#!/bin/bash
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../afr.rc
+
+cleanup;
+
+TEST glusterd;
+TEST pidof glusterd;
+TEST $CLI volume create $V0 replica 3 $H0:$B0/${V0}{0,1,2}
+TEST $CLI volume set $V0 features.shard enable
+TEST $CLI volume set $V0 features.shard-block-size 4MB
+
+TEST $CLI volume start $V0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V0 $H0 $B0/${V0}0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V0 $H0 $B0/${V0}1
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V0 $H0 $B0/${V0}2
+TEST glusterfs --volfile-server=$H0 --volfile-id=/$V0 $M0
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status $V0 0
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status $V0 1
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status $V0 2
+
+#Create split-brain by setting afr xattrs/gfids manually.
+#file1 is non-sharded and will be in data split-brain.
+#file2 will have one shard which will be in data split-brain.
+#file3 will have one shard which will be in gfid split-brain.
+#file4 will have one shard which will be in data & metadata split-brain.
+TEST dd if=/dev/zero of=$M0/file1 bs=1024 count=1024 oflag=direct
+TEST dd if=/dev/zero of=$M0/file2 bs=1M count=6 oflag=direct
+TEST dd if=/dev/zero of=$M0/file3 bs=1M count=6 oflag=direct
+TEST dd if=/dev/zero of=$M0/file4 bs=1M count=6 oflag=direct
+EXPECT_WITHIN $UMOUNT_TIMEOUT "Y" force_umount $M0
+
+#-------------------------------------------------------------------------------
+TEST setfattr -n trusted.afr.$V0-client-1 -v 0x000000010000000000000000 $B0/${V0}0/file1
+TEST setfattr -n trusted.afr.$V0-client-2 -v 0x000000010000000000000000 $B0/${V0}0/file1
+TEST setfattr -n trusted.afr.$V0-client-0 -v 0x000000010000000000000000 $B0/${V0}1/file1
+TEST setfattr -n trusted.afr.$V0-client-2 -v 0x000000010000000000000000 $B0/${V0}1/file1
+TEST setfattr -n trusted.afr.$V0-client-0 -v 0x000000010000000000000000 $B0/${V0}2/file1
+TEST setfattr -n trusted.afr.$V0-client-1 -v 0x000000010000000000000000 $B0/${V0}2/file1
+
+#-------------------------------------------------------------------------------
+gfid_f2=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $B0/${V0}0/file2))
+TEST setfattr -n trusted.afr.$V0-client-1 -v 0x000000010000000000000000 $B0/${V0}0/.shard/$gfid_f2.1
+TEST setfattr -n trusted.afr.$V0-client-2 -v 0x000000010000000000000000 $B0/${V0}0/.shard/$gfid_f2.1
+TEST setfattr -n trusted.afr.$V0-client-0 -v 0x000000010000000000000000 $B0/${V0}1/.shard/$gfid_f2.1
+TEST setfattr -n trusted.afr.$V0-client-2 -v 0x000000010000000000000000 $B0/${V0}1/.shard/$gfid_f2.1
+TEST setfattr -n trusted.afr.$V0-client-0 -v 0x000000010000000000000000 $B0/${V0}2/.shard/$gfid_f2.1
+TEST setfattr -n trusted.afr.$V0-client-1 -v 0x000000010000000000000000 $B0/${V0}2/.shard/$gfid_f2.1
+
+#-------------------------------------------------------------------------------
+TESTS_EXPECTED_IN_LOOP=5
+function assign_new_gfid {
+ brickpath=$1
+ filename=$2
+ gfid=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $brickpath/$filename))
+ gfid_shard=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $brickpath/.shard/$gfid.1))
+
+ TEST rm $brickpath/.glusterfs/${gfid_shard:0:2}/${gfid_shard:2:2}/$gfid_shard
+ TEST setfattr -x trusted.gfid $brickpath/.shard/$gfid.1
+ new_gfid=$(get_random_gfid)
+ new_gfid_str=$(gf_gfid_xattr_to_str $new_gfid)
+ TEST setfattr -n trusted.gfid -v $new_gfid $brickpath/.shard/$gfid.1
+ TEST mkdir -p $brickpath/.glusterfs/${new_gfid_str:0:2}/${new_gfid_str:2:2}
+ TEST ln $brickpath/.shard/$gfid.1 $brickpath/.glusterfs/${new_gfid_str:0:2}/${new_gfid_str:2:2}/$new_gfid_str
+}
+assign_new_gfid $B0/$V0"1" file3
+assign_new_gfid $B0/$V0"2" file3
+
+#-------------------------------------------------------------------------------
+gfid_f4=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $B0/${V0}0/file4))
+TEST setfattr -n trusted.afr.$V0-client-1 -v 0x000000010000000100000000 $B0/${V0}0/.shard/$gfid_f4.1
+TEST setfattr -n trusted.afr.$V0-client-2 -v 0x000000010000000100000000 $B0/${V0}0/.shard/$gfid_f4.1
+TEST setfattr -n trusted.afr.$V0-client-0 -v 0x000000010000000100000000 $B0/${V0}1/.shard/$gfid_f4.1
+TEST setfattr -n trusted.afr.$V0-client-2 -v 0x000000010000000100000000 $B0/${V0}1/.shard/$gfid_f4.1
+TEST setfattr -n trusted.afr.$V0-client-0 -v 0x000000010000000100000000 $B0/${V0}2/.shard/$gfid_f4.1
+TEST setfattr -n trusted.afr.$V0-client-1 -v 0x000000010000000100000000 $B0/${V0}2/.shard/$gfid_f4.1
+
+#-------------------------------------------------------------------------------
+#Add entry to xattrop dir on first brick and check for split-brain.
+xattrop_dir0=$(afr_get_index_path $B0/$V0"0")
+base_entry_b0=`ls $xattrop_dir0`
+
+gfid_f1=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $B0/$V0"0"/file1))
+TEST ln $xattrop_dir0/$base_entry_b0 $xattrop_dir0/$gfid_f1
+
+gfid_f2_shard1=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $B0/$V0"0"/.shard/$gfid_f2.1))
+TEST ln $xattrop_dir0/$base_entry_b0 $xattrop_dir0/$gfid_f2_shard1
+
+gfid_f3=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $B0/${V0}0/file3))
+gfid_f3_shard1=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $B0/$V0"0"/.shard/$gfid_f3.1))
+TEST ln $xattrop_dir0/$base_entry_b0 $xattrop_dir0/$gfid_f3_shard1
+
+gfid_f4_shard1=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $B0/$V0"0"/.shard/$gfid_f4.1))
+TEST ln $xattrop_dir0/$base_entry_b0 $xattrop_dir0/$gfid_f4_shard1
+
+#-------------------------------------------------------------------------------
+#gfid split-brain won't show up in split-brain count.
+EXPECT "3" afr_get_split_brain_count $V0
+EXPECT_NOT "^0$" get_pending_heal_count $V0
+
+#Resolve split-brains
+TEST $CLI volume heal $V0 split-brain source-brick $H0:$B0/${V0}1 /file1
+GFIDSTR="gfid:$gfid_f2_shard1"
+TEST $CLI volume heal $V0 split-brain source-brick $H0:$B0/${V0}1 $GFIDSTR
+TEST $CLI volume heal $V0 split-brain source-brick $H0:$B0/${V0}1 /.shard/$gfid_f3.1
+TEST $CLI volume heal $V0 split-brain source-brick $H0:$B0/${V0}1 /.shard/$gfid_f4.1
+TEST $CLI volume heal $V0
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" get_pending_heal_count $V0
+cleanup;
diff --git a/xlators/cluster/afr/src/afr-common.c b/xlators/cluster/afr/src/afr-common.c
index eef7fd2..32127c6 100644
--- a/xlators/cluster/afr/src/afr-common.c
+++ b/xlators/cluster/afr/src/afr-common.c
@@ -2250,7 +2250,8 @@ afr_attempt_readsubvol_set(call_frame_t *frame, xlator_t *this,
if ((spb_choice >= 0) &&
(AFR_COUNT(success_replies, child_count) == child_count)) {
*read_subvol = spb_choice;
- } else if (!priv->quorum_count) {
+ } else if (!priv->quorum_count ||
+ frame->root->pid == GF_CLIENT_PID_GLFS_HEAL) {
*read_subvol = afr_first_up_child(frame, this);
} else if (priv->quorum_count &&
afr_has_quorum(data_readable, this, NULL)) {
diff --git a/xlators/features/shard/src/shard.c b/xlators/features/shard/src/shard.c
index 2e2ef5d..16d557b 100644
--- a/xlators/features/shard/src/shard.c
+++ b/xlators/features/shard/src/shard.c
@@ -1472,7 +1472,8 @@ int shard_lookup(call_frame_t *frame, xlator_t *this, loc_t *loc,
shard_local_t *local = NULL;
this->itable = loc->inode->table;
- if (frame->root->pid != GF_CLIENT_PID_GSYNCD) {
+ if ((frame->root->pid != GF_CLIENT_PID_GSYNCD) &&
+ (frame->root->pid != GF_CLIENT_PID_GLFS_HEAL)) {
SHARD_ENTRY_FOP_CHECK(loc, op_errno, err);
}
--
1.8.3.1

60
SOURCES/0407-geo-rep-Improving-help-message-in-schedule_georep.py.patch Normal file
View File

@ -0,0 +1,60 @@
From de31f2b0cb09a59941892c9981cb8a8b3aced9ec Mon Sep 17 00:00:00 2001
From: kshithijiyer <kshithij.ki@gmail.com>
Date: Tue, 24 Dec 2019 13:02:21 +0530
Subject: [PATCH 407/449] [geo-rep] Improving help message in
schedule_georep.py.in
SLAVE positional argument doesn't provide a clear
picture of what it is when compared to mastervol and slavevol
in schedule_georep.py.in. It would be better if we change it to
something like "Slave hostame (<username>@SLAVEHOST or SLAVEHOST)"
Present:
----------
positional arguments:
mastervol Master Volume Name
SLAVE SLAVEHOST or root@SLAVEHOST or user@SLAVEHOST
slavevol Slave Volume Name
Suggested:
-----------
positional arguments:
mastervol Master Volume Name
SLAVE Slave hostname (<username>@SLAVEHOST or SLAVEHOST)
slavevol Slave Volume Name
Backport of:
>Upstream Patch: https://review.gluster.org/#/c/glusterfs/+/23919/
>fixes: bz#1786276
>Change-Id: I73d52247997d623f77d55e51cbb6eccc08eb95ff
>Signed-off-by: kshithijiyer <kshithij.ki@gmail.com>
BUG: 1787994
Change-Id: I73d52247997d623f77d55e51cbb6eccc08eb95ff
Signed-off-by: kshithijiyer <kshithij.ki@gmail.com>
Signed-off-by: Sunny Kumar <sunkumar@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202454
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
extras/geo-rep/schedule_georep.py.in | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/extras/geo-rep/schedule_georep.py.in b/extras/geo-rep/schedule_georep.py.in
index f29ae02..ac93716 100644
--- a/extras/geo-rep/schedule_georep.py.in
+++ b/extras/geo-rep/schedule_georep.py.in
@@ -459,8 +459,8 @@ if __name__ == "__main__":
description=__doc__)
parser.add_argument("mastervol", help="Master Volume Name")
parser.add_argument("slave",
- help="SLAVEHOST or root@SLAVEHOST "
- "or user@SLAVEHOST",
+ help="Slave hostname "
+ "(<username>@SLAVEHOST or SLAVEHOST)",
metavar="SLAVE")
parser.add_argument("slavevol", help="Slave Volume Name")
parser.add_argument("--interval", help="Interval in Seconds. "
--
1.8.3.1

107
SOURCES/0408-geo-rep-Fix-ssh-port-validation.patch Normal file
View File

@ -0,0 +1,107 @@
From 07ab5a460da007fc3809b1a943614d1c7f5fcfef Mon Sep 17 00:00:00 2001
From: Sunny Kumar <sunkumar@redhat.com>
Date: Fri, 17 Jan 2020 11:03:46 +0000
Subject: [PATCH 408/449] geo-rep: Fix ssh-port validation
If non-standard ssh-port is used, Geo-rep can be configured to use ssh port
by using config option, the value should be in allowed port range and non negative.
At present it can accept negative value and outside allowed port range which is incorrect.
Many Linux kernels use the port range 32768 to 61000.
IANA suggests it should be in the range 1 to 2^16 - 1, so keeping the same.
$ gluster volume geo-replication master 127.0.0.1::slave config ssh-port -22
geo-replication config updated successfully
$ gluster volume geo-replication master 127.0.0.1::slave config ssh-port 22222222
geo-replication config updated successfully
This patch fixes the above issue and have added few validations around this
in test cases.
Backport of:
>Upstream Patch: https://review.gluster.org/#/c/glusterfs/+/24035/
>Change-Id: I9875ab3f00d7257370fbac6f5ed4356d2fed3f3c
>Fixes: bz#1792276
>Signed-off-by: Sunny Kumar <sunkumar@redhat.com>
BUG: 1796814
Change-Id: I9875ab3f00d7257370fbac6f5ed4356d2fed3f3c
Signed-off-by: Sunny Kumar <sunkumar@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202453
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
geo-replication/gsyncd.conf.in | 4 +++-
tests/00-geo-rep/00-georep-verify-non-root-setup.t | 16 ++++++++++++++++
tests/00-geo-rep/georep-basic-dr-rsync.t | 13 +++++++++++++
3 files changed, 32 insertions(+), 1 deletion(-)
diff --git a/geo-replication/gsyncd.conf.in b/geo-replication/gsyncd.conf.in
index 9155cd8..11e57fd 100644
--- a/geo-replication/gsyncd.conf.in
+++ b/geo-replication/gsyncd.conf.in
@@ -266,7 +266,9 @@ allowed_values=ERROR,INFO,WARNING,DEBUG
[ssh-port]
value=22
-validation=int
+validation=minmax
+min=1
+max=65535
help=Set SSH port
type=int
diff --git a/tests/00-geo-rep/00-georep-verify-non-root-setup.t b/tests/00-geo-rep/00-georep-verify-non-root-setup.t
index c9fd8b2..12f0c01 100644
--- a/tests/00-geo-rep/00-georep-verify-non-root-setup.t
+++ b/tests/00-geo-rep/00-georep-verify-non-root-setup.t
@@ -223,6 +223,22 @@ TEST $GEOREP_CLI $master $slave_url resume
#Validate failure of volume stop when geo-rep is running
TEST ! $CLI volume stop $GMV0
+#Negative test for ssh-port
+#Port should be integer and between 1-65535 range
+
+TEST ! $GEOREP_CLI $master $slave_url config ssh-port -22
+
+TEST ! $GEOREP_CLI $master $slave_url config ssh-port abc
+
+TEST ! $GEOREP_CLI $master $slave_url config ssh-port 6875943
+
+TEST ! $GEOREP_CLI $master $slave_url config ssh-port 4.5
+
+TEST ! $GEOREP_CLI $master $slave_url config ssh-port 22a
+
+#Config Set ssh-port to validate int validation
+TEST $GEOREP_CLI $master $slave config ssh-port 22
+
#Hybrid directory rename test BZ#1763439
TEST $GEOREP_CLI $master $slave_url config change_detector xsync
mkdir ${master_mnt}/dir1
diff --git a/tests/00-geo-rep/georep-basic-dr-rsync.t b/tests/00-geo-rep/georep-basic-dr-rsync.t
index b6fbf18..d785aa5 100644
--- a/tests/00-geo-rep/georep-basic-dr-rsync.t
+++ b/tests/00-geo-rep/georep-basic-dr-rsync.t
@@ -71,6 +71,19 @@ EXPECT_WITHIN $GEO_REP_TIMEOUT 4 check_status_num_rows "Created"
#Config gluster-command-dir
TEST $GEOREP_CLI $master $slave config gluster-command-dir ${GLUSTER_CMD_DIR}
+#Negative test for ssh-port
+#Port should be integer and between 1-65535 range
+
+TEST ! $GEOREP_CLI $master $slave config ssh-port -22
+
+TEST ! $GEOREP_CLI $master $slave config ssh-port abc
+
+TEST ! $GEOREP_CLI $master $slave config ssh-port 6875943
+
+TEST ! $GEOREP_CLI $master $slave config ssh-port 4.5
+
+TEST ! $GEOREP_CLI $master $slave config ssh-port 22a
+
#Config Set ssh-port to validate int validation
TEST $GEOREP_CLI $master $slave config ssh-port 22
--
1.8.3.1

52
SOURCES/0409-system-posix-acl-update-ctx-only-if-iatt-is-non-NULL.patch Normal file
View File

@ -0,0 +1,52 @@
From a92b4f6373cb18544325436cf86abfebd6780d79 Mon Sep 17 00:00:00 2001
From: Homma <homma@allworks.co.jp>
Date: Fri, 5 Jul 2019 16:10:41 +0530
Subject: [PATCH 409/449] system/posix-acl: update ctx only if iatt is non-NULL
We need to safe-guard against possible zero'ing out of iatt
structure in acl ctx, which can cause many issues.
> upstream patch: https://review.gluster.org/#/c/glusterfs/+/23003/
> fixes: 1668286
> Change-Id: Ie81a57d7453a6624078de3be8c0845bf4d432773
> Signed-off-by: Amar Tumballi <amarts@redhat.com>
BUG: 1781649
Change-Id: I655b61551d30215b9f23cafc3ef9a5c0d98a43d0
Signed-off-by: Raghavendra M <raghavendra@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202446
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/system/posix-acl/src/posix-acl.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/xlators/system/posix-acl/src/posix-acl.c b/xlators/system/posix-acl/src/posix-acl.c
index 38e48b8..c6ba281 100644
--- a/xlators/system/posix-acl/src/posix-acl.c
+++ b/xlators/system/posix-acl/src/posix-acl.c
@@ -875,6 +875,13 @@ posix_acl_ctx_update(inode_t *inode, xlator_t *this, struct iatt *buf,
int ret = 0;
int i = 0;
+ if (!buf || !buf->ia_ctime) {
+ /* No need to update ctx if buf is empty */
+ gf_log_callingfn(this->name, GF_LOG_DEBUG, "iatt struct is empty (%d)",
+ fop);
+ goto out;
+ }
+
LOCK(&inode->lock);
{
ctx = __posix_acl_ctx_get(inode, this, _gf_true);
@@ -928,6 +935,7 @@ posix_acl_ctx_update(inode_t *inode, xlator_t *this, struct iatt *buf,
}
unlock:
UNLOCK(&inode->lock);
+out:
return ret;
}
--
1.8.3.1

16
SOURCES/0386-afr-prevent-spurious-entry-heals-leading-to-gfid-spl.patch → SOURCES/0410-afr-prevent-spurious-entry-heals-leading-to-gfid-spl.patch
View File

@ -1,7 +1,7 @@
From 7689fbb4be83f0e0657ec2729c4d66ed341b5751 Mon Sep 17 00:00:00 2001
From 2b2eb846c49caba13ab92ec66af20292e7780fc1 Mon Sep 17 00:00:00 2001
From: Ravishankar N <ravishankar@redhat.com>
Date: Tue, 11 Feb 2020 14:34:48 +0530
Subject: [PATCH 386/392] afr: prevent spurious entry heals leading to gfid
Subject: [PATCH 410/449] afr: prevent spurious entry heals leading to gfid
split-brain
Problem:
@ -18,7 +18,7 @@ Fix:
Proceed with entry heal only when shd can connect to all bricks of the replica,
just like in data and metadata heal.
BUG: 1848893
BUG: 1804164
> Upstream patch:https://review.gluster.org/#/c/glusterfs/+/24109/
> fixes: bz#1801624
@ -27,10 +27,8 @@ BUG: 1848893
Change-Id: I23f57e543cff1e3f35eb8dbc60a2babfae6838c7
Signed-off-by: Ravishankar N <ravishankar@redhat.com>
(cherry picked from commit 2b2eb846c49caba13ab92ec66af20292e7780fc1)
Reviewed-on: https://code.engineering.redhat.com/gerrit/203692
Reviewed-on: https://code.engineering.redhat.com/gerrit/202395
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Tested-by: Karthik Subrahmanya <ksubrahm@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
.../bug-1433571-undo-pending-only-on-up-bricks.t | 18 ++-----
@ -142,10 +140,10 @@ index 0000000..94b4651
+EXPECT_WITHIN $UMOUNT_TIMEOUT "Y" force_umount $M0
+cleanup;
diff --git a/xlators/cluster/afr/src/afr-common.c b/xlators/cluster/afr/src/afr-common.c
index d6b70e9..939246e 100644
index 32127c6..5806556 100644
--- a/xlators/cluster/afr/src/afr-common.c
+++ b/xlators/cluster/afr/src/afr-common.c
@@ -6632,7 +6632,7 @@ afr_fav_child_reset_sink_xattrs(void *opaque)
@@ -6629,7 +6629,7 @@ afr_fav_child_reset_sink_xattrs(void *opaque)
ret = afr_selfheal_inodelk(heal_frame, this, inode, this->name, 0, 0,
locked_on);
{
@ -154,7 +152,7 @@ index d6b70e9..939246e 100644
goto data_unlock;
ret = __afr_selfheal_data_prepare(
heal_frame, this, inode, locked_on, sources, sinks,
@@ -6649,7 +6649,7 @@ afr_fav_child_reset_sink_xattrs(void *opaque)
@@ -6646,7 +6646,7 @@ afr_fav_child_reset_sink_xattrs(void *opaque)
ret = afr_selfheal_inodelk(heal_frame, this, inode, this->name,
LLONG_MAX - 1, 0, locked_on);
{

116
SOURCES/0411-tools-glusterfind-validate-session-name.patch Normal file
View File

@ -0,0 +1,116 @@
From 854defb4ff5e0d53f51545d20796aff662f9850f Mon Sep 17 00:00:00 2001
From: Saravanakumar Arumugam <sarumuga@redhat.com>
Date: Thu, 9 Jul 2015 15:56:28 +0530
Subject: [PATCH 411/449] tools/glusterfind : validate session name
Validate a session name(during create) for the following:
1. minimum 2 character length.
2. Maximum 256 characters.
3. No special characters apart from underscore, hyphen allowed.
Also, validate volume(expect, while using glusterfind list).
>Change-Id: I1b1e64e218f93d0a531d3cf69fc2ce7e2ed11d01
>BUG: 1241494
>Signed-off-by: Saravanakumar Arumugam <sarumuga@redhat.com>
>Signed-off-by: Shwetha K Acharya <sacharya@redhat.com>
backport of https://review.gluster.org/#/c/glusterfs/+/11602/
BUG: 1234220
Change-Id: I1b1e64e218f93d0a531d3cf69fc2ce7e2ed11d01
Signed-off-by: Shwetha K Acharya <sacharya@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202469
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tools/glusterfind/src/main.py | 50 ++++++++++++++++++++++++++++++++++++-------
1 file changed, 42 insertions(+), 8 deletions(-)
diff --git a/tools/glusterfind/src/main.py b/tools/glusterfind/src/main.py
index 5ca1fec..4b5466d 100644
--- a/tools/glusterfind/src/main.py
+++ b/tools/glusterfind/src/main.py
@@ -23,6 +23,7 @@ import tempfile
import signal
from datetime import datetime
import codecs
+import re
from utils import execute, is_host_local, mkdirp, fail
from utils import setup_logger, human_time, handle_rm_error
@@ -520,11 +521,8 @@ def write_output(outfile, outfilemerger, field_separator):
else:
gfind_write(f, row[0], field_separator, p_rep)
-def mode_create(session_dir, args):
- logger.debug("Init is called - Session: %s, Volume: %s"
- % (args.session, args.volume))
-
- cmd = ["gluster", 'volume', 'info', args.volume, "--xml"]
+def validate_volume(volume):
+ cmd = ["gluster", 'volume', 'info', volume, "--xml"]
_, data, _ = execute(cmd,
exit_msg="Failed to Run Gluster Volume Info",
logger=logger)
@@ -532,11 +530,42 @@ def mode_create(session_dir, args):
tree = etree.fromstring(data)
statusStr = tree.find('volInfo/volumes/volume/statusStr').text
except (ParseError, AttributeError) as e:
- fail("Invalid Volume: %s" % e, logger=logger)
-
+ fail("Invalid Volume: Check the Volume name! %s" % e)
if statusStr != "Started":
- fail("Volume %s is not online" % args.volume, logger=logger)
+ fail("Volume %s is not online" % volume)
+
+# The rules for a valid session name.
+SESSION_NAME_RULES = {
+ 'min_length': 2,
+ 'max_length': 256, # same as maximum volume length
+ # Specifies all alphanumeric characters, underscore, hyphen.
+ 'valid_chars': r'0-9a-zA-Z_-',
+}
+
+
+# checks valid session name, fail otherwise
+def validate_session_name(session):
+ # Check for minimum length
+ if len(session) < SESSION_NAME_RULES['min_length']:
+ fail('session_name must be at least ' +
+ str(SESSION_NAME_RULES['min_length']) + ' characters long.')
+ # Check for maximum length
+ if len(session) > SESSION_NAME_RULES['max_length']:
+ fail('session_name must not exceed ' +
+ str(SESSION_NAME_RULES['max_length']) + ' characters length.')
+
+ # Matches strings composed entirely of characters specified within
+ if not re.match(r'^[' + SESSION_NAME_RULES['valid_chars'] +
+ ']+$', session):
+ fail('Session name can only contain these characters: ' +
+ SESSION_NAME_RULES['valid_chars'])
+
+
+def mode_create(session_dir, args):
+ validate_session_name(args.session)
+ logger.debug("Init is called - Session: %s, Volume: %s"
+ % (args.session, args.volume))
mkdirp(session_dir, exit_on_err=True, logger=logger)
mkdirp(os.path.join(session_dir, args.volume), exit_on_err=True,
logger=logger)
@@ -850,6 +879,11 @@ def main():
args.mode not in ["create", "list", "query"]:
fail("Invalid session %s" % args.session)
+ # volume involved, validate the volume first
+ if args.mode not in ["list"]:
+ validate_volume(args.volume)
+
+
# "default" is a system defined session name
if args.mode in ["create", "post", "pre", "delete"] and \
args.session == "default":
--
1.8.3.1

46
SOURCES/0412-gluster-smb-add-smb-parameter-when-access-gluster-by.patch Normal file
View File

@ -0,0 +1,46 @@
From 0769c5ddc78ea37b9a43ac35dd71ec8cea4b8da8 Mon Sep 17 00:00:00 2001
From: yinkui <13965432176@163.com>
Date: Fri, 16 Aug 2019 10:15:07 +0800
Subject: [PATCH 412/449] gluster-smb:add smb parameter when access gluster by
cifs
Backport of https://review.gluster.org/23240
Change-Id: I9ff54f2ca6f86bb5b2f4740485a0159e1fd7785f
BUG: 1783232
Signed-off-by: yinkui <13965432176@163.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202472
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
extras/hook-scripts/set/post/S30samba-set.sh | 1 +
extras/hook-scripts/start/post/S30samba-start.sh | 1 +
2 files changed, 2 insertions(+)
diff --git a/extras/hook-scripts/set/post/S30samba-set.sh b/extras/hook-scripts/set/post/S30samba-set.sh
index d2a62d3..e73f00f 100755
--- a/extras/hook-scripts/set/post/S30samba-set.sh
+++ b/extras/hook-scripts/set/post/S30samba-set.sh
@@ -90,6 +90,7 @@ function add_samba_share () {
STRING+="path = /\n"
STRING+="read only = no\n"
STRING+="guest ok = yes\n"
+ STRING+="kernel share modes = no\n"
printf "$STRING" >> ${CONFIGFILE}
}
diff --git a/extras/hook-scripts/start/post/S30samba-start.sh b/extras/hook-scripts/start/post/S30samba-start.sh
index 2854bdd..0d5a5ed 100755
--- a/extras/hook-scripts/start/post/S30samba-start.sh
+++ b/extras/hook-scripts/start/post/S30samba-start.sh
@@ -89,6 +89,7 @@ function add_samba_share () {
STRING+="path = /\n"
STRING+="read only = no\n"
STRING+="guest ok = yes\n"
+ STRING+="kernel share modes = no\n"
printf "$STRING" >> "${CONFIGFILE}"
}
--
1.8.3.1

46
SOURCES/0413-extras-hooks-Remove-smb.conf-parameter-allowing-gues.patch Normal file
View File

@ -0,0 +1,46 @@
From aec3dd00fa76547316fddd07e6ded428d945986c Mon Sep 17 00:00:00 2001
From: Anoop C S <anoopcs@redhat.com>
Date: Fri, 22 Nov 2019 17:36:55 +0530
Subject: [PATCH 413/449] extras/hooks: Remove smb.conf parameter allowing
guest access
Backport of https://review.gluster.org/23745
Change-Id: I88f494f16153d27ab6e2f2faf4d557e075671b10
BUG: 1775637
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202473
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
extras/hook-scripts/set/post/S30samba-set.sh | 1 -
extras/hook-scripts/start/post/S30samba-start.sh | 1 -
2 files changed, 2 deletions(-)
diff --git a/extras/hook-scripts/set/post/S30samba-set.sh b/extras/hook-scripts/set/post/S30samba-set.sh
index e73f00f..854f131 100755
--- a/extras/hook-scripts/set/post/S30samba-set.sh
+++ b/extras/hook-scripts/set/post/S30samba-set.sh
@@ -89,7 +89,6 @@ function add_samba_share () {
STRING+="glusterfs:loglevel = 7\n"
STRING+="path = /\n"
STRING+="read only = no\n"
- STRING+="guest ok = yes\n"
STRING+="kernel share modes = no\n"
printf "$STRING" >> ${CONFIGFILE}
}
diff --git a/extras/hook-scripts/start/post/S30samba-start.sh b/extras/hook-scripts/start/post/S30samba-start.sh
index 0d5a5ed..cac0cbf 100755
--- a/extras/hook-scripts/start/post/S30samba-start.sh
+++ b/extras/hook-scripts/start/post/S30samba-start.sh
@@ -88,7 +88,6 @@ function add_samba_share () {
STRING+="glusterfs:loglevel = 7\n"
STRING+="path = /\n"
STRING+="read only = no\n"
- STRING+="guest ok = yes\n"
STRING+="kernel share modes = no\n"
printf "$STRING" >> "${CONFIGFILE}"
}
--
1.8.3.1

28
SOURCES/0391-cluster-syncop-avoid-duplicate-unlock-of-inodelk-ent.patch → SOURCES/0414-cluster-syncop-avoid-duplicate-unlock-of-inodelk-ent.patch
View File

@ -1,23 +1,31 @@
From e9cb714d66a7926a746b8cd5f9288d59aefee918 Mon Sep 17 00:00:00 2001
From 5b549cbf3f1873054c6d187b09aa9f9313971b1f Mon Sep 17 00:00:00 2001
From: Kinglong Mee <kinglongmee@gmail.com>
Date: Mon, 18 Mar 2019 20:47:54 +0800
Subject: [PATCH 391/392] cluster-syncop: avoid duplicate unlock of
Subject: [PATCH 414/449] cluster-syncop: avoid duplicate unlock of
inodelk/entrylk
When using ec, there are many messages at brick log as,
When using ec, there are many spam messages in brick and client
logs files.
[inodelk.c:514:__inode_unlock_lock] 0-test-locks: Matching lock not found for unlock 0-9223372036854775807, lo=68e040a84b7f0000 on 0x7f208c006f78
[MSGID: 115053] [server-rpc-fops_v2.c:280:server4_inodelk_cbk] 0-test-server: 2557439: INODELK <gfid:df4e41be-723f-4289-b7af-b4272b3e880c> (df4e41be-723f-4289-b7af-b4272b3e880c), client: CTX_ID:67d4a7f3-605a-4965-89a5-31309d62d1fa-GRAPH_ID:0-PID:1659-HOST:openfs-node2-PC_NAME:test-client-1-RECON_NO:-28, error-xlator: test-locks [Invalid argument]
When shd does entry heal, it takes lock on a directory using
cluster_tiebreaker_inodelk(). If it does not get lock on all
the bricks because other clients has got lock on some bricks,
it will unlock the locks on those bricks which it got and then
will try blocking locks (If any one of the previous was successful).
The problem come here. In case we do not get locks on all the
required bricks, we are sending unlock request twice on those
bricks where we got the locks.
BUG: 1750211
> Upstream patch: https://review.gluster.org/#/c/glusterfs/+/22377/
> Change-Id: Ib164d29ebb071f620a4ca9679c4345ef7c88512a
> Updates: bz#1689920
> Signed-off-by: Kinglong Mee <mijinlong@open-fs.com>
> Reviewed on upstream link https://review.gluster.org/#/c/glusterfs/+/22377/
BUG: 1848890
Change-Id: Ib164d29ebb071f620a4ca9679c4345ef7c88512a
Signed-off-by: Sheetal Pamecha <spamecha@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/203852
Change-Id: I1647548ba75fdd27fd4e20dec08db67774f43375
Reviewed-on: https://code.engineering.redhat.com/gerrit/202477
Tested-by: Ashish Pandey <aspandey@redhat.com>
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---

523
SOURCES/0415-dht-Fix-stale-layout-and-create-issue.patch Normal file
View File

@ -0,0 +1,523 @@
From ba23e6d8f4eff11a228816149a8a1ccd6df41146 Mon Sep 17 00:00:00 2001
From: Susant Palai <spalai@redhat.com>
Date: Fri, 27 Dec 2019 12:06:19 +0530
Subject: [PATCH 415/449] dht: Fix stale-layout and create issue
Problem: With lookup-optimize set to on by default, a client with
stale-layout can create a new file on a wrong subvol. This will lead to
possible duplicate files if two different clients attempt to create the
same file with two different layouts.
Solution: Send in-memory layout to be cross checked at posix before
commiting a "create". In case of a mismatch, sync the client layout with
that of the server and attempt the create fop one more time.
test: Manual, testcase(attached)
(Backport of https://review.gluster.org/#/c/glusterfs/+/23927/)
BUG: 1748865
Change-Id: I6c82c97418654ae8eb3b81ab65f1247aa4002ceb
Signed-off-by: Susant Palai <spalai@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202465
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/bugs/distribute/bug-1786679.t | 69 +++++++++++
xlators/cluster/dht/src/dht-common.c | 147 ++++++++++++++++++++---
xlators/cluster/dht/src/dht-common.h | 6 +
xlators/protocol/client/src/client-rpc-fops_v2.c | 9 +-
xlators/storage/posix/src/posix-entry-ops.c | 29 ++++-
xlators/storage/posix/src/posix-helpers.c | 76 ++++++++++++
xlators/storage/posix/src/posix.h | 4 +
7 files changed, 321 insertions(+), 19 deletions(-)
create mode 100755 tests/bugs/distribute/bug-1786679.t
diff --git a/tests/bugs/distribute/bug-1786679.t b/tests/bugs/distribute/bug-1786679.t
new file mode 100755
index 0000000..219ce51
--- /dev/null
+++ b/tests/bugs/distribute/bug-1786679.t
@@ -0,0 +1,69 @@
+#!/bin/bash
+
+SCRIPT_TIMEOUT=250
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../dht.rc
+
+
+# create 2 subvols
+# create a dir
+# create a file
+# change layout
+# remove the file
+# execute create from a different mount
+# Without the patch, the file will be present on both of the bricks
+
+cleanup
+
+function get_layout () {
+
+layout=`getfattr -n trusted.glusterfs.dht -e hex $1 2>&1 | grep dht | gawk -F"=" '{print $2}'`
+
+echo $layout
+
+}
+
+function set_layout()
+{
+ setfattr -n "trusted.glusterfs.dht" -v $1 $2
+}
+
+TEST glusterd
+TEST pidof glusterd
+
+BRICK1=$B0/${V0}-0
+BRICK2=$B0/${V0}-1
+
+TEST $CLI volume create $V0 $H0:$BRICK1 $H0:$BRICK2
+TEST $CLI volume start $V0
+
+# Mount FUSE and create symlink
+TEST glusterfs -s $H0 --volfile-id $V0 $M0
+TEST mkdir $M0/dir
+TEST touch $M0/dir/file
+TEST ! stat "$BRICK1/dir/file"
+TEST stat "$BRICK2/dir/file"
+
+layout1="$(get_layout "$BRICK1/dir")"
+layout2="$(get_layout "$BRICK2/dir")"
+
+TEST set_layout $layout1 "$BRICK2/dir"
+TEST set_layout $layout2 "$BRICK1/dir"
+
+TEST rm $M0/dir/file -f
+TEST gluster v set $V0 client-log-level DEBUG
+
+#Without the patch in place, this client will create the file in $BRICK2
+#which will lead to two files being on both the bricks when a new client
+#create the file with the same name
+TEST touch $M0/dir/file
+
+TEST glusterfs -s $H0 --volfile-id $V0 $M1
+TEST touch $M1/dir/file
+
+TEST stat "$BRICK1/dir/file"
+TEST ! stat "$BRICK2/dir/file"
+
+cleanup
diff --git a/xlators/cluster/dht/src/dht-common.c b/xlators/cluster/dht/src/dht-common.c
index 7890e7a..6aa18f3 100644
--- a/xlators/cluster/dht/src/dht-common.c
+++ b/xlators/cluster/dht/src/dht-common.c
@@ -8262,6 +8262,11 @@ dht_create_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
xlator_t *prev = NULL;
int ret = -1;
dht_local_t *local = NULL;
+ gf_boolean_t parent_layout_changed = _gf_false;
+ char pgfid[GF_UUID_BUF_SIZE] = {0};
+ xlator_t *subvol = NULL;
+
+ local = frame->local;
local = frame->local;
if (!local) {
@@ -8270,8 +8275,69 @@ dht_create_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
goto out;
}
- if (op_ret == -1)
+ if (op_ret == -1) {
+ local->op_errno = op_errno;
+ parent_layout_changed = (xdata &&
+ dict_get(xdata, GF_PREOP_CHECK_FAILED))
+ ? _gf_true
+ : _gf_false;
+
+ if (parent_layout_changed) {
+ if (local && local->lock[0].layout.parent_layout.locks) {
+ /* Returning failure as the layout could not be fixed even under
+ * the lock */
+ goto out;
+ }
+
+ gf_uuid_unparse(local->loc.parent->gfid, pgfid);
+ gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_PARENT_LAYOUT_CHANGED,
+ "create (%s/%s) (path: %s): parent layout "
+ "changed. Attempting a layout refresh and then a "
+ "retry",
+ pgfid, local->loc.name, local->loc.path);
+
+ /*
+ dht_refresh_layout needs directory info in local->loc.Hence,
+ storing the parent_loc in local->loc and storing the create
+ context in local->loc2. We will restore this information in
+ dht_creation_do.
+ */
+
+ loc_wipe(&local->loc2);
+
+ ret = loc_copy(&local->loc2, &local->loc);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_NO_MEMORY,
+ "loc_copy failed %s", local->loc.path);
+
+ goto out;
+ }
+
+ loc_wipe(&local->loc);
+
+ ret = dht_build_parent_loc(this, &local->loc, &local->loc2,
+ &op_errno);
+
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_LOC_FAILED,
+ "parent loc build failed");
+ goto out;
+ }
+
+ subvol = dht_subvol_get_hashed(this, &local->loc2);
+
+ ret = dht_create_lock(frame, subvol);
+ if (ret < 0) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_INODE_LK_ERROR,
+ "locking parent failed");
+ goto out;
+ }
+
+ return 0;
+ }
+
goto out;
+ }
prev = cookie;
@@ -8392,6 +8458,8 @@ dht_create_wind_to_avail_subvol(call_frame_t *frame, xlator_t *this,
gf_msg_debug(this->name, 0, "creating %s on %s", loc->path,
subvol->name);
+ dht_set_parent_layout_in_dict(loc, this, local);
+
STACK_WIND_COOKIE(frame, dht_create_cbk, subvol, subvol,
subvol->fops->create, loc, flags, mode, umask, fd,
params);
@@ -8400,10 +8468,6 @@ dht_create_wind_to_avail_subvol(call_frame_t *frame, xlator_t *this,
avail_subvol = dht_free_disk_available_subvol(this, subvol, local);
if (avail_subvol != subvol) {
- local->params = dict_ref(params);
- local->flags = flags;
- local->mode = mode;
- local->umask = umask;
local->cached_subvol = avail_subvol;
local->hashed_subvol = subvol;
@@ -8419,6 +8483,8 @@ dht_create_wind_to_avail_subvol(call_frame_t *frame, xlator_t *this,
gf_msg_debug(this->name, 0, "creating %s on %s", loc->path,
subvol->name);
+ dht_set_parent_layout_in_dict(loc, this, local);
+
STACK_WIND_COOKIE(frame, dht_create_cbk, subvol, subvol,
subvol->fops->create, loc, flags, mode, umask, fd,
params);
@@ -8680,6 +8746,60 @@ err:
}
int
+dht_set_parent_layout_in_dict(loc_t *loc, xlator_t *this, dht_local_t *local)
+{
+ dht_conf_t *conf = this->private;
+ dht_layout_t *parent_layout = NULL;
+ int *parent_disk_layout = NULL;
+ xlator_t *hashed_subvol = NULL;
+ char pgfid[GF_UUID_BUF_SIZE] = {0};
+ int ret = 0;
+
+ gf_uuid_unparse(loc->parent->gfid, pgfid);
+
+ parent_layout = dht_layout_get(this, loc->parent);
+ hashed_subvol = dht_subvol_get_hashed(this, loc);
+
+ ret = dht_disk_layout_extract_for_subvol(this, parent_layout, hashed_subvol,
+ &parent_disk_layout);
+ if (ret == -1) {
+ gf_msg(this->name, GF_LOG_WARNING, local->op_errno,
+ DHT_MSG_PARENT_LAYOUT_CHANGED,
+ "%s (%s/%s) (path: %s): "
+ "extracting in-memory layout of parent failed. ",
+ gf_fop_list[local->fop], pgfid, loc->name, loc->path);
+ goto err;
+ }
+
+ ret = dict_set_str_sizen(local->params, GF_PREOP_PARENT_KEY,
+ conf->xattr_name);
+ if (ret < 0) {
+ gf_msg(this->name, GF_LOG_WARNING, local->op_errno,
+ DHT_MSG_PARENT_LAYOUT_CHANGED,
+ "%s (%s/%s) (path: %s): "
+ "setting %s key in params dictionary failed. ",
+ gf_fop_list[local->fop], pgfid, loc->name, loc->path,
+ GF_PREOP_PARENT_KEY);
+ goto err;
+ }
+
+ ret = dict_set_bin(local->params, conf->xattr_name, parent_disk_layout,
+ 4 * 4);
+ if (ret < 0) {
+ gf_msg(this->name, GF_LOG_WARNING, local->op_errno,
+ DHT_MSG_PARENT_LAYOUT_CHANGED,
+ "%s (%s/%s) (path: %s): "
+ "setting parent-layout in params dictionary failed. ",
+ gf_fop_list[local->fop], pgfid, loc->name, loc->path);
+ goto err;
+ }
+
+err:
+ dht_layout_unref(this, parent_layout);
+ return ret;
+}
+
+int
dht_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
mode_t mode, mode_t umask, fd_t *fd, dict_t *params)
{
@@ -8705,6 +8825,11 @@ dht_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
goto err;
}
+ local->params = dict_ref(params);
+ local->flags = flags;
+ local->mode = mode;
+ local->umask = umask;
+
if (dht_filter_loc_subvol_key(this, loc, &local->loc, &subvol)) {
gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_SUBVOL_INFO,
"creating %s on %s (got create on %s)", local->loc.path,
@@ -8720,10 +8845,6 @@ dht_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
if (hashed_subvol && (hashed_subvol != subvol)) {
/* Create the linkto file and then the data file */
- local->params = dict_ref(params);
- local->flags = flags;
- local->mode = mode;
- local->umask = umask;
local->cached_subvol = subvol;
local->hashed_subvol = hashed_subvol;
@@ -8736,6 +8857,9 @@ dht_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
* file as we expect a lookup everywhere if there are problems
* with the parent layout
*/
+
+ dht_set_parent_layout_in_dict(loc, this, local);
+
STACK_WIND_COOKIE(frame, dht_create_cbk, subvol, subvol,
subvol->fops->create, &local->loc, flags, mode, umask,
fd, params);
@@ -8787,11 +8911,6 @@ dht_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
goto err;
}
- local->params = dict_ref(params);
- local->flags = flags;
- local->mode = mode;
- local->umask = umask;
-
loc_wipe(&local->loc);
ret = dht_build_parent_loc(this, &local->loc, loc, &op_errno);
diff --git a/xlators/cluster/dht/src/dht-common.h b/xlators/cluster/dht/src/dht-common.h
index 8e65111..1b3e826 100644
--- a/xlators/cluster/dht/src/dht-common.h
+++ b/xlators/cluster/dht/src/dht-common.h
@@ -1549,4 +1549,10 @@ dht_check_remote_fd_failed_error(dht_local_t *local, int op_ret, int op_errno);
int
dht_dir_layout_error_check(xlator_t *this, inode_t *inode);
+int32_t
+dht_create_lock(call_frame_t *frame, xlator_t *subvol);
+
+int
+dht_set_parent_layout_in_dict(loc_t *loc, xlator_t *this, dht_local_t *local);
+
#endif /* _DHT_H */
diff --git a/xlators/protocol/client/src/client-rpc-fops_v2.c b/xlators/protocol/client/src/client-rpc-fops_v2.c
index 2673b6e..613dda8 100644
--- a/xlators/protocol/client/src/client-rpc-fops_v2.c
+++ b/xlators/protocol/client/src/client-rpc-fops_v2.c
@@ -2094,11 +2094,12 @@ client4_0_create_cbk(struct rpc_req *req, struct iovec *iov, int count,
goto out;
}
+ ret = client_post_create_v2(this, &rsp, &stbuf, &preparent, &postparent,
+ local, &xdata);
+ if (ret < 0)
+ goto out;
+
if (-1 != rsp.op_ret) {
- ret = client_post_create_v2(this, &rsp, &stbuf, &preparent, &postparent,
- local, &xdata);
- if (ret < 0)
- goto out;
ret = client_add_fd_to_saved_fds(frame->this, fd, &local->loc,
local->flags, rsp.fd, 0);
if (ret) {
diff --git a/xlators/storage/posix/src/posix-entry-ops.c b/xlators/storage/posix/src/posix-entry-ops.c
index bea0bbf..65650b3 100644
--- a/xlators/storage/posix/src/posix-entry-ops.c
+++ b/xlators/storage/posix/src/posix-entry-ops.c
@@ -2070,6 +2070,8 @@ posix_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
gf_boolean_t entry_created = _gf_false, gfid_set = _gf_false;
mode_t mode_bit = 0;
+ dict_t *xdata_rsp = dict_ref(xdata);
+
DECLARE_OLD_FS_ID_VAR;
VALIDATE_OR_GOTO(frame, out);
@@ -2118,6 +2120,28 @@ posix_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
was_present = 0;
}
+ if (!was_present) {
+ if (posix_is_layout_stale(xdata, par_path, this)) {
+ op_ret = -1;
+ op_errno = EIO;
+ if (!xdata_rsp) {
+ xdata_rsp = dict_new();
+ if (!xdata_rsp) {
+ op_errno = ENOMEM;
+ goto out;
+ }
+ }
+
+ if (dict_set_int32_sizen(xdata_rsp, GF_PREOP_CHECK_FAILED, 1) ==
+ -1) {
+ gf_msg(this->name, GF_LOG_ERROR, errno, P_MSG_DICT_SET_FAILED,
+ "setting key %s in dict failed", GF_PREOP_CHECK_FAILED);
+ }
+
+ goto out;
+ }
+ }
+
if (priv->o_direct)
_flags |= O_DIRECT;
@@ -2239,7 +2263,10 @@ out:
STACK_UNWIND_STRICT(create, frame, op_ret, op_errno, fd,
(loc) ? loc->inode : NULL, &stbuf, &preparent,
- &postparent, xdata);
+ &postparent, xdata_rsp);
+
+ if (xdata_rsp)
+ dict_unref(xdata_rsp);
return 0;
}
diff --git a/xlators/storage/posix/src/posix-helpers.c b/xlators/storage/posix/src/posix-helpers.c
index 35dd3b6..2c27d22 100644
--- a/xlators/storage/posix/src/posix-helpers.c
+++ b/xlators/storage/posix/src/posix-helpers.c
@@ -3559,3 +3559,79 @@ posix_update_iatt_buf(struct iatt *buf, int fd, char *loc, dict_t *xattr_req)
}
}
}
+
+gf_boolean_t
+posix_is_layout_stale(dict_t *xdata, char *par_path, xlator_t *this)
+{
+ int op_ret = 0;
+ ssize_t size = 0;
+ char value_buf[4096] = {
+ 0,
+ };
+ gf_boolean_t have_val = _gf_false;
+ data_t *arg_data = NULL;
+ char *xattr_name = NULL;
+ gf_boolean_t is_stale = _gf_false;
+
+ op_ret = dict_get_str_sizen(xdata, GF_PREOP_PARENT_KEY, &xattr_name);
+ if (xattr_name == NULL) {
+ op_ret = 0;
+ goto out;
+ }
+
+ arg_data = dict_get(xdata, xattr_name);
+ if (!arg_data) {
+ op_ret = 0;
+ goto out;
+ }
+
+ size = sys_lgetxattr(par_path, xattr_name, value_buf,
+ sizeof(value_buf) - 1);
+
+ if (size >= 0) {
+ have_val = _gf_true;
+ } else {
+ if (errno == ERANGE) {
+ gf_msg(this->name, GF_LOG_INFO, errno, P_MSG_PREOP_CHECK_FAILED,
+ "getxattr on key (%s) path (%s) failed due to"
+ " buffer overflow",
+ xattr_name, par_path);
+ size = sys_lgetxattr(par_path, xattr_name, NULL, 0);
+ }
+ if (size < 0) {
+ op_ret = -1;
+ gf_msg(this->name, GF_LOG_ERROR, errno, P_MSG_PREOP_CHECK_FAILED,
+ "getxattr on key (%s) failed, path : %s", xattr_name,
+ par_path);
+ goto out;
+ }
+ }
+
+ if (!have_val) {
+ size = sys_lgetxattr(par_path, xattr_name, value_buf, size);
+ if (size < 0) {
+ gf_msg(this->name, GF_LOG_ERROR, errno, P_MSG_PREOP_CHECK_FAILED,
+ "getxattr on key (%s) failed (%s)", xattr_name,
+ strerror(errno));
+ goto out;
+ }
+ }
+
+ if ((arg_data->len != size) || (memcmp(arg_data->data, value_buf, size))) {
+ gf_msg(this->name, GF_LOG_INFO, EIO, P_MSG_PREOP_CHECK_FAILED,
+ "failing preop as on-disk xattr value differs from argument "
+ "value for key %s",
+ xattr_name);
+ op_ret = -1;
+ }
+
+out:
+ dict_del_sizen(xdata, xattr_name);
+ dict_del_sizen(xdata, GF_PREOP_PARENT_KEY);
+
+ if (op_ret == -1) {
+ is_stale = _gf_true;
+ }
+
+ return is_stale;
+}
diff --git a/xlators/storage/posix/src/posix.h b/xlators/storage/posix/src/posix.h
index dd51062..ac9d83c 100644
--- a/xlators/storage/posix/src/posix.h
+++ b/xlators/storage/posix/src/posix.h
@@ -671,4 +671,8 @@ posix_spawn_ctx_janitor_thread(xlator_t *this);
void
posix_update_iatt_buf(struct iatt *buf, int fd, char *loc, dict_t *xdata);
+
+gf_boolean_t
+posix_is_layout_stale(dict_t *xdata, char *par_path, xlator_t *this);
+
#endif /* _POSIX_H */
--
1.8.3.1

72
SOURCES/0416-tests-fix-spurious-failure-of-bug-1402841.t-mt-dir-s.patch Normal file
View File

@ -0,0 +1,72 @@
From 63cfdd987b1dfbf97486f0f884380faee0ae25d0 Mon Sep 17 00:00:00 2001
From: Ravishankar N <ravishankar@redhat.com>
Date: Wed, 4 Sep 2019 11:27:30 +0530
Subject: [PATCH 416/449] tests: fix spurious failure of
bug-1402841.t-mt-dir-scan-race.t
Upstream patch: https://review.gluster.org/23352
Problem:
Since commit 600ba94183333c4af9b4a09616690994fd528478, shd starts
healing as soon as it is toggled from disabled to enabled. This was
causing the following line in the .t to fail on a 'fast' machine (always
on my laptop and sometimes on the jenkins slaves).
EXPECT_NOT "^0$" get_pending_heal_count $V0
because by the time shd was disabled, the heal was already completed.
Fix:
Increase the no. of files to be healed and make it a variable called
FILE_COUNT, should we need to bump it up further because the machines
become even faster. Also created pending metadata heals to increase the
time taken to heal a file.
>fixes: bz#1748744
>Change-Id: I5a26b08e45b8c19bce3c01ce67bdcc28ed48198d
Signed-off-by: Ravishankar N <ravishankar@redhat.com>
BUG: 1844359
Change-Id: Ie3676c6c2c27e7574b958d2eaac23801dfaed3a9
Reviewed-on: https://code.engineering.redhat.com/gerrit/202481
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/bugs/core/bug-1402841.t-mt-dir-scan-race.t | 9 +++++----
1 file changed, 5 insertions(+), 4 deletions(-)
diff --git a/tests/bugs/core/bug-1402841.t-mt-dir-scan-race.t b/tests/bugs/core/bug-1402841.t-mt-dir-scan-race.t
index 6351ba2..a1b9a85 100755
--- a/tests/bugs/core/bug-1402841.t-mt-dir-scan-race.t
+++ b/tests/bugs/core/bug-1402841.t-mt-dir-scan-race.t
@@ -3,6 +3,8 @@
. $(dirname $0)/../../volume.rc
cleanup;
+FILE_COUNT=500
+
TEST glusterd
TEST pidof glusterd
TEST $CLI volume create $V0 replica 2 $H0:$B0/${V0}{0,1}
@@ -11,15 +13,14 @@ TEST $CLI volume set $V0 cluster.shd-wait-qlength 100
TEST $CLI volume start $V0
TEST glusterfs --volfile-id=$V0 --volfile-server=$H0 $M0;
-touch $M0/file{1..200}
-
+for i in `seq 1 $FILE_COUNT`; do touch $M0/file$i; done
TEST kill_brick $V0 $H0 $B0/${V0}1
-for i in {1..200}; do echo hello>$M0/file$i; done
+for i in `seq 1 $FILE_COUNT`; do echo hello>$M0/file$i; chmod -x $M0/file$i; done
TEST $CLI volume start $V0 force
EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V0 $H0 $B0/${V0}1
EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 1
-EXPECT "200" get_pending_heal_count $V0
+EXPECT "$FILE_COUNT" get_pending_heal_count $V0
TEST $CLI volume set $V0 self-heal-daemon on
EXPECT_WITHIN $PROCESS_UP_TIMEOUT "Y" glustershd_up_status
--
1.8.3.1

153
SOURCES/0417-events-fix-IPv6-memory-corruption.patch Normal file
View File

@ -0,0 +1,153 @@
From 5e231ceb35bb763d6fafc7c3efe1c3c582929cc2 Mon Sep 17 00:00:00 2001
From: Xavi Hernandez <xhernandez@redhat.com>
Date: Tue, 14 Jan 2020 13:28:47 +0100
Subject: [PATCH 417/449] events: fix IPv6 memory corruption
When an event was generated and the target host was resolved to an IPv6
address, there was a memory overflow when that address was copied to a
fixed IPv4 structure (IPv6 addresses are longer than IPv4 ones).
This fix correctly handles IPv4 and IPv6 addresses returned by
getaddrinfo()
Backport of:
> Upstream-patch-link: https://review.gluster.org/24014
> Change-Id: I5864a0c6e6f1b405bd85988529570140cf23b250
> Fixes: bz#1790870
> Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
BUG: 1792873
Change-Id: I5864a0c6e6f1b405bd85988529570140cf23b250
Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202486
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
libglusterfs/src/events.c | 56 +++++++++++++----------------------------------
1 file changed, 15 insertions(+), 41 deletions(-)
diff --git a/libglusterfs/src/events.c b/libglusterfs/src/events.c
index 4e2f8f9..6d1e383 100644
--- a/libglusterfs/src/events.c
+++ b/libglusterfs/src/events.c
@@ -34,7 +34,6 @@ _gf_event(eventtypes_t event, const char *fmt, ...)
int ret = 0;
int sock = -1;
char *eventstr = NULL;
- struct sockaddr_in server;
va_list arguments;
char *msg = NULL;
glusterfs_ctx_t *ctx = NULL;
@@ -42,11 +41,10 @@ _gf_event(eventtypes_t event, const char *fmt, ...)
struct addrinfo hints;
struct addrinfo *result = NULL;
xlator_t *this = THIS;
- int sin_family = AF_INET;
char *volfile_server_transport = NULL;
/* Global context */
- ctx = THIS->ctx;
+ ctx = this->ctx;
if (event < 0 || event >= EVENT_LAST) {
ret = EVENT_ERROR_INVALID_INPUTS;
@@ -60,48 +58,31 @@ _gf_event(eventtypes_t event, const char *fmt, ...)
goto out;
}
- memset(&hints, 0, sizeof(hints));
- hints.ai_family = AF_UNSPEC;
-
if (ctx) {
volfile_server_transport = ctx->cmd_args.volfile_server_transport;
}
-
if (!volfile_server_transport) {
volfile_server_transport = "tcp";
}
- /* Get Host name to send message */
+
+ /* host = NULL returns localhost */
+ host = NULL;
if (ctx && ctx->cmd_args.volfile_server &&
(strcmp(volfile_server_transport, "unix"))) {
/* If it is client code then volfile_server is set
use that information to push the events. */
- if ((getaddrinfo(ctx->cmd_args.volfile_server, NULL, &hints,
- &result)) != 0) {
- ret = EVENT_ERROR_RESOLVE;
- goto out;
- }
-
- if (get_ip_from_addrinfo(result, &host) == NULL) {
- ret = EVENT_ERROR_RESOLVE;
- goto out;
- }
-
- sin_family = result->ai_family;
- } else {
- /* Localhost, Use the defined IP for localhost */
- host = gf_strdup(EVENT_HOST);
+ host = ctx->cmd_args.volfile_server;
}
- /* Socket Configurations */
- server.sin_family = sin_family;
- server.sin_port = htons(EVENT_PORT);
- ret = inet_pton(server.sin_family, host, &server.sin_addr);
- if (ret <= 0) {
- gf_msg(this->name, GF_LOG_ERROR, EINVAL, LG_MSG_INVALID_ARG,
- "inet_pton failed with return code %d", ret);
+ memset(&hints, 0, sizeof(hints));
+ hints.ai_family = AF_UNSPEC;
+ hints.ai_socktype = SOCK_DGRAM;
+ hints.ai_flags = AI_ADDRCONFIG;
+
+ if ((getaddrinfo(host, TOSTRING(EVENT_PORT), &hints, &result)) != 0) {
+ ret = EVENT_ERROR_RESOLVE;
goto out;
}
- memset(&server.sin_zero, '\0', sizeof(server.sin_zero));
va_start(arguments, fmt);
ret = gf_vasprintf(&msg, fmt, arguments);
@@ -113,15 +94,15 @@ _gf_event(eventtypes_t event, const char *fmt, ...)
}
ret = gf_asprintf(&eventstr, "%u %d %s", (unsigned)time(NULL), event, msg);
-
+ GF_FREE(msg);
if (ret <= 0) {
ret = EVENT_ERROR_MSG_FORMAT;
goto out;
}
/* Send Message */
- if (sendto(sock, eventstr, strlen(eventstr), 0, (struct sockaddr *)&server,
- sizeof(server)) <= 0) {
+ if (sendto(sock, eventstr, strlen(eventstr), 0, result->ai_addr,
+ result->ai_addrlen) <= 0) {
ret = EVENT_ERROR_SEND;
goto out;
}
@@ -133,17 +114,10 @@ out:
sys_close(sock);
}
- /* Allocated by gf_vasprintf */
- if (msg)
- GF_FREE(msg);
-
/* Allocated by gf_asprintf */
if (eventstr)
GF_FREE(eventstr);
- if (host)
- GF_FREE(host);
-
if (result)
freeaddrinfo(result);
--
1.8.3.1

439
SOURCES/0418-md-cache-avoid-clearing-cache-when-not-necessary.patch Normal file
View File

@ -0,0 +1,439 @@
From 7ad8c03a28fca67150972cda964ebe9233766b54 Mon Sep 17 00:00:00 2001
From: Xavi Hernandez <xhernandez@redhat.com>
Date: Mon, 30 Mar 2020 11:09:39 +0200
Subject: [PATCH 418/449] md-cache: avoid clearing cache when not necessary
mdc_inode_xatt_set() blindly cleared current cache when dict was not
NULL, even if there was no xattr requested.
This patch fixes this by only calling mdc_inode_xatt_set() when we have
explicitly requested something to cache.
Backport of:
> Upstream-patch-link: https://review.gluster.org/24267
> Change-Id: Idc91a4693f1ff39f7059acde26682ccc361b947d
> Fixes: #1140
> Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
BUG: 1815434
Change-Id: Idc91a4693f1ff39f7059acde26682ccc361b947d
Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202487
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/performance/md-cache/src/md-cache.c | 165 ++++++++++++++++------------
1 file changed, 93 insertions(+), 72 deletions(-)
diff --git a/xlators/performance/md-cache/src/md-cache.c b/xlators/performance/md-cache/src/md-cache.c
index a6b363f..bbbee3b 100644
--- a/xlators/performance/md-cache/src/md-cache.c
+++ b/xlators/performance/md-cache/src/md-cache.c
@@ -133,6 +133,7 @@ struct mdc_local {
char *key;
dict_t *xattr;
uint64_t incident_time;
+ bool update_cache;
};
int
@@ -969,7 +970,7 @@ out:
return ret;
}
-void
+static bool
mdc_load_reqs(xlator_t *this, dict_t *dict)
{
struct mdc_conf *conf = this->private;
@@ -978,6 +979,7 @@ mdc_load_reqs(xlator_t *this, dict_t *dict)
char *tmp = NULL;
char *tmp1 = NULL;
int ret = 0;
+ bool loaded = false;
tmp1 = conf->mdc_xattr_str;
if (!tmp1)
@@ -995,13 +997,17 @@ mdc_load_reqs(xlator_t *this, dict_t *dict)
conf->mdc_xattr_str = NULL;
gf_msg("md-cache", GF_LOG_ERROR, 0, MD_CACHE_MSG_NO_XATTR_CACHE,
"Disabled cache for xattrs, dict_set failed");
+ goto out;
}
pattern = strtok_r(NULL, ",", &tmp);
}
- GF_FREE(mdc_xattr_str);
+ loaded = true;
+
out:
- return;
+ GF_FREE(mdc_xattr_str);
+
+ return loaded;
}
struct checkpair {
@@ -1092,6 +1098,25 @@ err:
return ret;
}
+static dict_t *
+mdc_prepare_request(xlator_t *this, mdc_local_t *local, dict_t *xdata)
+{
+ if (xdata == NULL) {
+ xdata = dict_new();
+ if (xdata == NULL) {
+ local->update_cache = false;
+
+ return NULL;
+ }
+ } else {
+ dict_ref(xdata);
+ }
+
+ local->update_cache = mdc_load_reqs(this, xdata);
+
+ return xdata;
+}
+
int
mdc_statfs_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
int32_t op_ret, int32_t op_errno, struct statvfs *buf,
@@ -1201,7 +1226,9 @@ mdc_lookup_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
if (local->loc.inode) {
mdc_inode_iatt_set(this, local->loc.inode, stbuf, local->incident_time);
- mdc_inode_xatt_set(this, local->loc.inode, dict);
+ if (local->update_cache) {
+ mdc_inode_xatt_set(this, local->loc.inode, dict);
+ }
}
out:
MDC_STACK_UNWIND(lookup, frame, op_ret, op_errno, inode, stbuf, dict,
@@ -1220,7 +1247,6 @@ mdc_lookup(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xdata)
0,
};
dict_t *xattr_rsp = NULL;
- dict_t *xattr_alloc = NULL;
mdc_local_t *local = NULL;
struct mdc_conf *conf = this->private;
@@ -1271,18 +1297,18 @@ mdc_lookup(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xdata)
return 0;
uncached:
- if (!xdata)
- xdata = xattr_alloc = dict_new();
- if (xdata)
- mdc_load_reqs(this, xdata);
+ xdata = mdc_prepare_request(this, local, xdata);
STACK_WIND(frame, mdc_lookup_cbk, FIRST_CHILD(this),
FIRST_CHILD(this)->fops->lookup, loc, xdata);
if (xattr_rsp)
dict_unref(xattr_rsp);
- if (xattr_alloc)
- dict_unref(xattr_alloc);
+
+ if (xdata != NULL) {
+ dict_unref(xdata);
+ }
+
return 0;
}
@@ -1305,7 +1331,9 @@ mdc_stat_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
}
mdc_inode_iatt_set(this, local->loc.inode, buf, local->incident_time);
- mdc_inode_xatt_set(this, local->loc.inode, xdata);
+ if (local->update_cache) {
+ mdc_inode_xatt_set(this, local->loc.inode, xdata);
+ }
out:
MDC_STACK_UNWIND(stat, frame, op_ret, op_errno, buf, xdata);
@@ -1319,7 +1347,6 @@ mdc_stat(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xdata)
int ret;
struct iatt stbuf;
mdc_local_t *local = NULL;
- dict_t *xattr_alloc = NULL;
struct mdc_conf *conf = this->private;
local = mdc_local_get(frame, loc->inode);
@@ -1343,17 +1370,16 @@ mdc_stat(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xdata)
return 0;
uncached:
- if (!xdata)
- xdata = xattr_alloc = dict_new();
- if (xdata)
- mdc_load_reqs(this, xdata);
+ xdata = mdc_prepare_request(this, local, xdata);
GF_ATOMIC_INC(conf->mdc_counter.stat_miss);
STACK_WIND(frame, mdc_stat_cbk, FIRST_CHILD(this),
FIRST_CHILD(this)->fops->stat, loc, xdata);
- if (xattr_alloc)
- dict_unref(xattr_alloc);
+ if (xdata != NULL) {
+ dict_unref(xdata);
+ }
+
return 0;
}
@@ -1376,7 +1402,9 @@ mdc_fstat_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
}
mdc_inode_iatt_set(this, local->fd->inode, buf, local->incident_time);
- mdc_inode_xatt_set(this, local->fd->inode, xdata);
+ if (local->update_cache) {
+ mdc_inode_xatt_set(this, local->fd->inode, xdata);
+ }
out:
MDC_STACK_UNWIND(fstat, frame, op_ret, op_errno, buf, xdata);
@@ -1390,7 +1418,6 @@ mdc_fstat(call_frame_t *frame, xlator_t *this, fd_t *fd, dict_t *xdata)
int ret;
struct iatt stbuf;
mdc_local_t *local = NULL;
- dict_t *xattr_alloc = NULL;
struct mdc_conf *conf = this->private;
local = mdc_local_get(frame, fd->inode);
@@ -1409,17 +1436,16 @@ mdc_fstat(call_frame_t *frame, xlator_t *this, fd_t *fd, dict_t *xdata)
return 0;
uncached:
- if (!xdata)
- xdata = xattr_alloc = dict_new();
- if (xdata)
- mdc_load_reqs(this, xdata);
+ xdata = mdc_prepare_request(this, local, xdata);
GF_ATOMIC_INC(conf->mdc_counter.stat_miss);
STACK_WIND(frame, mdc_fstat_cbk, FIRST_CHILD(this),
FIRST_CHILD(this)->fops->fstat, fd, xdata);
- if (xattr_alloc)
- dict_unref(xattr_alloc);
+ if (xdata != NULL) {
+ dict_unref(xdata);
+ }
+
return 0;
}
@@ -2393,7 +2419,9 @@ mdc_getxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
goto out;
}
- mdc_inode_xatt_set(this, local->loc.inode, xdata);
+ if (local->update_cache) {
+ mdc_inode_xatt_set(this, local->loc.inode, xdata);
+ }
out:
MDC_STACK_UNWIND(getxattr, frame, op_ret, op_errno, xattr, xdata);
@@ -2410,7 +2438,6 @@ mdc_getxattr(call_frame_t *frame, xlator_t *this, loc_t *loc, const char *key,
mdc_local_t *local = NULL;
dict_t *xattr = NULL;
struct mdc_conf *conf = this->private;
- dict_t *xattr_alloc = NULL;
gf_boolean_t key_satisfied = _gf_true;
local = mdc_local_get(frame, loc->inode);
@@ -2443,18 +2470,17 @@ mdc_getxattr(call_frame_t *frame, xlator_t *this, loc_t *loc, const char *key,
uncached:
if (key_satisfied) {
- if (!xdata)
- xdata = xattr_alloc = dict_new();
- if (xdata)
- mdc_load_reqs(this, xdata);
+ xdata = mdc_prepare_request(this, local, xdata);
}
GF_ATOMIC_INC(conf->mdc_counter.xattr_miss);
STACK_WIND(frame, mdc_getxattr_cbk, FIRST_CHILD(this),
FIRST_CHILD(this)->fops->getxattr, loc, key, xdata);
- if (xattr_alloc)
- dict_unref(xattr_alloc);
+ if (key_satisfied && (xdata != NULL)) {
+ dict_unref(xdata);
+ }
+
return 0;
}
@@ -2481,7 +2507,9 @@ mdc_fgetxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
goto out;
}
- mdc_inode_xatt_set(this, local->fd->inode, xdata);
+ if (local->update_cache) {
+ mdc_inode_xatt_set(this, local->fd->inode, xdata);
+ }
out:
MDC_STACK_UNWIND(fgetxattr, frame, op_ret, op_errno, xattr, xdata);
@@ -2498,7 +2526,6 @@ mdc_fgetxattr(call_frame_t *frame, xlator_t *this, fd_t *fd, const char *key,
dict_t *xattr = NULL;
int op_errno = ENODATA;
struct mdc_conf *conf = this->private;
- dict_t *xattr_alloc = NULL;
gf_boolean_t key_satisfied = _gf_true;
local = mdc_local_get(frame, fd->inode);
@@ -2531,18 +2558,17 @@ mdc_fgetxattr(call_frame_t *frame, xlator_t *this, fd_t *fd, const char *key,
uncached:
if (key_satisfied) {
- if (!xdata)
- xdata = xattr_alloc = dict_new();
- if (xdata)
- mdc_load_reqs(this, xdata);
+ xdata = mdc_prepare_request(this, local, xdata);
}
GF_ATOMIC_INC(conf->mdc_counter.xattr_miss);
STACK_WIND(frame, mdc_fgetxattr_cbk, FIRST_CHILD(this),
FIRST_CHILD(this)->fops->fgetxattr, fd, key, xdata);
- if (xattr_alloc)
- dict_unref(xattr_alloc);
+ if (key_satisfied && (xdata != NULL)) {
+ dict_unref(xdata);
+ }
+
return 0;
}
@@ -2752,27 +2778,22 @@ int
mdc_opendir(call_frame_t *frame, xlator_t *this, loc_t *loc, fd_t *fd,
dict_t *xdata)
{
- dict_t *xattr_alloc = NULL;
mdc_local_t *local = NULL;
local = mdc_local_get(frame, loc->inode);
loc_copy(&local->loc, loc);
- if (!xdata)
- xdata = xattr_alloc = dict_new();
-
- if (xdata) {
- /* Tell readdir-ahead to include these keys in xdata when it
- * internally issues readdirp() in it's opendir_cbk */
- mdc_load_reqs(this, xdata);
- }
+ /* Tell readdir-ahead to include these keys in xdata when it
+ * internally issues readdirp() in it's opendir_cbk */
+ xdata = mdc_prepare_request(this, local, xdata);
STACK_WIND(frame, mdc_opendir_cbk, FIRST_CHILD(this),
FIRST_CHILD(this)->fops->opendir, loc, fd, xdata);
- if (xattr_alloc)
- dict_unref(xattr_alloc);
+ if (xdata != NULL) {
+ dict_unref(xdata);
+ }
return 0;
}
@@ -2800,7 +2821,9 @@ mdc_readdirp_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
continue;
mdc_inode_iatt_set(this, entry->inode, &entry->d_stat,
local->incident_time);
- mdc_inode_xatt_set(this, entry->inode, entry->dict);
+ if (local->update_cache) {
+ mdc_inode_xatt_set(this, entry->inode, entry->dict);
+ }
}
unwind:
@@ -2812,7 +2835,6 @@ int
mdc_readdirp(call_frame_t *frame, xlator_t *this, fd_t *fd, size_t size,
off_t offset, dict_t *xdata)
{
- dict_t *xattr_alloc = NULL;
mdc_local_t *local = NULL;
local = mdc_local_get(frame, fd->inode);
@@ -2821,15 +2843,15 @@ mdc_readdirp(call_frame_t *frame, xlator_t *this, fd_t *fd, size_t size,
local->fd = fd_ref(fd);
- if (!xdata)
- xdata = xattr_alloc = dict_new();
- if (xdata)
- mdc_load_reqs(this, xdata);
+ xdata = mdc_prepare_request(this, local, xdata);
STACK_WIND(frame, mdc_readdirp_cbk, FIRST_CHILD(this),
FIRST_CHILD(this)->fops->readdirp, fd, size, offset, xdata);
- if (xattr_alloc)
- dict_unref(xattr_alloc);
+
+ if (xdata != NULL) {
+ dict_unref(xdata);
+ }
+
return 0;
out:
MDC_STACK_UNWIND(readdirp, frame, -1, ENOMEM, NULL, NULL);
@@ -2860,7 +2882,6 @@ int
mdc_readdir(call_frame_t *frame, xlator_t *this, fd_t *fd, size_t size,
off_t offset, dict_t *xdata)
{
- int need_unref = 0;
mdc_local_t *local = NULL;
struct mdc_conf *conf = this->private;
@@ -2876,19 +2897,14 @@ mdc_readdir(call_frame_t *frame, xlator_t *this, fd_t *fd, size_t size,
return 0;
}
- if (!xdata) {
- xdata = dict_new();
- need_unref = 1;
- }
-
- if (xdata)
- mdc_load_reqs(this, xdata);
+ xdata = mdc_prepare_request(this, local, xdata);
STACK_WIND(frame, mdc_readdirp_cbk, FIRST_CHILD(this),
FIRST_CHILD(this)->fops->readdirp, fd, size, offset, xdata);
- if (need_unref && xdata)
+ if (xdata != NULL) {
dict_unref(xdata);
+ }
return 0;
unwind:
@@ -3468,7 +3484,12 @@ mdc_register_xattr_inval(xlator_t *this)
goto out;
}
- mdc_load_reqs(this, xattr);
+ if (!mdc_load_reqs(this, xattr)) {
+ gf_msg(this->name, GF_LOG_WARNING, ENOMEM, MD_CACHE_MSG_NO_MEMORY,
+ "failed to populate cache entries");
+ ret = -1;
+ goto out;
+ }
frame = create_frame(this, this->ctx->pool);
if (!frame) {
--
1.8.3.1

104
SOURCES/0419-cluster-afr-fix-race-when-bricks-come-up.patch Normal file
View File

@ -0,0 +1,104 @@
From b9b479de2a7fd1c5eefa7aa1142e0a39e0c96ca9 Mon Sep 17 00:00:00 2001
From: Xavi Hernandez <xhernandez@redhat.com>
Date: Sun, 1 Mar 2020 19:49:04 +0100
Subject: [PATCH 419/449] cluster/afr: fix race when bricks come up
The was a problem when self-heal was sending lookups at the same time
that one of the bricks was coming up. In this case there was a chance
that the number of 'up' bricks changes in the middle of sending the
requests to subvolumes which caused a discrepancy in the expected
number of replies and the actual number of sent requests.
This discrepancy caused that AFR continued executing requests before
all requests were complete. Eventually, the frame of the pending
request was destroyed when the operation terminated, causing a use-
after-free issue when the answer was finally received.
In theory the same thing could happen in the reverse way, i.e. AFR
tries to wait for more replies than sent requests, causing a hang.
Backport of:
> Upstream-patch-link: https://review.gluster.org/24191
> Change-Id: I7ed6108554ca379d532efb1a29b2de8085410b70
> Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
> Fixes: bz#1808875
BUG: 1794663
Change-Id: I7ed6108554ca379d532efb1a29b2de8085410b70
Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202489
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/cluster/afr/src/afr-self-heal-common.c | 6 +++---
xlators/cluster/afr/src/afr-self-heal-name.c | 4 +++-
xlators/cluster/afr/src/afr-self-heal.h | 7 +++++--
3 files changed, 11 insertions(+), 6 deletions(-)
diff --git a/xlators/cluster/afr/src/afr-self-heal-common.c b/xlators/cluster/afr/src/afr-self-heal-common.c
index ce1ea50..d942ccf 100644
--- a/xlators/cluster/afr/src/afr-self-heal-common.c
+++ b/xlators/cluster/afr/src/afr-self-heal-common.c
@@ -1869,12 +1869,12 @@ int
afr_selfheal_unlocked_discover(call_frame_t *frame, inode_t *inode, uuid_t gfid,
struct afr_reply *replies)
{
- afr_private_t *priv = NULL;
+ afr_local_t *local = NULL;
- priv = frame->this->private;
+ local = frame->local;
return afr_selfheal_unlocked_discover_on(frame, inode, gfid, replies,
- priv->child_up);
+ local->child_up);
}
unsigned int
diff --git a/xlators/cluster/afr/src/afr-self-heal-name.c b/xlators/cluster/afr/src/afr-self-heal-name.c
index 7d4f208..dace071 100644
--- a/xlators/cluster/afr/src/afr-self-heal-name.c
+++ b/xlators/cluster/afr/src/afr-self-heal-name.c
@@ -560,13 +560,15 @@ afr_selfheal_name_unlocked_inspect(call_frame_t *frame, xlator_t *this,
struct afr_reply *replies = NULL;
inode_t *inode = NULL;
int first_idx = -1;
+ afr_local_t *local = NULL;
priv = this->private;
+ local = frame->local;
replies = alloca0(sizeof(*replies) * priv->child_count);
inode = afr_selfheal_unlocked_lookup_on(frame, parent, bname, replies,
- priv->child_up, NULL);
+ local->child_up, NULL);
if (!inode)
return -ENOMEM;
diff --git a/xlators/cluster/afr/src/afr-self-heal.h b/xlators/cluster/afr/src/afr-self-heal.h
index 8234cec..f7ecf5d 100644
--- a/xlators/cluster/afr/src/afr-self-heal.h
+++ b/xlators/cluster/afr/src/afr-self-heal.h
@@ -46,13 +46,16 @@
afr_local_t *__local = frame->local; \
afr_private_t *__priv = frame->this->private; \
int __i = 0; \
- int __count = AFR_COUNT(list, __priv->child_count); \
+ int __count = 0; \
+ unsigned char *__list = alloca(__priv->child_count); \
\
+ memcpy(__list, list, sizeof(*__list) * __priv->child_count); \
+ __count = AFR_COUNT(__list, __priv->child_count); \
__local->barrier.waitfor = __count; \
afr_local_replies_wipe(__local, __priv); \
\
for (__i = 0; __i < __priv->child_count; __i++) { \
- if (!list[__i]) \
+ if (!__list[__i]) \
continue; \
STACK_WIND_COOKIE(frame, rfn, (void *)(long)__i, \
__priv->children[__i], \
--
1.8.3.1

46
SOURCES/0420-scripts-quota_fsck-script-TypeError-d-format-not-dic.patch Normal file
View File

@ -0,0 +1,46 @@
From 42a05c7f8464f529f53bced31a64ea373e16f58b Mon Sep 17 00:00:00 2001
From: Hari Gowtham <hgowtham@redhat.com>
Date: Thu, 24 Oct 2019 17:40:44 +0530
Subject: [PATCH 420/449] scripts: quota_fsck script TypeError: %d format:not
dict
Problem: One of the prints in the script have been using
%i as the format for printing which doesn't work.
Fix: use %s as the format in the place of %i
>Fixes: bz#1764129
>Change-Id: I4480ede7bf62906ddedbe5f880a1e89c76946641
>Signed-off-by: Hari Gowtham <hgowtham@redhat.com>
>Upstream patch: https://review.gluster.org/#/c/glusterfs/+/23586/
BUG: 1786681
Change-Id: I4480ede7bf62906ddedbe5f880a1e89c76946641
Signed-off-by: hari gowtham <hgowtham@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202484
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunny Kumar <sunkumar@redhat.com>
---
extras/quota/quota_fsck.py | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/extras/quota/quota_fsck.py b/extras/quota/quota_fsck.py
index 485a37a..174f2a2 100755
--- a/extras/quota/quota_fsck.py
+++ b/extras/quota/quota_fsck.py
@@ -58,10 +58,10 @@ def print_msg(log_type, path, xattr_dict = {}, stbuf = "", dir_size = None):
elif log_type == QUOTA_SIZE_MISMATCH:
print("mismatch")
if dir_size is not None:
- print('%24s %60s %12s %12s' % ("Size Mismatch", path,
+ print('%24s %60s %12s %12s' % ("Size Mismatch", path,
xattr_dict, dir_size))
else:
- print('%-24s %-60s %-12i %-12i' % ("Size Mismatch", path, xattr_dict,
+ print('%-24s %-60s %-12s %-12s' % ("Size Mismatch", path, xattr_dict,
stbuf.st_size))
def size_differs_lot(s1, s2):
--
1.8.3.1

93
SOURCES/0421-Improve-logging-in-EC-client-and-lock-translator.patch Normal file
View File

@ -0,0 +1,93 @@
From 8267e5e97327633bf21fd02df8d52e3a97f0f9ea Mon Sep 17 00:00:00 2001
From: Ashish Pandey <aspandey@redhat.com>
Date: Wed, 4 Dec 2019 17:06:18 +0530
Subject: [PATCH 421/449] Improve logging in EC, client and lock translator
BUG: 1787294
> Upstream patch: https://review.gluster.org/#/c/glusterfs/+/23814/
> Change-Id: I98af8672a25ff9fd9dba91a2e1384719f9155255
> Fixes: bz#1779760
Change-Id: I5cb04993f12d6248f2349a0c5a9e2c0ceecaf528
Reviewed-on: https://code.engineering.redhat.com/gerrit/202533
Tested-by: Ashish Pandey <aspandey@redhat.com>
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/cluster/ec/src/ec-combine.c | 5 +++--
xlators/cluster/ec/src/ec-common.c | 2 +-
xlators/features/locks/src/inodelk.c | 12 ++++++++----
3 files changed, 12 insertions(+), 7 deletions(-)
diff --git a/xlators/cluster/ec/src/ec-combine.c b/xlators/cluster/ec/src/ec-combine.c
index c5af2ab..99e5534 100644
--- a/xlators/cluster/ec/src/ec-combine.c
+++ b/xlators/cluster/ec/src/ec-combine.c
@@ -179,13 +179,14 @@ ec_iatt_combine(ec_fop_data_t *fop, struct iatt *dst, struct iatt *src,
"links: %u-%u, uid: %u-%u, gid: %u-%u, "
"rdev: %" PRIu64 "-%" PRIu64 ", size: %" PRIu64 "-%" PRIu64
", "
- "mode: %o-%o)",
+ "mode: %o-%o), %s",
dst[i].ia_ino, src[i].ia_ino, dst[i].ia_nlink,
src[i].ia_nlink, dst[i].ia_uid, src[i].ia_uid, dst[i].ia_gid,
src[i].ia_gid, dst[i].ia_rdev, src[i].ia_rdev,
dst[i].ia_size, src[i].ia_size,
st_mode_from_ia(dst[i].ia_prot, dst[i].ia_type),
- st_mode_from_ia(src[i].ia_prot, dst[i].ia_type));
+ st_mode_from_ia(src[i].ia_prot, dst[i].ia_type),
+ ec_msg_str(fop));
return 0;
}
diff --git a/xlators/cluster/ec/src/ec-common.c b/xlators/cluster/ec/src/ec-common.c
index 5cae37b..e580bfb 100644
--- a/xlators/cluster/ec/src/ec-common.c
+++ b/xlators/cluster/ec/src/ec-common.c
@@ -2240,7 +2240,7 @@ ec_unlocked(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
if (op_ret < 0) {
gf_msg(this->name, GF_LOG_WARNING, op_errno, EC_MSG_UNLOCK_FAILED,
- "entry/inode unlocking failed (%s)", ec_fop_name(link->fop->id));
+ "entry/inode unlocking failed :(%s)", ec_msg_str(link->fop));
} else {
ec_trace("UNLOCKED", link->fop, "lock=%p", link->lock);
}
diff --git a/xlators/features/locks/src/inodelk.c b/xlators/features/locks/src/inodelk.c
index df00ede..a9c42f1 100644
--- a/xlators/features/locks/src/inodelk.c
+++ b/xlators/features/locks/src/inodelk.c
@@ -502,22 +502,26 @@ static pl_inode_lock_t *
__inode_unlock_lock(xlator_t *this, pl_inode_lock_t *lock, pl_dom_list_t *dom)
{
pl_inode_lock_t *conf = NULL;
+ inode_t *inode = NULL;
+
+ inode = lock->pl_inode->inode;
conf = find_matching_inodelk(lock, dom);
if (!conf) {
gf_log(this->name, GF_LOG_ERROR,
" Matching lock not found for unlock %llu-%llu, by %s "
- "on %p",
+ "on %p for gfid:%s",
(unsigned long long)lock->fl_start,
(unsigned long long)lock->fl_end, lkowner_utoa(&lock->owner),
- lock->client);
+ lock->client, inode ? uuid_utoa(inode->gfid) : "UNKNOWN");
goto out;
}
__delete_inode_lock(conf);
gf_log(this->name, GF_LOG_DEBUG,
- " Matching lock found for unlock %llu-%llu, by %s on %p",
+ " Matching lock found for unlock %llu-%llu, by %s on %p for gfid:%s",
(unsigned long long)lock->fl_start, (unsigned long long)lock->fl_end,
- lkowner_utoa(&lock->owner), lock->client);
+ lkowner_utoa(&lock->owner), lock->client,
+ inode ? uuid_utoa(inode->gfid) : "UNKNOWN");
out:
return conf;
--
1.8.3.1

17
SOURCES/0385-cluster-afr-Prioritize-ENOSPC-over-other-errors.patch → SOURCES/0422-cluster-afr-Prioritize-ENOSPC-over-other-errors.patch
View File

@ -1,7 +1,7 @@
From cdd067dcc0cd70d4f57e173b4050d8e2eb79725a Mon Sep 17 00:00:00 2001
From 8b11ac1575ef167af2a47a96f7b7ed0f32bb5897 Mon Sep 17 00:00:00 2001
From: karthik-us <ksubrahm@redhat.com>
Date: Fri, 5 Jun 2020 17:20:04 +0530
Subject: [PATCH 385/392] cluster/afr: Prioritize ENOSPC over other errors
Subject: [PATCH 422/449] cluster/afr: Prioritize ENOSPC over other errors
Backport of: https://review.gluster.org/#/c/glusterfs/+/24477/
@ -23,9 +23,8 @@ on quorum number of bricks in arbiter configuration.
Change-Id: I4dd2bff17e6812bc7c8372130976e365e2407d88
Signed-off-by: karthik-us <ksubrahm@redhat.com>
BUG: 1848895
(cherry picked from commit 8b11ac1575ef167af2a47a96f7b7ed0f32bb5897)
Reviewed-on: https://code.engineering.redhat.com/gerrit/203691
BUG: 1837467
Reviewed-on: https://code.engineering.redhat.com/gerrit/202526
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
@ -123,10 +122,10 @@ index 0000000..fab94b7
+
+cleanup
diff --git a/xlators/cluster/afr/src/afr-common.c b/xlators/cluster/afr/src/afr-common.c
index 3690b84..d6b70e9 100644
index 5806556..59710aa 100644
--- a/xlators/cluster/afr/src/afr-common.c
+++ b/xlators/cluster/afr/src/afr-common.c
@@ -2463,7 +2463,7 @@ error:
@@ -2464,7 +2464,7 @@ error:
* others in that they must be given higher priority while
* returning to the user.
*
@ -135,7 +134,7 @@ index 3690b84..d6b70e9 100644
*/
int
@@ -2475,6 +2475,8 @@ afr_higher_errno(int32_t old_errno, int32_t new_errno)
@@ -2476,6 +2476,8 @@ afr_higher_errno(int32_t old_errno, int32_t new_errno)
return ENOENT;
if (old_errno == ESTALE || new_errno == ESTALE)
return ESTALE;
@ -220,7 +219,7 @@ index 15f3a7e..8e65ae2 100644
if (afr_need_dirty_marking(frame, this))
goto set_response;
diff --git a/xlators/storage/posix/src/posix-helpers.c b/xlators/storage/posix/src/posix-helpers.c
index 35dd3b6..aca0df6 100644
index 2c27d22..949c799 100644
--- a/xlators/storage/posix/src/posix-helpers.c
+++ b/xlators/storage/posix/src/posix-helpers.c
@@ -1059,7 +1059,7 @@ verify_handle:

128
SOURCES/0423-ctime-Fix-ctime-inconsisteny-with-utimensat.patch Normal file
View File

@ -0,0 +1,128 @@
From c140d30382306d08eaf2bc5c53e5be26d3e381e1 Mon Sep 17 00:00:00 2001
From: Kotresh HR <khiremat@redhat.com>
Date: Mon, 18 Nov 2019 05:24:33 -0500
Subject: [PATCH 423/449] ctime: Fix ctime inconsisteny with utimensat
Problem:
When touch is used to create a file, the ctime is not matching
atime and mtime which ideally should match. There is a difference
in nano seconds.
Cause:
When touch is used modify atime or mtime to current time (UTIME_NOW),
the current time is taken from kernel. The ctime gets updated to current
time when atime or mtime is updated. But the current time to update
ctime is taken from utime xlator. Hence the difference in nano seconds.
Fix:
When utimesat uses UTIME_NOW, use the current time from kernel.
>fixes: bz#1773530
>Change-Id: I9ccfa47dcd39df23396852b4216f1773c49250ce
>Signed-off-by: Kotresh HR <khiremat@redhat.com>
backport of: https://review.gluster.org/#/c/glusterfs/+/23719/
BUG: 1761932
Change-Id: I9ccfa47dcd39df23396852b4216f1773c49250ce
Signed-off-by: Kotresh HR <khiremat@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202541
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
libglusterfs/src/glusterfs/xlator.h | 2 ++
tests/basic/ctime/ctime-utimesat.t | 28 ++++++++++++++++++++++++++
xlators/features/utime/src/utime-gen-fops-c.py | 10 +++++++++
xlators/mount/fuse/src/fuse-bridge.c | 8 ++++++++
4 files changed, 48 insertions(+)
create mode 100644 tests/basic/ctime/ctime-utimesat.t
diff --git a/libglusterfs/src/glusterfs/xlator.h b/libglusterfs/src/glusterfs/xlator.h
index da551e9..db04c4d 100644
--- a/libglusterfs/src/glusterfs/xlator.h
+++ b/libglusterfs/src/glusterfs/xlator.h
@@ -35,6 +35,8 @@
#define GF_SET_ATTR_ATIME 0x10
#define GF_SET_ATTR_MTIME 0x20
#define GF_SET_ATTR_CTIME 0x40
+#define GF_ATTR_ATIME_NOW 0x80
+#define GF_ATTR_MTIME_NOW 0x100
#define gf_attr_mode_set(mode) ((mode)&GF_SET_ATTR_MODE)
#define gf_attr_uid_set(mode) ((mode)&GF_SET_ATTR_UID)
diff --git a/tests/basic/ctime/ctime-utimesat.t b/tests/basic/ctime/ctime-utimesat.t
new file mode 100644
index 0000000..540e57a
--- /dev/null
+++ b/tests/basic/ctime/ctime-utimesat.t
@@ -0,0 +1,28 @@
+#!/bin/bash
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../afr.rc
+cleanup;
+
+TEST glusterd
+TEST pidof glusterd
+TEST $CLI volume create $V0 replica 3 $H0:$B0/${V0}{0,1,2}
+TEST $CLI volume set $V0 performance.stat-prefetch off
+TEST $CLI volume set $V0 performance.read-ahead off
+TEST $CLI volume set $V0 performance.quick-read off
+TEST $CLI volume set $V0 performance.read-after-open off
+TEST $CLI volume set $V0 performance.open-behind off
+TEST $CLI volume set $V0 performance.write-behind off
+TEST $CLI volume set $V0 performance.io-cache off
+
+TEST $CLI volume start $V0
+
+TEST glusterfs --volfile-id=$V0 --volfile-server=$H0 --entry-timeout=0 $M0;
+
+touch $M0/FILE
+
+atime=$(stat -c "%.X" $M0/FILE)
+EXPECT $atime stat -c "%.Y" $M0/FILE
+EXPECT $atime stat -c "%.Z" $M0/FILE
+
+cleanup
diff --git a/xlators/features/utime/src/utime-gen-fops-c.py b/xlators/features/utime/src/utime-gen-fops-c.py
index 8730a51..9fb3e1b 100755
--- a/xlators/features/utime/src/utime-gen-fops-c.py
+++ b/xlators/features/utime/src/utime-gen-fops-c.py
@@ -95,6 +95,16 @@ gf_utime_@NAME@ (call_frame_t *frame, xlator_t *this,
frame->root->flags |= MDATA_CTIME;
}
+ if (valid & (GF_SET_ATTR_ATIME | GF_SET_ATTR_MTIME)) {
+ if (valid & GF_ATTR_ATIME_NOW) {
+ frame->root->ctime.tv_sec = stbuf->ia_atime;
+ frame->root->ctime.tv_nsec = stbuf->ia_atime_nsec;
+ } else if (valid & GF_ATTR_MTIME_NOW) {
+ frame->root->ctime.tv_sec = stbuf->ia_mtime;
+ frame->root->ctime.tv_nsec = stbuf->ia_mtime_nsec;
+ }
+ }
+
STACK_WIND (frame, gf_utime_@NAME@_cbk, FIRST_CHILD(this),
FIRST_CHILD(this)->fops->@NAME@, @SHORT_ARGS@);
return 0;
diff --git a/xlators/mount/fuse/src/fuse-bridge.c b/xlators/mount/fuse/src/fuse-bridge.c
index 6e99053..fdeec49 100644
--- a/xlators/mount/fuse/src/fuse-bridge.c
+++ b/xlators/mount/fuse/src/fuse-bridge.c
@@ -1706,6 +1706,14 @@ fattr_to_gf_set_attr(int32_t valid)
gf_valid |= GF_SET_ATTR_CTIME;
#endif
+#if FUSE_KERNEL_MINOR_VERSION >= 9
+ if (valid & FATTR_ATIME_NOW)
+ gf_valid |= GF_ATTR_ATIME_NOW;
+
+ if (valid & FATTR_MTIME_NOW)
+ gf_valid |= GF_ATTR_MTIME_NOW;
+#endif
+
if (valid & FATTR_SIZE)
gf_valid |= GF_SET_ATTR_SIZE;
--
1.8.3.1

884
SOURCES/0424-afr-make-heal-info-lockless.patch Normal file
View File

@ -0,0 +1,884 @@
From 54d4ea44fec96560aad9c41f7e4f5aad164ffb8b Mon Sep 17 00:00:00 2001
From: Ravishankar N <ravishankar@redhat.com>
Date: Fri, 5 Jun 2020 14:14:15 +0530
Subject: [PATCH 424/449] afr: make heal info lockless
Changes in locks xlator:
Added support for per-domain inodelk count requests.
Caller needs to set GLUSTERFS_MULTIPLE_DOM_LK_CNT_REQUESTS key in the
dict and then set each key with name
'GLUSTERFS_INODELK_DOM_PREFIX:<domain name>'.
In the response dict, the xlator will send the per domain count as
values for each of these keys.
Changes in AFR:
Replaced afr_selfheal_locked_inspect() with afr_lockless_inspect(). Logic has
been added to make the latter behave same as the former, thus not
breaking the current heal info output behaviour.
> Upstream patch: https://review.gluster.org/#/c/glusterfs/+/23771/
> fixes: bz#1774011
> Change-Id: Ie9e83c162aa77f44a39c2ba7115de558120ada4d
BUG: 1721355
Change-Id: I8ed4b504880b19e00068312efd90cd0706787404
Signed-off-by: Ravishankar N <ravishankar@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202490
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Karthik Subrahmanya <ksubrahm@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
heal/src/glfs-heal.c | 17 +-
libglusterfs/src/glusterfs/glusterfs.h | 2 +
xlators/cluster/afr/src/afr-common.c | 367 +++++++++++--------------
xlators/cluster/afr/src/afr-self-heal-common.c | 43 ++-
xlators/cluster/afr/src/afr-self-heal.h | 3 +-
xlators/features/locks/src/common.h | 4 +
xlators/features/locks/src/locks.h | 8 +
xlators/features/locks/src/posix.c | 117 +++++++-
8 files changed, 338 insertions(+), 223 deletions(-)
diff --git a/heal/src/glfs-heal.c b/heal/src/glfs-heal.c
index 125b12c..5af9e31 100644
--- a/heal/src/glfs-heal.c
+++ b/heal/src/glfs-heal.c
@@ -775,7 +775,8 @@ static int
glfsh_process_entries(xlator_t *xl, fd_t *fd, gf_dirent_t *entries,
uint64_t *offset, num_entries_t *num_entries,
print_status glfsh_print_status,
- gf_boolean_t ignore_dirty, glfsh_fail_mode_t mode)
+ gf_boolean_t ignore_dirty, glfsh_fail_mode_t mode,
+ dict_t *xattr_req)
{
gf_dirent_t *entry = NULL;
gf_dirent_t *tmp = NULL;
@@ -807,7 +808,7 @@ glfsh_process_entries(xlator_t *xl, fd_t *fd, gf_dirent_t *entries,
gf_uuid_parse(entry->d_name, gfid);
gf_uuid_copy(loc.gfid, gfid);
- ret = syncop_getxattr(this, &loc, &dict, GF_HEAL_INFO, NULL, NULL);
+ ret = syncop_getxattr(this, &loc, &dict, GF_HEAL_INFO, xattr_req, NULL);
if (ret) {
if ((mode != GLFSH_MODE_CONTINUE_ON_ERROR) && (ret == -ENOTCONN))
goto out;
@@ -876,19 +877,19 @@ glfsh_crawl_directory(glfs_t *fs, xlator_t *top_subvol, loc_t *rootloc,
if (heal_op == GF_SHD_OP_INDEX_SUMMARY) {
ret = glfsh_process_entries(readdir_xl, fd, &entries, &offset,
num_entries, glfsh_print_heal_status,
- ignore, mode);
+ ignore, mode, xattr_req);
if (ret < 0)
goto out;
} else if (heal_op == GF_SHD_OP_SPLIT_BRAIN_FILES) {
ret = glfsh_process_entries(readdir_xl, fd, &entries, &offset,
num_entries, glfsh_print_spb_status,
- ignore, mode);
+ ignore, mode, xattr_req);
if (ret < 0)
goto out;
} else if (heal_op == GF_SHD_OP_HEAL_SUMMARY) {
ret = glfsh_process_entries(readdir_xl, fd, &entries, &offset,
num_entries, glfsh_print_summary_status,
- ignore, mode);
+ ignore, mode, xattr_req);
if (ret < 0)
goto out;
} else if (heal_op == GF_SHD_OP_SBRAIN_HEAL_FROM_BRICK) {
@@ -897,7 +898,7 @@ glfsh_crawl_directory(glfs_t *fs, xlator_t *top_subvol, loc_t *rootloc,
} else if (heal_op == GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE) {
ret = glfsh_process_entries(readdir_xl, fd, &entries, &offset,
num_entries, glfsh_heal_status_boolean,
- ignore, mode);
+ ignore, mode, xattr_req);
if (ret < 0)
goto out;
}
@@ -951,6 +952,10 @@ glfsh_print_pending_heals_type(glfs_t *fs, xlator_t *top_subvol, loc_t *rootloc,
int32_t op_errno = 0;
gf_boolean_t ignore = _gf_false;
+ ret = dict_set_str(xattr_req, "index-vgfid", vgfid);
+ if (ret)
+ return ret;
+
if (!strcmp(vgfid, GF_XATTROP_DIRTY_GFID))
ignore = _gf_true;
diff --git a/libglusterfs/src/glusterfs/glusterfs.h b/libglusterfs/src/glusterfs/glusterfs.h
index 3b594c0..177a020 100644
--- a/libglusterfs/src/glusterfs/glusterfs.h
+++ b/libglusterfs/src/glusterfs/glusterfs.h
@@ -217,6 +217,8 @@ enum gf_internal_fop_indicator {
#define GLUSTERFS_POSIXLK_COUNT "glusterfs.posixlk-count"
#define GLUSTERFS_PARENT_ENTRYLK "glusterfs.parent-entrylk"
#define GLUSTERFS_INODELK_DOM_COUNT "glusterfs.inodelk-dom-count"
+#define GLUSTERFS_INODELK_DOM_PREFIX "glusterfs.inodelk-dom-prefix"
+#define GLUSTERFS_MULTIPLE_DOM_LK_CNT_REQUESTS "glusterfs.multi-dom-lk-cnt-req"
#define GFID_TO_PATH_KEY "glusterfs.gfid2path"
#define GF_XATTR_STIME_PATTERN "trusted.glusterfs.*.stime"
#define GF_XATTR_XTIME_PATTERN "trusted.glusterfs.*.xtime"
diff --git a/xlators/cluster/afr/src/afr-common.c b/xlators/cluster/afr/src/afr-common.c
index 59710aa..c355ec5 100644
--- a/xlators/cluster/afr/src/afr-common.c
+++ b/xlators/cluster/afr/src/afr-common.c
@@ -5908,259 +5908,218 @@ out:
return _gf_true;
}
-int
-afr_selfheal_locked_metadata_inspect(call_frame_t *frame, xlator_t *this,
- inode_t *inode, gf_boolean_t *msh,
- unsigned char *pending)
+static dict_t *
+afr_set_heal_info(char *status)
{
+ dict_t *dict = NULL;
int ret = -1;
- unsigned char *locked_on = NULL;
- unsigned char *sources = NULL;
- unsigned char *sinks = NULL;
- unsigned char *healed_sinks = NULL;
- unsigned char *undid_pending = NULL;
- struct afr_reply *locked_replies = NULL;
-
- afr_private_t *priv = this->private;
- locked_on = alloca0(priv->child_count);
- sources = alloca0(priv->child_count);
- sinks = alloca0(priv->child_count);
- healed_sinks = alloca0(priv->child_count);
- undid_pending = alloca0(priv->child_count);
+ dict = dict_new();
+ if (!dict) {
+ ret = -ENOMEM;
+ goto out;
+ }
- locked_replies = alloca0(sizeof(*locked_replies) * priv->child_count);
+ ret = dict_set_dynstr_sizen(dict, "heal-info", status);
+ if (ret)
+ gf_msg("", GF_LOG_WARNING, -ret, AFR_MSG_DICT_SET_FAILED,
+ "Failed to set heal-info key to "
+ "%s",
+ status);
+out:
+ /* Any error other than EINVAL, dict_set_dynstr frees status */
+ if (ret == -ENOMEM || ret == -EINVAL) {
+ GF_FREE(status);
+ }
- ret = afr_selfheal_inodelk(frame, this, inode, this->name, LLONG_MAX - 1, 0,
- locked_on);
- {
- if (ret == 0) {
- /* Not a single lock */
- ret = -afr_final_errno(frame->local, priv);
- if (ret == 0)
- ret = -ENOTCONN; /* all invalid responses */
- goto out;
- }
- ret = __afr_selfheal_metadata_prepare(
- frame, this, inode, locked_on, sources, sinks, healed_sinks,
- undid_pending, locked_replies, pending);
- *msh = afr_decide_heal_info(priv, sources, ret);
+ if (ret && dict) {
+ dict_unref(dict);
+ dict = NULL;
}
- afr_selfheal_uninodelk(frame, this, inode, this->name, LLONG_MAX - 1, 0,
- locked_on);
-out:
- if (locked_replies)
- afr_replies_wipe(locked_replies, priv->child_count);
- return ret;
+ return dict;
}
-int
-afr_selfheal_locked_data_inspect(call_frame_t *frame, xlator_t *this, fd_t *fd,
- gf_boolean_t *dsh, unsigned char *pflag)
+static gf_boolean_t
+afr_is_dirty_count_non_unary_for_txn(xlator_t *this, struct afr_reply *replies,
+ afr_transaction_type type)
{
- int ret = -1;
- unsigned char *data_lock = NULL;
- unsigned char *sources = NULL;
- unsigned char *sinks = NULL;
- unsigned char *healed_sinks = NULL;
- unsigned char *undid_pending = NULL;
- afr_private_t *priv = NULL;
- struct afr_reply *locked_replies = NULL;
- inode_t *inode = fd->inode;
+ afr_private_t *priv = this->private;
+ int *dirty = alloca0(priv->child_count * sizeof(int));
+ int i = 0;
- priv = this->private;
- data_lock = alloca0(priv->child_count);
- sources = alloca0(priv->child_count);
- sinks = alloca0(priv->child_count);
- healed_sinks = alloca0(priv->child_count);
- undid_pending = alloca0(priv->child_count);
+ afr_selfheal_extract_xattr(this, replies, type, dirty, NULL);
+ for (i = 0; i < priv->child_count; i++) {
+ if (dirty[i] > 1)
+ return _gf_true;
+ }
- locked_replies = alloca0(sizeof(*locked_replies) * priv->child_count);
+ return _gf_false;
+}
- ret = afr_selfheal_inodelk(frame, this, inode, this->name, 0, 0, data_lock);
- {
- if (ret == 0) {
- ret = -afr_final_errno(frame->local, priv);
- if (ret == 0)
- ret = -ENOTCONN; /* all invalid responses */
- goto out;
- }
- ret = __afr_selfheal_data_prepare(frame, this, inode, data_lock,
- sources, sinks, healed_sinks,
- undid_pending, locked_replies, pflag);
- *dsh = afr_decide_heal_info(priv, sources, ret);
+static gf_boolean_t
+afr_is_dirty_count_non_unary(xlator_t *this, struct afr_reply *replies,
+ ia_type_t ia_type)
+{
+ gf_boolean_t data_chk = _gf_false;
+ gf_boolean_t mdata_chk = _gf_false;
+ gf_boolean_t entry_chk = _gf_false;
+
+ switch (ia_type) {
+ case IA_IFDIR:
+ mdata_chk = _gf_true;
+ entry_chk = _gf_true;
+ break;
+ case IA_IFREG:
+ mdata_chk = _gf_true;
+ data_chk = _gf_true;
+ break;
+ default:
+ /*IA_IFBLK, IA_IFCHR, IA_IFLNK, IA_IFIFO, IA_IFSOCK*/
+ mdata_chk = _gf_true;
+ break;
}
- afr_selfheal_uninodelk(frame, this, inode, this->name, 0, 0, data_lock);
-out:
- if (locked_replies)
- afr_replies_wipe(locked_replies, priv->child_count);
- return ret;
+
+ if (data_chk && afr_is_dirty_count_non_unary_for_txn(
+ this, replies, AFR_DATA_TRANSACTION)) {
+ return _gf_true;
+ } else if (mdata_chk && afr_is_dirty_count_non_unary_for_txn(
+ this, replies, AFR_METADATA_TRANSACTION)) {
+ return _gf_true;
+ } else if (entry_chk && afr_is_dirty_count_non_unary_for_txn(
+ this, replies, AFR_ENTRY_TRANSACTION)) {
+ return _gf_true;
+ }
+
+ return _gf_false;
}
-int
-afr_selfheal_locked_entry_inspect(call_frame_t *frame, xlator_t *this,
- inode_t *inode, gf_boolean_t *esh,
- unsigned char *pflag)
+static int
+afr_update_heal_status(xlator_t *this, struct afr_reply *replies,
+ char *index_vgfid, ia_type_t ia_type, gf_boolean_t *esh,
+ gf_boolean_t *dsh, gf_boolean_t *msh)
{
int ret = -1;
- int source = -1;
+ GF_UNUSED int ret1 = 0;
+ int i = 0;
+ int io_domain_lk_count = 0;
+ int shd_domain_lk_count = 0;
afr_private_t *priv = NULL;
- unsigned char *locked_on = NULL;
- unsigned char *data_lock = NULL;
- unsigned char *sources = NULL;
- unsigned char *sinks = NULL;
- unsigned char *healed_sinks = NULL;
- struct afr_reply *locked_replies = NULL;
- gf_boolean_t granular_locks = _gf_false;
+ char *key1 = NULL;
+ char *key2 = NULL;
priv = this->private;
- granular_locks = priv->granular_locks; /*Assign to local variable so that
- reconfigure doesn't change this
- value between locking and unlocking
- below*/
- locked_on = alloca0(priv->child_count);
- data_lock = alloca0(priv->child_count);
- sources = alloca0(priv->child_count);
- sinks = alloca0(priv->child_count);
- healed_sinks = alloca0(priv->child_count);
-
- locked_replies = alloca0(sizeof(*locked_replies) * priv->child_count);
+ key1 = alloca0(strlen(GLUSTERFS_INODELK_DOM_PREFIX) + 2 +
+ strlen(this->name));
+ key2 = alloca0(strlen(GLUSTERFS_INODELK_DOM_PREFIX) + 2 +
+ strlen(priv->sh_domain));
+ sprintf(key1, "%s:%s", GLUSTERFS_INODELK_DOM_PREFIX, this->name);
+ sprintf(key2, "%s:%s", GLUSTERFS_INODELK_DOM_PREFIX, priv->sh_domain);
- if (!granular_locks) {
- ret = afr_selfheal_tryentrylk(frame, this, inode, priv->sh_domain, NULL,
- locked_on);
- }
- {
- if (!granular_locks && ret == 0) {
- ret = -afr_final_errno(frame->local, priv);
- if (ret == 0)
- ret = -ENOTCONN; /* all invalid responses */
- goto out;
+ for (i = 0; i < priv->child_count; i++) {
+ if ((replies[i].valid != 1) || (replies[i].op_ret != 0))
+ continue;
+ if (!io_domain_lk_count) {
+ ret1 = dict_get_int32(replies[i].xdata, key1, &io_domain_lk_count);
}
+ if (!shd_domain_lk_count) {
+ ret1 = dict_get_int32(replies[i].xdata, key2, &shd_domain_lk_count);
+ }
+ }
- ret = afr_selfheal_entrylk(frame, this, inode, this->name, NULL,
- data_lock);
- {
- if (ret == 0) {
- ret = -afr_final_errno(frame->local, priv);
- if (ret == 0)
- ret = -ENOTCONN;
- /* all invalid responses */
- goto unlock;
- }
- ret = __afr_selfheal_entry_prepare(frame, this, inode, data_lock,
- sources, sinks, healed_sinks,
- locked_replies, &source, pflag);
- if ((ret == 0) && (*pflag & PFLAG_SBRAIN))
- ret = -EIO;
- *esh = afr_decide_heal_info(priv, sources, ret);
+ if (!strcmp(index_vgfid, GF_XATTROP_INDEX_GFID)) {
+ if (shd_domain_lk_count) {
+ ret = -EAGAIN; /*For 'possibly-healing'. */
+ } else {
+ ret = 0; /*needs heal. Just set a non -ve value so that it is
+ assumed as the source index.*/
+ }
+ } else if (!strcmp(index_vgfid, GF_XATTROP_DIRTY_GFID)) {
+ if ((afr_is_dirty_count_non_unary(this, replies, ia_type)) ||
+ (!io_domain_lk_count)) {
+ /* Needs heal. */
+ ret = 0;
+ } else {
+ /* No heal needed. */
+ *dsh = *esh = *msh = 0;
}
- afr_selfheal_unentrylk(frame, this, inode, this->name, NULL, data_lock,
- NULL);
}
-unlock:
- if (!granular_locks)
- afr_selfheal_unentrylk(frame, this, inode, priv->sh_domain, NULL,
- locked_on, NULL);
-out:
- if (locked_replies)
- afr_replies_wipe(locked_replies, priv->child_count);
return ret;
}
+/*return EIO, EAGAIN or pending*/
int
-afr_selfheal_locked_inspect(call_frame_t *frame, xlator_t *this, uuid_t gfid,
- inode_t **inode, gf_boolean_t *entry_selfheal,
- gf_boolean_t *data_selfheal,
- gf_boolean_t *metadata_selfheal,
- unsigned char *pending)
-
+afr_lockless_inspect(call_frame_t *frame, xlator_t *this, uuid_t gfid,
+ inode_t **inode, char *index_vgfid,
+ gf_boolean_t *entry_selfheal, gf_boolean_t *data_selfheal,
+ gf_boolean_t *metadata_selfheal, unsigned char *pending)
{
int ret = -1;
- fd_t *fd = NULL;
+ int i = 0;
+ afr_private_t *priv = NULL;
+ struct afr_reply *replies = NULL;
gf_boolean_t dsh = _gf_false;
gf_boolean_t msh = _gf_false;
gf_boolean_t esh = _gf_false;
+ unsigned char *sources = NULL;
+ unsigned char *sinks = NULL;
+ unsigned char *valid_on = NULL;
+ uint64_t *witness = NULL;
+
+ priv = this->private;
+ replies = alloca0(sizeof(*replies) * priv->child_count);
+ sources = alloca0(sizeof(*sources) * priv->child_count);
+ sinks = alloca0(sizeof(*sinks) * priv->child_count);
+ witness = alloca0(sizeof(*witness) * priv->child_count);
+ valid_on = alloca0(sizeof(*valid_on) * priv->child_count);
ret = afr_selfheal_unlocked_inspect(frame, this, gfid, inode, &dsh, &msh,
- &esh);
+ &esh, replies);
if (ret)
goto out;
-
- /* For every heal type hold locks and check if it indeed needs heal */
-
- /* Heal-info does an open() on the file being examined so that the
- * current eager-lock holding client, if present, at some point sees
- * open-fd count being > 1 and releases the eager-lock so that heal-info
- * doesn't remain blocked forever until IO completes.
- */
- if ((*inode)->ia_type == IA_IFREG) {
- ret = afr_selfheal_data_open(this, *inode, &fd);
- if (ret < 0) {
- gf_msg_debug(this->name, -ret, "%s: Failed to open",
- uuid_utoa((*inode)->gfid));
- goto out;
+ for (i = 0; i < priv->child_count; i++) {
+ if (replies[i].valid && replies[i].op_ret == 0) {
+ valid_on[i] = 1;
}
}
-
if (msh) {
- ret = afr_selfheal_locked_metadata_inspect(frame, this, *inode, &msh,
- pending);
- if (ret == -EIO)
+ ret = afr_selfheal_find_direction(frame, this, replies,
+ AFR_METADATA_TRANSACTION, valid_on,
+ sources, sinks, witness, pending);
+ if (*pending & PFLAG_SBRAIN)
+ ret = -EIO;
+ if (ret)
goto out;
}
-
if (dsh) {
- ret = afr_selfheal_locked_data_inspect(frame, this, fd, &dsh, pending);
- if (ret == -EIO || (ret == -EAGAIN))
+ ret = afr_selfheal_find_direction(frame, this, replies,
+ AFR_DATA_TRANSACTION, valid_on,
+ sources, sinks, witness, pending);
+ if (*pending & PFLAG_SBRAIN)
+ ret = -EIO;
+ if (ret)
goto out;
}
-
if (esh) {
- ret = afr_selfheal_locked_entry_inspect(frame, this, *inode, &esh,
- pending);
+ ret = afr_selfheal_find_direction(frame, this, replies,
+ AFR_ENTRY_TRANSACTION, valid_on,
+ sources, sinks, witness, pending);
+ if (*pending & PFLAG_SBRAIN)
+ ret = -EIO;
+ if (ret)
+ goto out;
}
+ ret = afr_update_heal_status(this, replies, index_vgfid, (*inode)->ia_type,
+ &esh, &dsh, &msh);
out:
*data_selfheal = dsh;
*entry_selfheal = esh;
*metadata_selfheal = msh;
- if (fd)
- fd_unref(fd);
+ if (replies)
+ afr_replies_wipe(replies, priv->child_count);
return ret;
}
-static dict_t *
-afr_set_heal_info(char *status)
-{
- dict_t *dict = NULL;
- int ret = -1;
-
- dict = dict_new();
- if (!dict) {
- ret = -ENOMEM;
- goto out;
- }
-
- ret = dict_set_dynstr_sizen(dict, "heal-info", status);
- if (ret)
- gf_msg("", GF_LOG_WARNING, -ret, AFR_MSG_DICT_SET_FAILED,
- "Failed to set heal-info key to "
- "%s",
- status);
-out:
- /* Any error other than EINVAL, dict_set_dynstr frees status */
- if (ret == -ENOMEM || ret == -EINVAL) {
- GF_FREE(status);
- }
-
- if (ret && dict) {
- dict_unref(dict);
- dict = NULL;
- }
- return dict;
-}
-
int
afr_get_heal_info(call_frame_t *frame, xlator_t *this, loc_t *loc)
{
@@ -6174,10 +6133,18 @@ afr_get_heal_info(call_frame_t *frame, xlator_t *this, loc_t *loc)
inode_t *inode = NULL;
char *substr = NULL;
char *status = NULL;
+ afr_local_t *local = NULL;
+ char *index_vgfid = NULL;
+
+ local = frame->local;
+ if (dict_get_str(local->xdata_req, "index-vgfid", &index_vgfid)) {
+ ret = -1;
+ goto out;
+ }
- ret = afr_selfheal_locked_inspect(frame, this, loc->gfid, &inode,
- &entry_selfheal, &data_selfheal,
- &metadata_selfheal, &pending);
+ ret = afr_lockless_inspect(frame, this, loc->gfid, &inode, index_vgfid,
+ &entry_selfheal, &data_selfheal,
+ &metadata_selfheal, &pending);
if (ret == -ENOMEM) {
ret = -1;
diff --git a/xlators/cluster/afr/src/afr-self-heal-common.c b/xlators/cluster/afr/src/afr-self-heal-common.c
index d942ccf..1608f75 100644
--- a/xlators/cluster/afr/src/afr-self-heal-common.c
+++ b/xlators/cluster/afr/src/afr-self-heal-common.c
@@ -1827,6 +1827,37 @@ afr_selfheal_unlocked_lookup_on(call_frame_t *frame, inode_t *parent,
return inode;
}
+static int
+afr_set_multi_dom_lock_count_request(xlator_t *this, dict_t *dict)
+{
+ int ret = 0;
+ afr_private_t *priv = NULL;
+ char *key1 = NULL;
+ char *key2 = NULL;
+
+ priv = this->private;
+ key1 = alloca0(strlen(GLUSTERFS_INODELK_DOM_PREFIX) + 2 +
+ strlen(this->name));
+ key2 = alloca0(strlen(GLUSTERFS_INODELK_DOM_PREFIX) + 2 +
+ strlen(priv->sh_domain));
+
+ ret = dict_set_uint32(dict, GLUSTERFS_MULTIPLE_DOM_LK_CNT_REQUESTS, 1);
+ if (ret)
+ return ret;
+
+ sprintf(key1, "%s:%s", GLUSTERFS_INODELK_DOM_PREFIX, this->name);
+ ret = dict_set_uint32(dict, key1, 1);
+ if (ret)
+ return ret;
+
+ sprintf(key2, "%s:%s", GLUSTERFS_INODELK_DOM_PREFIX, priv->sh_domain);
+ ret = dict_set_uint32(dict, key2, 1);
+ if (ret)
+ return ret;
+
+ return 0;
+}
+
int
afr_selfheal_unlocked_discover_on(call_frame_t *frame, inode_t *inode,
uuid_t gfid, struct afr_reply *replies,
@@ -1851,6 +1882,11 @@ afr_selfheal_unlocked_discover_on(call_frame_t *frame, inode_t *inode,
return -ENOMEM;
}
+ if (afr_set_multi_dom_lock_count_request(frame->this, xattr_req)) {
+ dict_unref(xattr_req);
+ return -1;
+ }
+
loc.inode = inode_ref(inode);
gf_uuid_copy(loc.gfid, gfid);
@@ -2241,7 +2277,8 @@ int
afr_selfheal_unlocked_inspect(call_frame_t *frame, xlator_t *this, uuid_t gfid,
inode_t **link_inode, gf_boolean_t *data_selfheal,
gf_boolean_t *metadata_selfheal,
- gf_boolean_t *entry_selfheal)
+ gf_boolean_t *entry_selfheal,
+ struct afr_reply *replies_dst)
{
afr_private_t *priv = NULL;
inode_t *inode = NULL;
@@ -2377,6 +2414,8 @@ afr_selfheal_unlocked_inspect(call_frame_t *frame, xlator_t *this, uuid_t gfid,
ret = 0;
out:
+ if (replies && replies_dst)
+ afr_replies_copy(replies_dst, replies, priv->child_count);
if (inode)
inode_unref(inode);
if (replies)
@@ -2493,7 +2532,7 @@ afr_selfheal_do(call_frame_t *frame, xlator_t *this, uuid_t gfid)
ret = afr_selfheal_unlocked_inspect(frame, this, gfid, &inode,
&data_selfheal, &metadata_selfheal,
- &entry_selfheal);
+ &entry_selfheal, NULL);
if (ret)
goto out;
diff --git a/xlators/cluster/afr/src/afr-self-heal.h b/xlators/cluster/afr/src/afr-self-heal.h
index f7ecf5d..b39af02 100644
--- a/xlators/cluster/afr/src/afr-self-heal.h
+++ b/xlators/cluster/afr/src/afr-self-heal.h
@@ -327,7 +327,8 @@ int
afr_selfheal_unlocked_inspect(call_frame_t *frame, xlator_t *this, uuid_t gfid,
inode_t **link_inode, gf_boolean_t *data_selfheal,
gf_boolean_t *metadata_selfheal,
- gf_boolean_t *entry_selfheal);
+ gf_boolean_t *entry_selfheal,
+ struct afr_reply *replies);
int
afr_selfheal_do(call_frame_t *frame, xlator_t *this, uuid_t gfid);
diff --git a/xlators/features/locks/src/common.h b/xlators/features/locks/src/common.h
index 3a74967..ea86b96 100644
--- a/xlators/features/locks/src/common.h
+++ b/xlators/features/locks/src/common.h
@@ -45,6 +45,10 @@
fd_unref(__local->fd); \
if (__local->inode) \
inode_unref(__local->inode); \
+ if (__local->xdata) { \
+ dict_unref(__local->xdata); \
+ __local->xdata = NULL; \
+ } \
mem_put(__local); \
} \
} while (0)
diff --git a/xlators/features/locks/src/locks.h b/xlators/features/locks/src/locks.h
index b817960..aa267de 100644
--- a/xlators/features/locks/src/locks.h
+++ b/xlators/features/locks/src/locks.h
@@ -239,6 +239,7 @@ typedef struct {
gf_boolean_t inodelk_count_req;
gf_boolean_t posixlk_count_req;
gf_boolean_t parent_entrylk_req;
+ gf_boolean_t multiple_dom_lk_requests;
int update_mlock_enforced_flag;
} pl_local_t;
@@ -260,6 +261,13 @@ typedef struct _locks_ctx {
struct list_head metalk_list;
} pl_ctx_t;
+typedef struct _multi_dom_lk_data {
+ xlator_t *this;
+ inode_t *inode;
+ dict_t *xdata_rsp;
+ gf_boolean_t keep_max;
+} multi_dom_lk_data;
+
typedef enum { DECREMENT, INCREMENT } pl_count_op_t;
pl_ctx_t *
diff --git a/xlators/features/locks/src/posix.c b/xlators/features/locks/src/posix.c
index 4592240..9a14c64 100644
--- a/xlators/features/locks/src/posix.c
+++ b/xlators/features/locks/src/posix.c
@@ -150,13 +150,20 @@ fetch_pathinfo(xlator_t *, inode_t *, int32_t *, char **);
gf_boolean_t
pl_has_xdata_requests(dict_t *xdata)
{
- static char *reqs[] = {GLUSTERFS_ENTRYLK_COUNT, GLUSTERFS_INODELK_COUNT,
- GLUSTERFS_INODELK_DOM_COUNT, GLUSTERFS_POSIXLK_COUNT,
- GLUSTERFS_PARENT_ENTRYLK, NULL};
- static int reqs_size[] = {
- SLEN(GLUSTERFS_ENTRYLK_COUNT), SLEN(GLUSTERFS_INODELK_COUNT),
- SLEN(GLUSTERFS_INODELK_DOM_COUNT), SLEN(GLUSTERFS_POSIXLK_COUNT),
- SLEN(GLUSTERFS_PARENT_ENTRYLK), 0};
+ static char *reqs[] = {GLUSTERFS_ENTRYLK_COUNT,
+ GLUSTERFS_INODELK_COUNT,
+ GLUSTERFS_INODELK_DOM_COUNT,
+ GLUSTERFS_POSIXLK_COUNT,
+ GLUSTERFS_PARENT_ENTRYLK,
+ GLUSTERFS_MULTIPLE_DOM_LK_CNT_REQUESTS,
+ NULL};
+ static int reqs_size[] = {SLEN(GLUSTERFS_ENTRYLK_COUNT),
+ SLEN(GLUSTERFS_INODELK_COUNT),
+ SLEN(GLUSTERFS_INODELK_DOM_COUNT),
+ SLEN(GLUSTERFS_POSIXLK_COUNT),
+ SLEN(GLUSTERFS_PARENT_ENTRYLK),
+ SLEN(GLUSTERFS_MULTIPLE_DOM_LK_CNT_REQUESTS),
+ 0};
int i = 0;
if (!xdata)
@@ -169,12 +176,22 @@ pl_has_xdata_requests(dict_t *xdata)
return _gf_false;
}
+static int
+dict_delete_domain_key(dict_t *dict, char *key, data_t *value, void *data)
+{
+ dict_del(dict, key);
+ return 0;
+}
+
void
pl_get_xdata_requests(pl_local_t *local, dict_t *xdata)
{
if (!local || !xdata)
return;
+ GF_ASSERT(local->xdata == NULL);
+ local->xdata = dict_copy_with_ref(xdata, NULL);
+
if (dict_get_sizen(xdata, GLUSTERFS_ENTRYLK_COUNT)) {
local->entrylk_count_req = 1;
dict_del_sizen(xdata, GLUSTERFS_ENTRYLK_COUNT);
@@ -183,6 +200,12 @@ pl_get_xdata_requests(pl_local_t *local, dict_t *xdata)
local->inodelk_count_req = 1;
dict_del_sizen(xdata, GLUSTERFS_INODELK_COUNT);
}
+ if (dict_get_sizen(xdata, GLUSTERFS_MULTIPLE_DOM_LK_CNT_REQUESTS)) {
+ local->multiple_dom_lk_requests = 1;
+ dict_del_sizen(xdata, GLUSTERFS_MULTIPLE_DOM_LK_CNT_REQUESTS);
+ dict_foreach_fnmatch(xdata, GLUSTERFS_INODELK_DOM_PREFIX "*",
+ dict_delete_domain_key, NULL);
+ }
local->inodelk_dom_count_req = dict_get_sizen(xdata,
GLUSTERFS_INODELK_DOM_COUNT);
@@ -210,7 +233,7 @@ pl_needs_xdata_response(pl_local_t *local)
if (local->parent_entrylk_req || local->entrylk_count_req ||
local->inodelk_dom_count_req || local->inodelk_count_req ||
- local->posixlk_count_req)
+ local->posixlk_count_req || local->multiple_dom_lk_requests)
return _gf_true;
return _gf_false;
@@ -411,6 +434,75 @@ pl_posixlk_xattr_fill(xlator_t *this, inode_t *inode, dict_t *dict,
}
void
+pl_inodelk_xattr_fill_each(xlator_t *this, inode_t *inode, dict_t *dict,
+ char *domname, gf_boolean_t keep_max, char *key)
+{
+ int32_t count = 0;
+ int32_t maxcount = -1;
+ int ret = -1;
+
+ if (keep_max) {
+ ret = dict_get_int32(dict, key, &maxcount);
+ if (ret < 0)
+ gf_msg_debug(this->name, 0, " Failed to fetch the value for key %s",
+ GLUSTERFS_INODELK_COUNT);
+ }
+ count = get_inodelk_count(this, inode, domname);
+ if (maxcount >= count)
+ return;
+
+ ret = dict_set_int32(dict, key, count);
+ if (ret < 0) {
+ gf_msg_debug(this->name, 0,
+ "Failed to set count for "
+ "key %s",
+ key);
+ }
+
+ return;
+}
+
+static int
+pl_inodelk_xattr_fill_multiple(dict_t *this, char *key, data_t *value,
+ void *data)
+{
+ multi_dom_lk_data *d = data;
+ char *tmp_key = NULL;
+ char *save_ptr = NULL;
+
+ tmp_key = gf_strdup(key);
+ strtok_r(tmp_key, ":", &save_ptr);
+ if (!*save_ptr) {
+ gf_msg(THIS->name, GF_LOG_ERROR, 0, EINVAL,
+ "Could not tokenize domain string from key %s", key);
+ return -1;
+ }
+
+ pl_inodelk_xattr_fill_each(d->this, d->inode, d->xdata_rsp, save_ptr,
+ d->keep_max, key);
+ if (tmp_key)
+ GF_FREE(tmp_key);
+
+ return 0;
+}
+
+void
+pl_fill_multiple_dom_lk_requests(xlator_t *this, pl_local_t *local,
+ inode_t *inode, dict_t *dict,
+ gf_boolean_t keep_max)
+{
+ multi_dom_lk_data data;
+
+ data.this = this;
+ data.inode = inode;
+ data.xdata_rsp = dict;
+ data.keep_max = keep_max;
+
+ dict_foreach_fnmatch(local->xdata, GLUSTERFS_INODELK_DOM_PREFIX "*",
+ pl_inodelk_xattr_fill_multiple, &data);
+}
+
+void
pl_set_xdata_response(xlator_t *this, pl_local_t *local, inode_t *parent,
inode_t *inode, char *name, dict_t *xdata,
gf_boolean_t max_lock)
@@ -437,6 +529,9 @@ pl_set_xdata_response(xlator_t *this, pl_local_t *local, inode_t *parent,
if (local->posixlk_count_req)
pl_posixlk_xattr_fill(this, inode, xdata, max_lock);
+
+ if (local->multiple_dom_lk_requests)
+ pl_fill_multiple_dom_lk_requests(this, local, inode, xdata, max_lock);
}
/* Checks whether the region where fop is acting upon conflicts
@@ -773,9 +868,6 @@ pl_truncate_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
{
pl_local_t *local = frame->local;
- if (local->xdata)
- dict_unref(local->xdata);
-
pl_track_io_fop_count(local, this, DECREMENT);
if (local->op == GF_FOP_TRUNCATE)
@@ -932,9 +1024,6 @@ unwind:
"ret: %d, error: %s",
op_ret, strerror(op_errno));
- if (local->xdata)
- dict_unref(local->xdata);
-
switch (local->op) {
case GF_FOP_TRUNCATE:
PL_STACK_UNWIND(truncate, xdata, frame, op_ret, op_errno, buf,
--
1.8.3.1

187
SOURCES/0425-tests-Fix-spurious-self-heald.t-failure.patch Normal file
View File

@ -0,0 +1,187 @@
From 2c582ea6c76031463501b31d9250e739d5aeda79 Mon Sep 17 00:00:00 2001
From: Ravishankar N <ravishankar@redhat.com>
Date: Fri, 5 Jun 2020 14:28:11 +0530
Subject: [PATCH 425/449] tests: Fix spurious self-heald.t failure
Problem:
heal-info code assumes that all indices in xattrop directory
definitely need heal. There is one corner case.
The very first xattrop on the file will lead to adding the
gfid to 'xattrop' index in fop path and in _cbk path it is
removed because the fop is zero-xattr xattrop in success case.
These gfids could be read by heal-info and shown as needing heal.
Fix:
Check the pending flag to see if the file definitely needs or
not instead of which index is being crawled at the moment.
> Upstream patch: https://review.gluster.org/#/c/glusterfs/+/24110/
> fixes: bz#1801623
> Change-Id: I79f00dc7366fedbbb25ec4bec838dba3b34c7ad5
> Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
BUG: 1721355
Change-Id: I7efdf45a5158fadfdbdd21c91837f193d80fa6c7
Signed-off-by: Ravishankar N <ravishankar@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202491
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Pranith Kumar Karampuri <pkarampu@redhat.com>
---
heal/src/glfs-heal.c | 17 ++++++----------
xlators/cluster/afr/src/afr-common.c | 38 ++++++++++++++----------------------
2 files changed, 21 insertions(+), 34 deletions(-)
diff --git a/heal/src/glfs-heal.c b/heal/src/glfs-heal.c
index 5af9e31..125b12c 100644
--- a/heal/src/glfs-heal.c
+++ b/heal/src/glfs-heal.c
@@ -775,8 +775,7 @@ static int
glfsh_process_entries(xlator_t *xl, fd_t *fd, gf_dirent_t *entries,
uint64_t *offset, num_entries_t *num_entries,
print_status glfsh_print_status,
- gf_boolean_t ignore_dirty, glfsh_fail_mode_t mode,
- dict_t *xattr_req)
+ gf_boolean_t ignore_dirty, glfsh_fail_mode_t mode)
{
gf_dirent_t *entry = NULL;
gf_dirent_t *tmp = NULL;
@@ -808,7 +807,7 @@ glfsh_process_entries(xlator_t *xl, fd_t *fd, gf_dirent_t *entries,
gf_uuid_parse(entry->d_name, gfid);
gf_uuid_copy(loc.gfid, gfid);
- ret = syncop_getxattr(this, &loc, &dict, GF_HEAL_INFO, xattr_req, NULL);
+ ret = syncop_getxattr(this, &loc, &dict, GF_HEAL_INFO, NULL, NULL);
if (ret) {
if ((mode != GLFSH_MODE_CONTINUE_ON_ERROR) && (ret == -ENOTCONN))
goto out;
@@ -877,19 +876,19 @@ glfsh_crawl_directory(glfs_t *fs, xlator_t *top_subvol, loc_t *rootloc,
if (heal_op == GF_SHD_OP_INDEX_SUMMARY) {
ret = glfsh_process_entries(readdir_xl, fd, &entries, &offset,
num_entries, glfsh_print_heal_status,
- ignore, mode, xattr_req);
+ ignore, mode);
if (ret < 0)
goto out;
} else if (heal_op == GF_SHD_OP_SPLIT_BRAIN_FILES) {
ret = glfsh_process_entries(readdir_xl, fd, &entries, &offset,
num_entries, glfsh_print_spb_status,
- ignore, mode, xattr_req);
+ ignore, mode);
if (ret < 0)
goto out;
} else if (heal_op == GF_SHD_OP_HEAL_SUMMARY) {
ret = glfsh_process_entries(readdir_xl, fd, &entries, &offset,
num_entries, glfsh_print_summary_status,
- ignore, mode, xattr_req);
+ ignore, mode);
if (ret < 0)
goto out;
} else if (heal_op == GF_SHD_OP_SBRAIN_HEAL_FROM_BRICK) {
@@ -898,7 +897,7 @@ glfsh_crawl_directory(glfs_t *fs, xlator_t *top_subvol, loc_t *rootloc,
} else if (heal_op == GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE) {
ret = glfsh_process_entries(readdir_xl, fd, &entries, &offset,
num_entries, glfsh_heal_status_boolean,
- ignore, mode, xattr_req);
+ ignore, mode);
if (ret < 0)
goto out;
}
@@ -952,10 +951,6 @@ glfsh_print_pending_heals_type(glfs_t *fs, xlator_t *top_subvol, loc_t *rootloc,
int32_t op_errno = 0;
gf_boolean_t ignore = _gf_false;
- ret = dict_set_str(xattr_req, "index-vgfid", vgfid);
- if (ret)
- return ret;
-
if (!strcmp(vgfid, GF_XATTROP_DIRTY_GFID))
ignore = _gf_true;
diff --git a/xlators/cluster/afr/src/afr-common.c b/xlators/cluster/afr/src/afr-common.c
index c355ec5..89e2483 100644
--- a/xlators/cluster/afr/src/afr-common.c
+++ b/xlators/cluster/afr/src/afr-common.c
@@ -5995,8 +5995,8 @@ afr_is_dirty_count_non_unary(xlator_t *this, struct afr_reply *replies,
static int
afr_update_heal_status(xlator_t *this, struct afr_reply *replies,
- char *index_vgfid, ia_type_t ia_type, gf_boolean_t *esh,
- gf_boolean_t *dsh, gf_boolean_t *msh)
+ ia_type_t ia_type, gf_boolean_t *esh, gf_boolean_t *dsh,
+ gf_boolean_t *msh, unsigned char pending)
{
int ret = -1;
GF_UNUSED int ret1 = 0;
@@ -6026,14 +6026,7 @@ afr_update_heal_status(xlator_t *this, struct afr_reply *replies,
}
}
- if (!strcmp(index_vgfid, GF_XATTROP_INDEX_GFID)) {
- if (shd_domain_lk_count) {
- ret = -EAGAIN; /*For 'possibly-healing'. */
- } else {
- ret = 0; /*needs heal. Just set a non -ve value so that it is
- assumed as the source index.*/
- }
- } else if (!strcmp(index_vgfid, GF_XATTROP_DIRTY_GFID)) {
+ if (!pending) {
if ((afr_is_dirty_count_non_unary(this, replies, ia_type)) ||
(!io_domain_lk_count)) {
/* Needs heal. */
@@ -6042,6 +6035,13 @@ afr_update_heal_status(xlator_t *this, struct afr_reply *replies,
/* No heal needed. */
*dsh = *esh = *msh = 0;
}
+ } else {
+ if (shd_domain_lk_count) {
+ ret = -EAGAIN; /*For 'possibly-healing'. */
+ } else {
+ ret = 0; /*needs heal. Just set a non -ve value so that it is
+ assumed as the source index.*/
+ }
}
return ret;
}
@@ -6049,8 +6049,8 @@ afr_update_heal_status(xlator_t *this, struct afr_reply *replies,
/*return EIO, EAGAIN or pending*/
int
afr_lockless_inspect(call_frame_t *frame, xlator_t *this, uuid_t gfid,
- inode_t **inode, char *index_vgfid,
- gf_boolean_t *entry_selfheal, gf_boolean_t *data_selfheal,
+ inode_t **inode, gf_boolean_t *entry_selfheal,
+ gf_boolean_t *data_selfheal,
gf_boolean_t *metadata_selfheal, unsigned char *pending)
{
int ret = -1;
@@ -6109,8 +6109,8 @@ afr_lockless_inspect(call_frame_t *frame, xlator_t *this, uuid_t gfid,
goto out;
}
- ret = afr_update_heal_status(this, replies, index_vgfid, (*inode)->ia_type,
- &esh, &dsh, &msh);
+ ret = afr_update_heal_status(this, replies, (*inode)->ia_type, &esh, &dsh,
+ &msh, *pending);
out:
*data_selfheal = dsh;
*entry_selfheal = esh;
@@ -6133,16 +6133,8 @@ afr_get_heal_info(call_frame_t *frame, xlator_t *this, loc_t *loc)
inode_t *inode = NULL;
char *substr = NULL;
char *status = NULL;
- afr_local_t *local = NULL;
- char *index_vgfid = NULL;
-
- local = frame->local;
- if (dict_get_str(local->xdata_req, "index-vgfid", &index_vgfid)) {
- ret = -1;
- goto out;
- }
- ret = afr_lockless_inspect(frame, this, loc->gfid, &inode, index_vgfid,
+ ret = afr_lockless_inspect(frame, this, loc->gfid, &inode,
&entry_selfheal, &data_selfheal,
&metadata_selfheal, &pending);
--
1.8.3.1

216
SOURCES/0426-geo-rep-Fix-for-Transport-End-Point-not-connected-is.patch Normal file
View File

@ -0,0 +1,216 @@
From 91936fe5ef854bd9d2f91e643795d0e7791b97ba Mon Sep 17 00:00:00 2001
From: Harpreet Kaur <hlalwani@redhat.com>
Date: Mon, 7 Jan 2019 16:38:25 +0530
Subject: [PATCH 426/449] geo-rep: Fix for "Transport End Point not connected"
issue
problem: Geo-rep gsyncd process mounts the master and slave volume
on master nodes and slave nodes respectively and starts
the sync. But it doesn't wait for the mount to be in ready
state to accept I/O. The gluster mount is considered to be
ready when all the distribute sub-volumes is up. If the all
the distribute subvolumes are not up, it can cause ENOTCONN
error, when lookup on file comes and file is on the subvol
that is down.
solution: Added a Virtual Xattr "dht.subvol.status" which returns "1"
if all subvols are up and "0" if all subvols are not up.
Geo-rep then uses this virtual xattr after a fresh mount, to
check whether all subvols are up or not and then starts the
I/O.
>fixes: bz#1664335
>Change-Id: If3ad01d728b1372da7c08ccbe75a45bdc1ab2a91
>Signed-off-by: Harpreet Kaur <hlalwani@redhat.com>
>Signed-off-by: Kotresh HR <khiremat@redhat.com>
backport of https://review.gluster.org/#/c/glusterfs/+/22001/
BUG: 1640573
Change-Id: If3ad01d728b1372da7c08ccbe75a45bdc1ab2a91
Signed-off-by: Shwetha K Acharya <sacharya@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202554
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
geo-replication/syncdaemon/resource.py | 11 ++++++
geo-replication/syncdaemon/syncdutils.py | 20 +++++++++--
xlators/cluster/dht/src/dht-common.c | 59 ++++++++++++++++++++++++++++++++
xlators/cluster/dht/src/dht-common.h | 4 +++
4 files changed, 91 insertions(+), 3 deletions(-)
diff --git a/geo-replication/syncdaemon/resource.py b/geo-replication/syncdaemon/resource.py
index 189d8a1..0c61de9 100644
--- a/geo-replication/syncdaemon/resource.py
+++ b/geo-replication/syncdaemon/resource.py
@@ -37,6 +37,7 @@ from syncdutils import ChangelogException, ChangelogHistoryNotAvailable
from syncdutils import get_changelog_log_level, get_rsync_version
from syncdutils import CHANGELOG_AGENT_CLIENT_VERSION
from syncdutils import GX_GFID_CANONICAL_LEN
+from syncdutils import gf_mount_ready
from gsyncdstatus import GeorepStatus
from syncdutils import lf, Popen, sup
from syncdutils import Xattr, matching_disk_gfid, get_gfid_from_mnt
@@ -950,6 +951,16 @@ class Mounter(object):
logging.exception('mount cleanup failure:')
rv = 200
os._exit(rv)
+
+ #Polling the dht.subvol.status value.
+ RETRIES = 10
+ while not gf_mount_ready():
+ if RETRIES < 0:
+ logging.error('Subvols are not up')
+ break
+ RETRIES -= 1
+ time.sleep(0.2)
+
logging.debug('auxiliary glusterfs mount prepared')
diff --git a/geo-replication/syncdaemon/syncdutils.py b/geo-replication/syncdaemon/syncdutils.py
index b08098e..7560fa1 100644
--- a/geo-replication/syncdaemon/syncdutils.py
+++ b/geo-replication/syncdaemon/syncdutils.py
@@ -21,8 +21,8 @@ import subprocess
import socket
from subprocess import PIPE
from threading import Lock, Thread as baseThread
-from errno import EACCES, EAGAIN, EPIPE, ENOTCONN, ECONNABORTED
-from errno import EINTR, ENOENT, ESTALE, EBUSY, errorcode
+from errno import EACCES, EAGAIN, EPIPE, ENOTCONN, ENOMEM, ECONNABORTED
+from errno import EINTR, ENOENT, ESTALE, EBUSY, ENODATA, errorcode
from signal import signal, SIGTERM
import select as oselect
from os import waitpid as owaitpid
@@ -55,6 +55,8 @@ from rconf import rconf
from hashlib import sha256 as sha256
+ENOTSUP = getattr(errno, 'ENOTSUP', 'EOPNOTSUPP')
+
# auxiliary gfid based access prefix
_CL_AUX_GFID_PFX = ".gfid/"
ROOT_GFID = "00000000-0000-0000-0000-000000000001"
@@ -100,6 +102,19 @@ def unescape_space_newline(s):
.replace(NEWLINE_ESCAPE_CHAR, "\n")\
.replace(PERCENTAGE_ESCAPE_CHAR, "%")
+# gf_mount_ready() returns 1 if all subvols are up, else 0
+def gf_mount_ready():
+ ret = errno_wrap(Xattr.lgetxattr,
+ ['.', 'dht.subvol.status', 16],
+ [ENOENT, ENOTSUP, ENODATA], [ENOMEM])
+
+ if isinstance(ret, int):
+ logging.error("failed to get the xattr value")
+ return 1
+ ret = ret.rstrip('\x00')
+ if ret == "1":
+ return 1
+ return 0
def norm(s):
if s:
@@ -564,7 +579,6 @@ def errno_wrap(call, arg=[], errnos=[], retry_errnos=[]):
def lstat(e):
return errno_wrap(os.lstat, [e], [ENOENT], [ESTALE, EBUSY])
-
def get_gfid_from_mnt(gfidpath):
return errno_wrap(Xattr.lgetxattr,
[gfidpath, 'glusterfs.gfid.string',
diff --git a/xlators/cluster/dht/src/dht-common.c b/xlators/cluster/dht/src/dht-common.c
index 6aa18f3..23cc80c 100644
--- a/xlators/cluster/dht/src/dht-common.c
+++ b/xlators/cluster/dht/src/dht-common.c
@@ -4858,6 +4858,60 @@ out:
return 0;
}
+/* Virtual Xattr which returns 1 if all subvols are up,
+ else returns 0. Geo-rep then uses this virtual xattr
+ after a fresh mount and starts the I/O.
+*/
+
+enum dht_vxattr_subvol {
+ DHT_VXATTR_SUBVOLS_UP = 1,
+ DHT_VXATTR_SUBVOLS_DOWN = 0,
+};
+
+int
+dht_vgetxattr_subvol_status(call_frame_t *frame, xlator_t *this,
+ const char *key)
+{
+ dht_local_t *local = NULL;
+ int ret = -1;
+ int op_errno = ENODATA;
+ int value = DHT_VXATTR_SUBVOLS_UP;
+ int i = 0;
+ dht_conf_t *conf = NULL;
+
+ conf = this->private;
+ local = frame->local;
+
+ if (!key) {
+ op_errno = EINVAL;
+ goto out;
+ }
+ local->xattr = dict_new();
+ if (!local->xattr) {
+ op_errno = ENOMEM;
+ goto out;
+ }
+ for (i = 0; i < conf->subvolume_cnt; i++) {
+ if (!conf->subvolume_status[i]) {
+ value = DHT_VXATTR_SUBVOLS_DOWN;
+ gf_msg_debug(this->name, 0, "subvol %s is down ",
+ conf->subvolumes[i]->name);
+ break;
+ }
+ }
+ ret = dict_set_int8(local->xattr, (char *)key, value);
+ if (ret < 0) {
+ op_errno = -ret;
+ ret = -1;
+ goto out;
+ }
+ ret = 0;
+
+out:
+ DHT_STACK_UNWIND(getxattr, frame, ret, op_errno, local->xattr, NULL);
+ return 0;
+}
+
int
dht_getxattr(call_frame_t *frame, xlator_t *this, loc_t *loc, const char *key,
dict_t *xdata)
@@ -4915,6 +4969,11 @@ dht_getxattr(call_frame_t *frame, xlator_t *this, loc_t *loc, const char *key,
goto err;
}
+ if (strncmp(key, DHT_SUBVOL_STATUS_KEY, SLEN(DHT_SUBVOL_STATUS_KEY)) == 0) {
+ dht_vgetxattr_subvol_status(frame, this, key);
+ return 0;
+ }
+
/* skip over code which is irrelevant if !DHT_IS_DIR(layout) */
if (!DHT_IS_DIR(layout))
goto no_dht_is_dir;
diff --git a/xlators/cluster/dht/src/dht-common.h b/xlators/cluster/dht/src/dht-common.h
index 1b3e826..9ec5b51 100644
--- a/xlators/cluster/dht/src/dht-common.h
+++ b/xlators/cluster/dht/src/dht-common.h
@@ -45,6 +45,10 @@
#define DHT_DIR_STAT_BLOCKS 8
#define DHT_DIR_STAT_SIZE 4096
+/* Virtual xattr for subvols status */
+
+#define DHT_SUBVOL_STATUS_KEY "dht.subvol.status"
+
/* Virtual xattrs for debugging */
#define DHT_DBG_HASHED_SUBVOL_PATTERN "dht.file.hashed-subvol.*"
--
1.8.3.1

38
SOURCES/0427-storage-posix-Fixing-a-coverity-issue.patch Normal file
View File

@ -0,0 +1,38 @@
From 3943fce5818a353117fc1c492e6383434d742979 Mon Sep 17 00:00:00 2001
From: Barak Sason <bsasonro@redhat.com>
Date: Sun, 18 Aug 2019 17:52:04 +0300
Subject: [PATCH 427/449] storage/posix - Fixing a coverity issue
Fixed a resource leak of variable 'pfd'
backport of https://review.gluster.org/#/c/glusterfs/+/23261/
>CID: 1400673
>Updates: bz#789278
>Change-Id: I78e1e8a89e0604b56e35a75c25d436b35db096c3
>Signed-off-by: Barak Sason <bsasonro@redhat.com>
BUG: 1787310
Change-Id: I78e1e8a89e0604b56e35a75c25d436b35db096c3
Signed-off-by: Barak Sason Rofman <bsasonro@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202563
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/storage/posix/src/posix-inode-fd-ops.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/xlators/storage/posix/src/posix-inode-fd-ops.c b/xlators/storage/posix/src/posix-inode-fd-ops.c
index bcce06e..5748b9f 100644
--- a/xlators/storage/posix/src/posix-inode-fd-ops.c
+++ b/xlators/storage/posix/src/posix-inode-fd-ops.c
@@ -1603,6 +1603,7 @@ posix_open(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
if (op_ret == -1) {
gf_msg(this->name, GF_LOG_ERROR, errno, P_MSG_FSTAT_FAILED,
"pre-operation fstat failed on fd=%p", fd);
+ GF_FREE(pfd);
goto out;
}
--
1.8.3.1

48
SOURCES/0428-glusterd-ganesha-fixing-resource-leak-in-tear_down_c.patch Normal file
View File

@ -0,0 +1,48 @@
From 7e1bf1e338a6effe209f57b1b92a70d5d25a73bf Mon Sep 17 00:00:00 2001
From: Jiffin Tony Thottan <jthottan@redhat.com>
Date: Mon, 26 Aug 2019 11:32:18 +0530
Subject: [PATCH 428/449] glusterd/ganesha: fixing resource leak in
tear_down_cluster()
backport of https://review.gluster.org/#/c/glusterfs/+/23295/
>CID: 1370947
>Updates: bz#789278
>Change-Id: Ib694056430ff0536ed705a0e77e5ace22486891e
>Signed-off-by: Jiffin Tony Thottan <jthottan@redhat.com>
BUG: 1787310
Change-Id: Ib694056430ff0536ed705a0e77e5ace22486891e
Signed-off-by: Barak Sason Rofman <bsasonro@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202561
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/mgmt/glusterd/src/glusterd-ganesha.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/xlators/mgmt/glusterd/src/glusterd-ganesha.c b/xlators/mgmt/glusterd/src/glusterd-ganesha.c
index 0a16925..06f028f 100644
--- a/xlators/mgmt/glusterd/src/glusterd-ganesha.c
+++ b/xlators/mgmt/glusterd/src/glusterd-ganesha.c
@@ -659,10 +659,18 @@ tear_down_cluster(gf_boolean_t run_teardown)
"Failed to close dir %s. Reason :"
" %s",
CONFDIR, strerror(errno));
+ goto exit;
}
}
out:
+ if (dir && sys_closedir(dir)) {
+ gf_msg_debug(THIS->name, 0,
+ "Failed to close dir %s. Reason :"
+ " %s",
+ CONFDIR, strerror(errno));
+ }
+exit:
return ret;
}
--
1.8.3.1

61
SOURCES/0429-dht-rebalance-fixing-failure-occurace-due-to-rebalan.patch Normal file
View File

@ -0,0 +1,61 @@
From 1370db202a2a60810409f74c390448bf8fbd6998 Mon Sep 17 00:00:00 2001
From: Barak Sason Rofman <bsasonro@redhat.com>
Date: Sun, 9 Feb 2020 15:09:30 +0200
Subject: [PATCH 429/449] dht/rebalance - fixing failure occurace due to
rebalance stop
Probelm description:
When topping rebalance, the following error messages appear in the
rebalance log file:
[2020-01-28 14:31:42.452070] W [dht-rebalance.c:3447:gf_defrag_process_dir] 0-distrep-dht: Found error from gf_defrag_get_entry
[2020-01-28 14:31:42.452764] E [MSGID: 109111] [dht-rebalance.c:3971:gf_defrag_fix_layout] 0-distrep-dht: gf_defrag_process_dir failed for directory: /0/1/2/3/4/5/6/7/8/9/10/11/12/13/14/15/16/17/18/19/20/21/22/23/24/25/26/27/28/29/30/31
[2020-01-28 14:31:42.453498] E [MSGID: 109016] [dht-rebalance.c:3906:gf_defrag_fix_layout] 0-distrep-dht: Fix layout failed for /0/1/2/3/4/5/6/7/8/9/10/11/12/13/14/15/16/17/18/19/20/21/22/23/24/25/26/27/28/29/30
In order to avoid seing these error messages, a modification to the
error handling mechanism has been made.
In addition, several log messages had been added in order to improve debugging efficiency
backport of https://review.gluster.org/#/c/glusterfs/+/24103/
>fixes: bz#1800956
>Change-Id: Ifc82dae79ab3da9fe22ee25088a2a6b855afcfcf
>Signed-off-by: Barak Sason Rofman <bsasonro@redhat.com>
BUG: 1286171
Change-Id: Ifc82dae79ab3da9fe22ee25088a2a6b855afcfcf
Signed-off-by: Barak Sason Rofman <bsasonro@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202562
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/cluster/dht/src/dht-rebalance.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/xlators/cluster/dht/src/dht-rebalance.c b/xlators/cluster/dht/src/dht-rebalance.c
index 8f31dca..88b6b54 100644
--- a/xlators/cluster/dht/src/dht-rebalance.c
+++ b/xlators/cluster/dht/src/dht-rebalance.c
@@ -3479,6 +3479,10 @@ gf_defrag_process_dir(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
migrate_data, dir_dfmeta, xattr_req,
&should_commit_hash, perrno);
+ if (defrag->defrag_status == GF_DEFRAG_STATUS_STOPPED) {
+ goto out;
+ }
+
if (ret) {
gf_log(this->name, GF_LOG_WARNING,
"Found "
@@ -3935,6 +3939,10 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
ret = gf_defrag_fix_layout(this, defrag, &entry_loc, fix_layout,
migrate_data);
+ if (defrag->defrag_status == GF_DEFRAG_STATUS_STOPPED) {
+ goto out;
+ }
+
if (ret && ret != 2) {
gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LAYOUT_FIX_FAILED,
"Fix layout failed for %s", entry_loc.path);
--
1.8.3.1

291
SOURCES/0430-Fix-some-Null-pointer-dereference-coverity-issues.patch Normal file
View File

@ -0,0 +1,291 @@
From 7fe500a03d42dba6082c28ef7284c950c44fbfa3 Mon Sep 17 00:00:00 2001
From: Xavi Hernandez <xhernandez@redhat.com>
Date: Wed, 22 May 2019 17:46:19 +0200
Subject: [PATCH 430/449] Fix some "Null pointer dereference" coverity issues
This patch fixes the following CID's:
* 1124829
* 1274075
* 1274083
* 1274128
* 1274135
* 1274141
* 1274143
* 1274197
* 1274205
* 1274210
* 1274211
* 1288801
* 1398629
Backport of:
> Upstream-patch-link: https://review.gluster.org/22767
> Change-Id: Ia7c86cfab3245b20777ffa296e1a59748040f558
> Updates: bz#789278
> Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
BUG: 1787310
Change-Id: Ia7c86cfab3245b20777ffa296e1a59748040f558
Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202616
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
cli/src/cli-cmd-system.c | 2 +-
cli/src/cli-xml-output.c | 2 +-
glusterfsd/src/glusterfsd.c | 24 +++++++++++++-----------
libglusterfs/src/inode.c | 3 +++
rpc/rpc-lib/src/rpcsvc.c | 4 ++++
xlators/cluster/dht/src/dht-shared.c | 4 ++++
xlators/cluster/dht/src/switch.c | 9 +++++++++
xlators/features/trash/src/trash.c | 2 +-
xlators/mgmt/glusterd/src/glusterd-geo-rep.c | 7 +++++--
xlators/nfs/server/src/mount3.c | 6 ++++++
xlators/protocol/client/src/client.c | 7 ++++++-
xlators/storage/posix/src/posix-helpers.c | 3 +++
12 files changed, 56 insertions(+), 17 deletions(-)
diff --git a/cli/src/cli-cmd-system.c b/cli/src/cli-cmd-system.c
index 8cd1542..cb3a9ea 100644
--- a/cli/src/cli-cmd-system.c
+++ b/cli/src/cli-cmd-system.c
@@ -446,7 +446,7 @@ cli_cmd_sys_exec_cbk(struct cli_state *state, struct cli_cmd_word *word,
dict_t *dict = NULL;
cli_local_t *local = NULL;
- if (wordcount < 3) {
+ if ((wordcount < 3) || (words[2] == NULL)) {
cli_usage_out(word->pattern);
goto out;
}
diff --git a/cli/src/cli-xml-output.c b/cli/src/cli-xml-output.c
index 006e2fb..903997c 100644
--- a/cli/src/cli-xml-output.c
+++ b/cli/src/cli-xml-output.c
@@ -64,7 +64,7 @@ cli_begin_xml_output(xmlTextWriterPtr *writer, xmlDocPtr *doc)
int ret = -1;
*writer = xmlNewTextWriterDoc(doc, 0);
- if (writer == NULL) {
+ if (*writer == NULL) {
ret = -1;
goto out;
}
diff --git a/glusterfsd/src/glusterfsd.c b/glusterfsd/src/glusterfsd.c
index 974fb88..9821180 100644
--- a/glusterfsd/src/glusterfsd.c
+++ b/glusterfsd/src/glusterfsd.c
@@ -1235,19 +1235,21 @@ parse_opts(int key, char *arg, struct argp_state *state)
case ARGP_BRICK_PORT_KEY:
n = 0;
- port_str = strtok_r(arg, ",", &tmp_str);
- if (gf_string2uint_base10(port_str, &n) == 0) {
- cmd_args->brick_port = n;
- port_str = strtok_r(NULL, ",", &tmp_str);
- if (port_str) {
- if (gf_string2uint_base10(port_str, &n) == 0) {
- cmd_args->brick_port2 = n;
- break;
+ if (arg != NULL) {
+ port_str = strtok_r(arg, ",", &tmp_str);
+ if (gf_string2uint_base10(port_str, &n) == 0) {
+ cmd_args->brick_port = n;
+ port_str = strtok_r(NULL, ",", &tmp_str);
+ if (port_str) {
+ if (gf_string2uint_base10(port_str, &n) == 0) {
+ cmd_args->brick_port2 = n;
+ break;
+ }
+ argp_failure(state, -1, 0,
+ "wrong brick (listen) port %s", arg);
}
- argp_failure(state, -1, 0, "wrong brick (listen) port %s",
- arg);
+ break;
}
- break;
}
argp_failure(state, -1, 0, "unknown brick (listen) port %s", arg);
diff --git a/libglusterfs/src/inode.c b/libglusterfs/src/inode.c
index 9dbb25b..4c3c546 100644
--- a/libglusterfs/src/inode.c
+++ b/libglusterfs/src/inode.c
@@ -899,6 +899,9 @@ inode_resolve(inode_table_t *table, char *path)
parent = inode_ref(table->root);
str = tmp = gf_strdup(path);
+ if (str == NULL) {
+ goto out;
+ }
while (1) {
bname = strtok_r(str, "/", &saveptr);
diff --git a/rpc/rpc-lib/src/rpcsvc.c b/rpc/rpc-lib/src/rpcsvc.c
index 5a35139..b058932 100644
--- a/rpc/rpc-lib/src/rpcsvc.c
+++ b/rpc/rpc-lib/src/rpcsvc.c
@@ -2874,6 +2874,10 @@ rpcsvc_transport_peer_check_search(dict_t *options, char *pattern, char *ip,
}
dup_addrstr = gf_strdup(addrstr);
+ if (dup_addrstr == NULL) {
+ ret = -1;
+ goto err;
+ }
addrtok = strtok_r(dup_addrstr, ",", &svptr);
while (addrtok) {
/* CASEFOLD not present on Solaris */
diff --git a/xlators/cluster/dht/src/dht-shared.c b/xlators/cluster/dht/src/dht-shared.c
index ea4b7c6..58e3339 100644
--- a/xlators/cluster/dht/src/dht-shared.c
+++ b/xlators/cluster/dht/src/dht-shared.c
@@ -278,6 +278,10 @@ dht_parse_decommissioned_bricks(xlator_t *this, dht_conf_t *conf,
goto out;
dup_brick = gf_strdup(bricks);
+ if (dup_brick == NULL) {
+ goto out;
+ }
+
node = strtok_r(dup_brick, ",", &tmpstr);
while (node) {
for (i = 0; i < conf->subvolume_cnt; i++) {
diff --git a/xlators/cluster/dht/src/switch.c b/xlators/cluster/dht/src/switch.c
index a782fcd..207d109 100644
--- a/xlators/cluster/dht/src/switch.c
+++ b/xlators/cluster/dht/src/switch.c
@@ -610,9 +610,15 @@ set_switch_pattern(xlator_t *this, dht_conf_t *conf, const char *pattern_str)
/* Get the pattern for considering switch case.
"option block-size *avi:10MB" etc */
option_string = gf_strdup(pattern_str);
+ if (option_string == NULL) {
+ goto err;
+ }
switch_str = strtok_r(option_string, ";", &tmp_str);
while (switch_str) {
dup_str = gf_strdup(switch_str);
+ if (dup_str == NULL) {
+ goto err;
+ }
switch_opt = GF_CALLOC(1, sizeof(struct switch_struct),
gf_switch_mt_switch_struct);
if (!switch_opt) {
@@ -647,6 +653,9 @@ set_switch_pattern(xlator_t *this, dht_conf_t *conf, const char *pattern_str)
if (childs) {
dup_childs = gf_strdup(childs);
+ if (dup_childs == NULL) {
+ goto err;
+ }
child = strtok_r(dup_childs, ",", &tmp);
while (child) {
if (gf_switch_valid_child(this, child)) {
diff --git a/xlators/features/trash/src/trash.c b/xlators/features/trash/src/trash.c
index d668436..f96ed73 100644
--- a/xlators/features/trash/src/trash.c
+++ b/xlators/features/trash/src/trash.c
@@ -170,7 +170,7 @@ store_eliminate_path(char *str, trash_elim_path **eliminate)
int ret = 0;
char *strtokptr = NULL;
- if (eliminate == NULL) {
+ if ((str == NULL) || (eliminate == NULL)) {
ret = EINVAL;
goto out;
}
diff --git a/xlators/mgmt/glusterd/src/glusterd-geo-rep.c b/xlators/mgmt/glusterd/src/glusterd-geo-rep.c
index 0f40bea..85c06c1 100644
--- a/xlators/mgmt/glusterd/src/glusterd-geo-rep.c
+++ b/xlators/mgmt/glusterd/src/glusterd-geo-rep.c
@@ -5981,7 +5981,7 @@ glusterd_get_slave_info(char *slave, char **slave_url, char **hostname,
GF_ASSERT(this);
ret = glusterd_urltransform_single(slave, "normalize", &linearr);
- if (ret == -1) {
+ if ((ret == -1) || (linearr[0] == NULL)) {
ret = snprintf(errmsg, sizeof(errmsg) - 1, "Invalid Url: %s", slave);
errmsg[ret] = '\0';
*op_errstr = gf_strdup(errmsg);
@@ -5992,7 +5992,10 @@ glusterd_get_slave_info(char *slave, char **slave_url, char **hostname,
tmp = strtok_r(linearr[0], "/", &save_ptr);
tmp = strtok_r(NULL, "/", &save_ptr);
- slave = strtok_r(tmp, ":", &save_ptr);
+ slave = NULL;
+ if (tmp != NULL) {
+ slave = strtok_r(tmp, ":", &save_ptr);
+ }
if (slave) {
ret = glusterd_geo_rep_parse_slave(slave, hostname, op_errstr);
if (ret) {
diff --git a/xlators/nfs/server/src/mount3.c b/xlators/nfs/server/src/mount3.c
index 396809c..734453c 100644
--- a/xlators/nfs/server/src/mount3.c
+++ b/xlators/nfs/server/src/mount3.c
@@ -3205,6 +3205,12 @@ mnt3_export_parse_auth_param(struct mnt3_export *exp, char *exportpath)
struct host_auth_spec *host = NULL;
int ret = 0;
+ if (exportpath == NULL) {
+ gf_msg(GF_MNT, GF_LOG_ERROR, EINVAL, NFS_MSG_PARSE_HOSTSPEC_FAIL,
+ "Export path is NULL");
+ return -1;
+ }
+
/* Using exportpath directly in strtok_r because we want
* to strip off AUTH parameter from exportpath. */
token = strtok_r(exportpath, "(", &savPtr);
diff --git a/xlators/protocol/client/src/client.c b/xlators/protocol/client/src/client.c
index e156d4d..ed855ca 100644
--- a/xlators/protocol/client/src/client.c
+++ b/xlators/protocol/client/src/client.c
@@ -1222,9 +1222,12 @@ client_set_remote_options(char *value, xlator_t *this)
char *remote_port_str = NULL;
char *tmp = NULL;
int remote_port = 0;
- int ret = 0;
+ int ret = -1;
dup_value = gf_strdup(value);
+ if (dup_value == NULL) {
+ goto out;
+ }
host = strtok_r(dup_value, ":", &tmp);
subvol = strtok_r(NULL, ":", &tmp);
remote_port_str = strtok_r(NULL, ":", &tmp);
@@ -1238,6 +1241,7 @@ client_set_remote_options(char *value, xlator_t *this)
if (ret) {
gf_msg(this->name, GF_LOG_WARNING, 0, PC_MSG_DICT_SET_FAILED,
"failed to set remote-host with %s", host);
+ GF_FREE(host_dup);
goto out;
}
}
@@ -1252,6 +1256,7 @@ client_set_remote_options(char *value, xlator_t *this)
if (ret) {
gf_msg(this->name, GF_LOG_WARNING, 0, PC_MSG_DICT_SET_FAILED,
"failed to set remote-host with %s", host);
+ GF_FREE(subvol_dup);
goto out;
}
}
diff --git a/xlators/storage/posix/src/posix-helpers.c b/xlators/storage/posix/src/posix-helpers.c
index 949c799..2336add 100644
--- a/xlators/storage/posix/src/posix-helpers.c
+++ b/xlators/storage/posix/src/posix-helpers.c
@@ -390,6 +390,9 @@ _posix_get_marker_quota_contributions(posix_xattr_filler_t *filler, char *key)
int i = 0, ret = 0;
tmp_key = ptr = gf_strdup(key);
+ if (tmp_key == NULL) {
+ return -1;
+ }
for (i = 0; i < 4; i++) {
token = strtok_r(tmp_key, ".", &saveptr);
tmp_key = NULL;
--
1.8.3.1

638
SOURCES/0431-glusterd-check-for-same-node-while-adding-bricks-in-.patch Normal file
View File

@ -0,0 +1,638 @@
From d7c52ddd2cbadb1d9a55767c2f7fe6ba38d9a2ed Mon Sep 17 00:00:00 2001
From: Sheetal Pamecha <spamecha@redhat.com>
Date: Wed, 20 Nov 2019 12:42:12 +0530
Subject: [PATCH 431/449] glusterd: check for same node while adding bricks in
disperse volume
The optimal way for configuring disperse and replicate volumes
is to have all bricks in different nodes.
During create operation it fails saying it is not optimal, user
must use force to over-ride this behavior. Implementing same
during add-brick operation to avoid situation where all the added
bricks end up from same host. Operation will error out accordingly.
and this can be over-ridden by using force same as create.
> Upstream Patch Link: https://review.gluster.org/#/c/glusterfs/+/23729
> fixes: #1047
> Change-Id: I3ee9c97c1a14b73f4532893bc00187ef9355238b
> Signed-off-by: Sheetal Pamecha <spamecha@redhat.com>
BUG: 1524457
Change-Id: I3ee9c97c1a14b73f4532893bc00187ef9355238b
Signed-off-by: Sheetal Pamecha <spamecha@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202621
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/mgmt/glusterd/src/glusterd-brick-ops.c | 20 +-
xlators/mgmt/glusterd/src/glusterd-utils.c | 224 ++++++++++++++++++
xlators/mgmt/glusterd/src/glusterd-utils.h | 4 +
xlators/mgmt/glusterd/src/glusterd-volume-ops.c | 293 +++---------------------
4 files changed, 276 insertions(+), 265 deletions(-)
diff --git a/xlators/mgmt/glusterd/src/glusterd-brick-ops.c b/xlators/mgmt/glusterd/src/glusterd-brick-ops.c
index c5141de..d424f31 100644
--- a/xlators/mgmt/glusterd/src/glusterd-brick-ops.c
+++ b/xlators/mgmt/glusterd/src/glusterd-brick-ops.c
@@ -21,7 +21,6 @@
#include "glusterd-messages.h"
#include "glusterd-server-quorum.h"
#include <glusterfs/run.h>
-#include "glusterd-volgen.h"
#include <glusterfs/syscall.h>
#include <sys/signal.h>
@@ -1575,6 +1574,25 @@ glusterd_op_stage_add_brick(dict_t *dict, char **op_errstr, dict_t *rsp_dict)
is_force = dict_get_str_boolean(dict, "force", _gf_false);
+ /* Check brick order if the volume type is replicate or disperse. If
+ * force at the end of command not given then check brick order.
+ */
+
+ if (!is_force) {
+ if ((volinfo->type == GF_CLUSTER_TYPE_REPLICATE) ||
+ (volinfo->type == GF_CLUSTER_TYPE_DISPERSE)) {
+ ret = glusterd_check_brick_order(dict, msg, volinfo->type);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_BAD_BRKORDER,
+ "Not adding brick because of "
+ "bad brick order. %s",
+ msg);
+ *op_errstr = gf_strdup(msg);
+ goto out;
+ }
+ }
+ }
+
if (volinfo->replica_count < replica_count && !is_force) {
cds_list_for_each_entry(brickinfo, &volinfo->bricks, brick_list)
{
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index a1299bc..14e23d1 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -14759,3 +14759,227 @@ glusterd_is_profile_on(glusterd_volinfo_t *volinfo)
return _gf_true;
return _gf_false;
}
+
+static gf_ai_compare_t
+glusterd_compare_addrinfo(struct addrinfo *first, struct addrinfo *next)
+{
+ int ret = -1;
+ struct addrinfo *tmp1 = NULL;
+ struct addrinfo *tmp2 = NULL;
+ char firstip[NI_MAXHOST] = {0.};
+ char nextip[NI_MAXHOST] = {
+ 0,
+ };
+
+ for (tmp1 = first; tmp1 != NULL; tmp1 = tmp1->ai_next) {
+ ret = getnameinfo(tmp1->ai_addr, tmp1->ai_addrlen, firstip, NI_MAXHOST,
+ NULL, 0, NI_NUMERICHOST);
+ if (ret)
+ return GF_AI_COMPARE_ERROR;
+ for (tmp2 = next; tmp2 != NULL; tmp2 = tmp2->ai_next) {
+ ret = getnameinfo(tmp2->ai_addr, tmp2->ai_addrlen, nextip,
+ NI_MAXHOST, NULL, 0, NI_NUMERICHOST);
+ if (ret)
+ return GF_AI_COMPARE_ERROR;
+ if (!strcmp(firstip, nextip)) {
+ return GF_AI_COMPARE_MATCH;
+ }
+ }
+ }
+ return GF_AI_COMPARE_NO_MATCH;
+}
+
+/* Check for non optimal brick order for Replicate/Disperse :
+ * Checks if bricks belonging to a replicate or disperse
+ * volume are present on the same server
+ */
+int32_t
+glusterd_check_brick_order(dict_t *dict, char *err_str, int32_t type)
+{
+ int ret = -1;
+ int i = 0;
+ int j = 0;
+ int k = 0;
+ xlator_t *this = NULL;
+ addrinfo_list_t *ai_list = NULL;
+ addrinfo_list_t *ai_list_tmp1 = NULL;
+ addrinfo_list_t *ai_list_tmp2 = NULL;
+ char *brick = NULL;
+ char *brick_list = NULL;
+ char *brick_list_dup = NULL;
+ char *brick_list_ptr = NULL;
+ char *tmpptr = NULL;
+ char *volname = NULL;
+ int32_t brick_count = 0;
+ int32_t sub_count = 0;
+ struct addrinfo *ai_info = NULL;
+ char brick_addr[128] = {
+ 0,
+ };
+ int addrlen = 0;
+
+ const char failed_string[2048] =
+ "Failed to perform brick order "
+ "check. Use 'force' at the end of the command"
+ " if you want to override this behavior. ";
+ const char found_string[2048] =
+ "Multiple bricks of a %s "
+ "volume are present on the same server. This "
+ "setup is not optimal. Bricks should be on "
+ "different nodes to have best fault tolerant "
+ "configuration. Use 'force' at the end of the "
+ "command if you want to override this "
+ "behavior. ";
+
+ this = THIS;
+
+ GF_ASSERT(this);
+
+ ai_list = MALLOC(sizeof(addrinfo_list_t));
+ ai_list->info = NULL;
+ CDS_INIT_LIST_HEAD(&ai_list->list);
+
+ ret = dict_get_strn(dict, "volname", SLEN("volname"), &volname);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_GET_FAILED,
+ "Unable to get volume name");
+ goto out;
+ }
+
+ ret = dict_get_strn(dict, "bricks", SLEN("bricks"), &brick_list);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_GET_FAILED,
+ "Bricks check : Could not "
+ "retrieve bricks list");
+ goto out;
+ }
+
+ ret = dict_get_int32n(dict, "count", SLEN("count"), &brick_count);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_GET_FAILED,
+ "Bricks check : Could not "
+ "retrieve brick count");
+ goto out;
+ }
+
+ if (type != GF_CLUSTER_TYPE_DISPERSE) {
+ ret = dict_get_int32n(dict, "replica-count", SLEN("replica-count"),
+ &sub_count);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_GET_FAILED,
+ "Bricks check : Could"
+ " not retrieve replica count");
+ goto out;
+ }
+ gf_msg_debug(this->name, 0,
+ "Replicate cluster type "
+ "found. Checking brick order.");
+ } else {
+ ret = dict_get_int32n(dict, "disperse-count", SLEN("disperse-count"),
+ &sub_count);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_GET_FAILED,
+ "Bricks check : Could"
+ " not retrieve disperse count");
+ goto out;
+ }
+ gf_msg(this->name, GF_LOG_INFO, 0, GD_MSG_DISPERSE_CLUSTER_FOUND,
+ "Disperse cluster type"
+ " found. Checking brick order.");
+ }
+ brick_list_dup = brick_list_ptr = gf_strdup(brick_list);
+ /* Resolve hostnames and get addrinfo */
+ while (i < brick_count) {
+ ++i;
+ brick = strtok_r(brick_list_dup, " \n", &tmpptr);
+ brick_list_dup = tmpptr;
+ if (brick == NULL)
+ goto check_failed;
+ tmpptr = strrchr(brick, ':');
+ if (tmpptr == NULL)
+ goto check_failed;
+ addrlen = strlen(brick) - strlen(tmpptr);
+ strncpy(brick_addr, brick, addrlen);
+ brick_addr[addrlen] = '\0';
+ ret = getaddrinfo(brick_addr, NULL, NULL, &ai_info);
+ if (ret != 0) {
+ ret = 0;
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_HOSTNAME_RESOLVE_FAIL,
+ "unable to resolve host name for addr %s", brick_addr);
+ goto out;
+ }
+ ai_list_tmp1 = MALLOC(sizeof(addrinfo_list_t));
+ if (ai_list_tmp1 == NULL) {
+ ret = 0;
+ gf_msg(this->name, GF_LOG_ERROR, ENOMEM, GD_MSG_NO_MEMORY,
+ "failed to allocate "
+ "memory");
+ freeaddrinfo(ai_info);
+ goto out;
+ }
+ ai_list_tmp1->info = ai_info;
+ cds_list_add_tail(&ai_list_tmp1->list, &ai_list->list);
+ ai_list_tmp1 = NULL;
+ }
+
+ i = 0;
+ ai_list_tmp1 = cds_list_entry(ai_list->list.next, addrinfo_list_t, list);
+
+ /* Check for bad brick order */
+ while (i < brick_count) {
+ ++i;
+ ai_info = ai_list_tmp1->info;
+ ai_list_tmp1 = cds_list_entry(ai_list_tmp1->list.next, addrinfo_list_t,
+ list);
+ if (0 == i % sub_count) {
+ j = 0;
+ continue;
+ }
+ ai_list_tmp2 = ai_list_tmp1;
+ k = j;
+ while (k < sub_count - 1) {
+ ++k;
+ ret = glusterd_compare_addrinfo(ai_info, ai_list_tmp2->info);
+ if (GF_AI_COMPARE_ERROR == ret)
+ goto check_failed;
+ if (GF_AI_COMPARE_MATCH == ret)
+ goto found_bad_brick_order;
+ ai_list_tmp2 = cds_list_entry(ai_list_tmp2->list.next,
+ addrinfo_list_t, list);
+ }
+ ++j;
+ }
+ gf_msg_debug(this->name, 0, "Brick order okay");
+ ret = 0;
+ goto out;
+
+check_failed:
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_BAD_BRKORDER_CHECK_FAIL,
+ "Failed bad brick order check");
+ snprintf(err_str, sizeof(failed_string), failed_string);
+ ret = -1;
+ goto out;
+
+found_bad_brick_order:
+ gf_msg(this->name, GF_LOG_INFO, 0, GD_MSG_BAD_BRKORDER,
+ "Bad brick order found");
+ if (type == GF_CLUSTER_TYPE_DISPERSE) {
+ snprintf(err_str, sizeof(found_string), found_string, "disperse");
+ } else {
+ snprintf(err_str, sizeof(found_string), found_string, "replicate");
+ }
+
+ ret = -1;
+out:
+ ai_list_tmp2 = NULL;
+ GF_FREE(brick_list_ptr);
+ cds_list_for_each_entry(ai_list_tmp1, &ai_list->list, list)
+ {
+ if (ai_list_tmp1->info)
+ freeaddrinfo(ai_list_tmp1->info);
+ free(ai_list_tmp2);
+ ai_list_tmp2 = ai_list_tmp1;
+ }
+ free(ai_list_tmp2);
+ return ret;
+}
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.h b/xlators/mgmt/glusterd/src/glusterd-utils.h
index ead16b2..e2e2454 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.h
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.h
@@ -881,4 +881,8 @@ glusterd_is_profile_on(glusterd_volinfo_t *volinfo);
char *
search_brick_path_from_proc(pid_t brick_pid, char *brickpath);
+
+int32_t
+glusterd_check_brick_order(dict_t *dict, char *err_str, int32_t type);
+
#endif
diff --git a/xlators/mgmt/glusterd/src/glusterd-volume-ops.c b/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
index 93042ab..8da2ff3 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
@@ -41,240 +41,6 @@
#define glusterd_op_start_volume_args_get(dict, volname, flags) \
glusterd_op_stop_volume_args_get(dict, volname, flags)
-gf_ai_compare_t
-glusterd_compare_addrinfo(struct addrinfo *first, struct addrinfo *next)
-{
- int ret = -1;
- struct addrinfo *tmp1 = NULL;
- struct addrinfo *tmp2 = NULL;
- char firstip[NI_MAXHOST] = {0.};
- char nextip[NI_MAXHOST] = {
- 0,
- };
-
- for (tmp1 = first; tmp1 != NULL; tmp1 = tmp1->ai_next) {
- ret = getnameinfo(tmp1->ai_addr, tmp1->ai_addrlen, firstip, NI_MAXHOST,
- NULL, 0, NI_NUMERICHOST);
- if (ret)
- return GF_AI_COMPARE_ERROR;
- for (tmp2 = next; tmp2 != NULL; tmp2 = tmp2->ai_next) {
- ret = getnameinfo(tmp2->ai_addr, tmp2->ai_addrlen, nextip,
- NI_MAXHOST, NULL, 0, NI_NUMERICHOST);
- if (ret)
- return GF_AI_COMPARE_ERROR;
- if (!strcmp(firstip, nextip)) {
- return GF_AI_COMPARE_MATCH;
- }
- }
- }
- return GF_AI_COMPARE_NO_MATCH;
-}
-
-/* Check for non optimal brick order for replicate :
- * Checks if bricks belonging to a replicate volume
- * are present on the same server
- */
-int32_t
-glusterd_check_brick_order(dict_t *dict, char *err_str)
-{
- int ret = -1;
- int i = 0;
- int j = 0;
- int k = 0;
- xlator_t *this = NULL;
- addrinfo_list_t *ai_list = NULL;
- addrinfo_list_t *ai_list_tmp1 = NULL;
- addrinfo_list_t *ai_list_tmp2 = NULL;
- char *brick = NULL;
- char *brick_list = NULL;
- char *brick_list_dup = NULL;
- char *brick_list_ptr = NULL;
- char *tmpptr = NULL;
- char *volname = NULL;
- int32_t brick_count = 0;
- int32_t type = GF_CLUSTER_TYPE_NONE;
- int32_t sub_count = 0;
- struct addrinfo *ai_info = NULL;
- char brick_addr[128] = {
- 0,
- };
- int addrlen = 0;
-
- const char failed_string[2048] =
- "Failed to perform brick order "
- "check. Use 'force' at the end of the command"
- " if you want to override this behavior. ";
- const char found_string[2048] =
- "Multiple bricks of a %s "
- "volume are present on the same server. This "
- "setup is not optimal. Bricks should be on "
- "different nodes to have best fault tolerant "
- "configuration. Use 'force' at the end of the "
- "command if you want to override this "
- "behavior. ";
-
- this = THIS;
-
- GF_ASSERT(this);
-
- ai_list = MALLOC(sizeof(addrinfo_list_t));
- ai_list->info = NULL;
- CDS_INIT_LIST_HEAD(&ai_list->list);
-
- ret = dict_get_strn(dict, "volname", SLEN("volname"), &volname);
- if (ret) {
- gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_GET_FAILED,
- "Unable to get volume name");
- goto out;
- }
-
- ret = dict_get_int32n(dict, "type", SLEN("type"), &type);
- if (ret) {
- snprintf(err_str, 512, "Unable to get type of volume %s", volname);
- gf_msg(this->name, GF_LOG_WARNING, 0, GD_MSG_DICT_GET_FAILED, "%s",
- err_str);
- goto out;
- }
-
- ret = dict_get_strn(dict, "bricks", SLEN("bricks"), &brick_list);
- if (ret) {
- gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_GET_FAILED,
- "Bricks check : Could not "
- "retrieve bricks list");
- goto out;
- }
-
- ret = dict_get_int32n(dict, "count", SLEN("count"), &brick_count);
- if (ret) {
- gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_GET_FAILED,
- "Bricks check : Could not "
- "retrieve brick count");
- goto out;
- }
-
- if (type != GF_CLUSTER_TYPE_DISPERSE) {
- ret = dict_get_int32n(dict, "replica-count", SLEN("replica-count"),
- &sub_count);
- if (ret) {
- gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_GET_FAILED,
- "Bricks check : Could"
- " not retrieve replica count");
- goto out;
- }
- gf_msg_debug(this->name, 0,
- "Replicate cluster type "
- "found. Checking brick order.");
- } else {
- ret = dict_get_int32n(dict, "disperse-count", SLEN("disperse-count"),
- &sub_count);
- if (ret) {
- gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_GET_FAILED,
- "Bricks check : Could"
- " not retrieve disperse count");
- goto out;
- }
- gf_msg(this->name, GF_LOG_INFO, 0, GD_MSG_DISPERSE_CLUSTER_FOUND,
- "Disperse cluster type"
- " found. Checking brick order.");
- }
-
- brick_list_dup = brick_list_ptr = gf_strdup(brick_list);
- /* Resolve hostnames and get addrinfo */
- while (i < brick_count) {
- ++i;
- brick = strtok_r(brick_list_dup, " \n", &tmpptr);
- brick_list_dup = tmpptr;
- if (brick == NULL)
- goto check_failed;
- tmpptr = strrchr(brick, ':');
- if (tmpptr == NULL)
- goto check_failed;
- addrlen = strlen(brick) - strlen(tmpptr);
- strncpy(brick_addr, brick, addrlen);
- brick_addr[addrlen] = '\0';
- ret = getaddrinfo(brick_addr, NULL, NULL, &ai_info);
- if (ret != 0) {
- ret = 0;
- gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_HOSTNAME_RESOLVE_FAIL,
- "unable to resolve host name for addr %s", brick_addr);
- goto out;
- }
- ai_list_tmp1 = MALLOC(sizeof(addrinfo_list_t));
- if (ai_list_tmp1 == NULL) {
- ret = 0;
- gf_msg(this->name, GF_LOG_ERROR, ENOMEM, GD_MSG_NO_MEMORY,
- "failed to allocate "
- "memory");
- freeaddrinfo(ai_info);
- goto out;
- }
- ai_list_tmp1->info = ai_info;
- cds_list_add_tail(&ai_list_tmp1->list, &ai_list->list);
- ai_list_tmp1 = NULL;
- }
-
- i = 0;
- ai_list_tmp1 = cds_list_entry(ai_list->list.next, addrinfo_list_t, list);
-
- /* Check for bad brick order */
- while (i < brick_count) {
- ++i;
- ai_info = ai_list_tmp1->info;
- ai_list_tmp1 = cds_list_entry(ai_list_tmp1->list.next, addrinfo_list_t,
- list);
- if (0 == i % sub_count) {
- j = 0;
- continue;
- }
- ai_list_tmp2 = ai_list_tmp1;
- k = j;
- while (k < sub_count - 1) {
- ++k;
- ret = glusterd_compare_addrinfo(ai_info, ai_list_tmp2->info);
- if (GF_AI_COMPARE_ERROR == ret)
- goto check_failed;
- if (GF_AI_COMPARE_MATCH == ret)
- goto found_bad_brick_order;
- ai_list_tmp2 = cds_list_entry(ai_list_tmp2->list.next,
- addrinfo_list_t, list);
- }
- ++j;
- }
- gf_msg_debug(this->name, 0, "Brick order okay");
- ret = 0;
- goto out;
-
-check_failed:
- gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_BAD_BRKORDER_CHECK_FAIL,
- "Failed bad brick order check");
- snprintf(err_str, sizeof(failed_string), failed_string);
- ret = -1;
- goto out;
-
-found_bad_brick_order:
- gf_msg(this->name, GF_LOG_INFO, 0, GD_MSG_BAD_BRKORDER,
- "Bad brick order found");
- if (type == GF_CLUSTER_TYPE_DISPERSE) {
- snprintf(err_str, sizeof(found_string), found_string, "disperse");
- } else {
- snprintf(err_str, sizeof(found_string), found_string, "replicate");
- }
-
- ret = -1;
-out:
- ai_list_tmp2 = NULL;
- GF_FREE(brick_list_ptr);
- cds_list_for_each_entry(ai_list_tmp1, &ai_list->list, list)
- {
- if (ai_list_tmp1->info)
- freeaddrinfo(ai_list_tmp1->info);
- free(ai_list_tmp2);
- ai_list_tmp2 = ai_list_tmp1;
- }
- free(ai_list_tmp2);
- return ret;
-}
-
int
__glusterd_handle_create_volume(rpcsvc_request_t *req)
{
@@ -1337,6 +1103,35 @@ glusterd_op_stage_create_volume(dict_t *dict, char **op_errstr,
}
}
+ /*Check brick order if the volume type is replicate or disperse. If
+ * force at the end of command not given then check brick order.
+ */
+ if (is_origin_glusterd(dict)) {
+ ret = dict_get_int32n(dict, "type", SLEN("type"), &type);
+ if (ret) {
+ snprintf(msg, sizeof(msg),
+ "Unable to get type of "
+ "volume %s",
+ volname);
+ gf_msg(this->name, GF_LOG_WARNING, 0, GD_MSG_DICT_GET_FAILED, "%s",
+ msg);
+ goto out;
+ }
+
+ if (!is_force) {
+ if ((type == GF_CLUSTER_TYPE_REPLICATE) ||
+ (type == GF_CLUSTER_TYPE_DISPERSE)) {
+ ret = glusterd_check_brick_order(dict, msg, type);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_BAD_BRKORDER,
+ "Not creating volume because of "
+ "bad brick order");
+ goto out;
+ }
+ }
+ }
+ }
+
while (i < brick_count) {
i++;
brick = strtok_r(brick_list, " \n", &tmpptr);
@@ -1423,36 +1218,6 @@ glusterd_op_stage_create_volume(dict_t *dict, char **op_errstr,
brick_info = NULL;
}
- /*Check brick order if the volume type is replicate or disperse. If
- * force at the end of command not given then check brick order.
- */
- if (is_origin_glusterd(dict)) {
- ret = dict_get_int32n(dict, "type", SLEN("type"), &type);
- if (ret) {
- snprintf(msg, sizeof(msg),
- "Unable to get type of "
- "volume %s",
- volname);
- gf_msg(this->name, GF_LOG_WARNING, 0, GD_MSG_DICT_GET_FAILED, "%s",
- msg);
- goto out;
- }
-
- if (!is_force) {
- if ((type == GF_CLUSTER_TYPE_REPLICATE) ||
- (type == GF_CLUSTER_TYPE_DISPERSE)) {
- ret = glusterd_check_brick_order(dict, msg);
- if (ret) {
- gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_BAD_BRKORDER,
- "Not "
- "creating volume because of "
- "bad brick order");
- goto out;
- }
- }
- }
- }
-
ret = dict_set_int32n(rsp_dict, "brick_count", SLEN("brick_count"),
local_brick_count);
if (ret) {
--
1.8.3.1

503
SOURCES/0432-glusterd-Fix-coverity-defects-put-coverity-annotatio.patch Normal file
View File

@ -0,0 +1,503 @@
From aa215163cb7d806dc98bef2386a4e282a5e54a31 Mon Sep 17 00:00:00 2001
From: Atin Mukherjee <amukherj@redhat.com>
Date: Thu, 25 Apr 2019 12:00:52 +0530
Subject: [PATCH 432/449] glusterd: Fix coverity defects & put coverity
annotations
Along with fixing few defect, put the required annotations for the defects which
are marked ignore/false positive/intentional as per the coverity defect sheet.
This should avoid the per component graph showing many defects as open in the
coverity glusterfs web page.
> upstream patch link: https://review.gluster.org/#/c/glusterfs/+/22619/
> Updates: bz#789278
> Change-Id: I19461dc3603a3bd8f88866a1ab3db43d783af8e4
> Signed-off-by: Atin Mukherjee <amukherj@redhat.com>
BUG: 1787310
Change-Id: I19461dc3603a3bd8f88866a1ab3db43d783af8e4
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202631
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/mgmt/glusterd/src/glusterd-brick-ops.c | 7 +++--
xlators/mgmt/glusterd/src/glusterd-geo-rep.c | 3 +-
.../glusterd/src/glusterd-gfproxyd-svc-helper.c | 2 +-
xlators/mgmt/glusterd/src/glusterd-handler.c | 8 ++++-
xlators/mgmt/glusterd/src/glusterd-mountbroker.c | 5 ++-
xlators/mgmt/glusterd/src/glusterd-op-sm.c | 8 +++++
xlators/mgmt/glusterd/src/glusterd-peer-utils.c | 2 ++
xlators/mgmt/glusterd/src/glusterd-server-quorum.c | 1 +
xlators/mgmt/glusterd/src/glusterd-store.c | 4 ---
xlators/mgmt/glusterd/src/glusterd-svc-helper.c | 4 +--
xlators/mgmt/glusterd/src/glusterd-syncop.c | 1 +
.../mgmt/glusterd/src/glusterd-tierd-svc-helper.c | 4 +--
xlators/mgmt/glusterd/src/glusterd-utils.c | 9 ++++--
xlators/mgmt/glusterd/src/glusterd-volgen.c | 36 +++++++++++++---------
14 files changed, 63 insertions(+), 31 deletions(-)
diff --git a/xlators/mgmt/glusterd/src/glusterd-brick-ops.c b/xlators/mgmt/glusterd/src/glusterd-brick-ops.c
index d424f31..121346c 100644
--- a/xlators/mgmt/glusterd/src/glusterd-brick-ops.c
+++ b/xlators/mgmt/glusterd/src/glusterd-brick-ops.c
@@ -2032,7 +2032,6 @@ glusterd_op_stage_remove_brick(dict_t *dict, char **op_errstr)
case GF_OP_CMD_STATUS:
ret = 0;
goto out;
-
case GF_OP_CMD_DETACH_START:
if (volinfo->type != GF_CLUSTER_TYPE_TIER) {
snprintf(msg, sizeof(msg),
@@ -2044,7 +2043,7 @@ glusterd_op_stage_remove_brick(dict_t *dict, char **op_errstr)
errstr);
goto out;
}
-
+ /* Fall through */
case GF_OP_CMD_START: {
if ((volinfo->type == GF_CLUSTER_TYPE_REPLICATE) &&
dict_getn(dict, "replica-count", SLEN("replica-count"))) {
@@ -2259,7 +2258,8 @@ out:
if (op_errstr)
*op_errstr = errstr;
}
-
+ if (!op_errstr && errstr)
+ GF_FREE(errstr);
return ret;
}
@@ -2687,6 +2687,7 @@ glusterd_op_remove_brick(dict_t *dict, char **op_errstr)
* Update defrag_cmd as well or it will only be done
* for nodes on which the brick to be removed exists.
*/
+ /* coverity[MIXED_ENUMS] */
volinfo->rebal.defrag_cmd = cmd;
volinfo->rebal.defrag_status = GF_DEFRAG_STATUS_NOT_STARTED;
ret = dict_get_strn(dict, GF_REMOVE_BRICK_TID_KEY,
diff --git a/xlators/mgmt/glusterd/src/glusterd-geo-rep.c b/xlators/mgmt/glusterd/src/glusterd-geo-rep.c
index 85c06c1..5a91df4 100644
--- a/xlators/mgmt/glusterd/src/glusterd-geo-rep.c
+++ b/xlators/mgmt/glusterd/src/glusterd-geo-rep.c
@@ -4107,6 +4107,7 @@ gd_pause_or_resume_gsync(dict_t *dict, char *master, char *slave,
out:
sys_close(pfd);
+ /* coverity[INTEGER_OVERFLOW] */
return ret;
}
@@ -4183,7 +4184,7 @@ stop_gsync(char *master, char *slave, char **msg, char *conf_path,
out:
sys_close(pfd);
-
+ /* coverity[INTEGER_OVERFLOW] */
return ret;
}
diff --git a/xlators/mgmt/glusterd/src/glusterd-gfproxyd-svc-helper.c b/xlators/mgmt/glusterd/src/glusterd-gfproxyd-svc-helper.c
index 67e3f41..e338bf4 100644
--- a/xlators/mgmt/glusterd/src/glusterd-gfproxyd-svc-helper.c
+++ b/xlators/mgmt/glusterd/src/glusterd-gfproxyd-svc-helper.c
@@ -111,7 +111,7 @@ glusterd_svc_get_gfproxyd_volfile(glusterd_volinfo_t *volinfo, char *svc_name,
goto out;
}
- /* coverity[secure_temp] mkstemp uses 0600 as the mode and is safe */
+ /* coverity[SECURE_TEMP] mkstemp uses 0600 as the mode and is safe */
tmp_fd = mkstemp(*tmpvol);
if (tmp_fd < 0) {
gf_msg("glusterd", GF_LOG_WARNING, errno, GD_MSG_FILE_OP_FAILED,
diff --git a/xlators/mgmt/glusterd/src/glusterd-handler.c b/xlators/mgmt/glusterd/src/glusterd-handler.c
index 2e73c98..1f31e72 100644
--- a/xlators/mgmt/glusterd/src/glusterd-handler.c
+++ b/xlators/mgmt/glusterd/src/glusterd-handler.c
@@ -930,6 +930,7 @@ __glusterd_handle_cluster_lock(rpcsvc_request_t *req)
op_ctx = dict_new();
if (!op_ctx) {
+ ret = -1;
gf_msg(this->name, GF_LOG_ERROR, ENOMEM, GD_MSG_DICT_CREATE_FAIL,
"Unable to set new dict");
goto out;
@@ -956,6 +957,9 @@ out:
glusterd_friend_sm();
glusterd_op_sm();
+ if (ret)
+ GF_FREE(ctx);
+
return ret;
}
@@ -3470,6 +3474,7 @@ glusterd_rpc_create(struct rpc_clnt **rpc, dict_t *options,
GF_ASSERT(this);
GF_ASSERT(options);
+ GF_VALIDATE_OR_GOTO(this->name, rpc, out);
if (force && rpc && *rpc) {
(void)rpc_clnt_unref(*rpc);
@@ -3482,7 +3487,6 @@ glusterd_rpc_create(struct rpc_clnt **rpc, dict_t *options,
goto out;
ret = rpc_clnt_register_notify(new_rpc, notify_fn, notify_data);
- *rpc = new_rpc;
if (ret)
goto out;
ret = rpc_clnt_start(new_rpc);
@@ -3491,6 +3495,8 @@ out:
if (new_rpc) {
(void)rpc_clnt_unref(new_rpc);
}
+ } else {
+ *rpc = new_rpc;
}
gf_msg_debug(this->name, 0, "returning %d", ret);
diff --git a/xlators/mgmt/glusterd/src/glusterd-mountbroker.c b/xlators/mgmt/glusterd/src/glusterd-mountbroker.c
index 332ddef..c017ccb 100644
--- a/xlators/mgmt/glusterd/src/glusterd-mountbroker.c
+++ b/xlators/mgmt/glusterd/src/glusterd-mountbroker.c
@@ -334,7 +334,10 @@ make_ghadoop_mountspec(gf_mount_spec_t *mspec, const char *volname, char *user,
if (ret == -1)
return ret;
- return parse_mount_pattern_desc(mspec, hadoop_mnt_desc);
+ ret = parse_mount_pattern_desc(mspec, hadoop_mnt_desc);
+ GF_FREE(hadoop_mnt_desc);
+
+ return ret;
}
static gf_boolean_t
diff --git a/xlators/mgmt/glusterd/src/glusterd-op-sm.c b/xlators/mgmt/glusterd/src/glusterd-op-sm.c
index 6475611..46fc607 100644
--- a/xlators/mgmt/glusterd/src/glusterd-op-sm.c
+++ b/xlators/mgmt/glusterd/src/glusterd-op-sm.c
@@ -2467,6 +2467,7 @@ glusterd_start_bricks(glusterd_volinfo_t *volinfo)
if (!brickinfo->start_triggered) {
pthread_mutex_lock(&brickinfo->restart_mutex);
{
+ /* coverity[SLEEP] */
ret = glusterd_brick_start(volinfo, brickinfo, _gf_false,
_gf_false);
}
@@ -3466,6 +3467,7 @@ _add_task_to_dict(dict_t *dict, glusterd_volinfo_t *volinfo, int op, int index)
switch (op) {
case GD_OP_REMOVE_TIER_BRICK:
+ /* Fall through */
case GD_OP_REMOVE_BRICK:
snprintf(key, sizeof(key), "task%d", index);
ret = _add_remove_bricks_to_dict(dict, volinfo, key);
@@ -7550,6 +7552,7 @@ glusterd_op_ac_send_brick_op(glusterd_op_sm_event_t *event, void *ctx)
glusterd_op_t op = GD_OP_NONE;
glusterd_req_ctx_t *req_ctx = NULL;
char *op_errstr = NULL;
+ gf_boolean_t free_req_ctx = _gf_false;
this = THIS;
priv = this->private;
@@ -7558,6 +7561,9 @@ glusterd_op_ac_send_brick_op(glusterd_op_sm_event_t *event, void *ctx)
req_ctx = ctx;
} else {
req_ctx = GF_CALLOC(1, sizeof(*req_ctx), gf_gld_mt_op_allack_ctx_t);
+ if (!req_ctx)
+ goto out;
+ free_req_ctx = _gf_true;
op = glusterd_op_get_op();
req_ctx->op = op;
gf_uuid_copy(req_ctx->uuid, MY_UUID);
@@ -7588,6 +7594,8 @@ glusterd_op_ac_send_brick_op(glusterd_op_sm_event_t *event, void *ctx)
}
out:
+ if (ret && req_ctx && free_req_ctx)
+ GF_FREE(req_ctx);
gf_msg_debug(this->name, 0, "Returning with %d", ret);
return ret;
diff --git a/xlators/mgmt/glusterd/src/glusterd-peer-utils.c b/xlators/mgmt/glusterd/src/glusterd-peer-utils.c
index 8c1feeb..1a65359 100644
--- a/xlators/mgmt/glusterd/src/glusterd-peer-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-peer-utils.c
@@ -82,6 +82,7 @@ glusterd_peerinfo_cleanup(glusterd_peerinfo_t *peerinfo)
call_rcu(&peerinfo->rcu_head.head, glusterd_peerinfo_destroy);
if (quorum_action)
+ /* coverity[SLEEP] */
glusterd_do_quorum_action();
return 0;
}
@@ -358,6 +359,7 @@ glusterd_uuid_to_hostname(uuid_t uuid)
if (!gf_uuid_compare(MY_UUID, uuid)) {
hostname = gf_strdup("localhost");
+ return hostname;
}
RCU_READ_LOCK;
if (!cds_list_empty(&priv->peers)) {
diff --git a/xlators/mgmt/glusterd/src/glusterd-server-quorum.c b/xlators/mgmt/glusterd/src/glusterd-server-quorum.c
index fd334e6..f378187 100644
--- a/xlators/mgmt/glusterd/src/glusterd-server-quorum.c
+++ b/xlators/mgmt/glusterd/src/glusterd-server-quorum.c
@@ -372,6 +372,7 @@ glusterd_do_volume_quorum_action(xlator_t *this, glusterd_volinfo_t *volinfo,
if (!brickinfo->start_triggered) {
pthread_mutex_lock(&brickinfo->restart_mutex);
{
+ /* coverity[SLEEP] */
ret = glusterd_brick_start(volinfo, brickinfo, _gf_false,
_gf_false);
}
diff --git a/xlators/mgmt/glusterd/src/glusterd-store.c b/xlators/mgmt/glusterd/src/glusterd-store.c
index b3b5ee9..4fa8116 100644
--- a/xlators/mgmt/glusterd/src/glusterd-store.c
+++ b/xlators/mgmt/glusterd/src/glusterd-store.c
@@ -4764,10 +4764,6 @@ glusterd_store_retrieve_peers(xlator_t *this)
*/
address = cds_list_entry(peerinfo->hostnames.next,
glusterd_peer_hostname_t, hostname_list);
- if (!address) {
- ret = -1;
- goto next;
- }
peerinfo->hostname = gf_strdup(address->hostname);
ret = glusterd_friend_add_from_peerinfo(peerinfo, 1, NULL);
diff --git a/xlators/mgmt/glusterd/src/glusterd-svc-helper.c b/xlators/mgmt/glusterd/src/glusterd-svc-helper.c
index ca19a75..1d1f42d 100644
--- a/xlators/mgmt/glusterd/src/glusterd-svc-helper.c
+++ b/xlators/mgmt/glusterd/src/glusterd-svc-helper.c
@@ -179,7 +179,7 @@ glusterd_svc_check_volfile_identical(char *svc_name,
goto out;
}
- /* coverity[secure_temp] mkstemp uses 0600 as the mode and is safe */
+ /* coverity[SECURE_TEMP] mkstemp uses 0600 as the mode and is safe */
tmp_fd = mkstemp(tmpvol);
if (tmp_fd < 0) {
gf_msg(this->name, GF_LOG_WARNING, errno, GD_MSG_FILE_OP_FAILED,
@@ -241,7 +241,7 @@ glusterd_svc_check_topology_identical(char *svc_name,
goto out;
}
- /* coverity[secure_temp] mkstemp uses 0600 as the mode and is safe */
+ /* coverity[SECURE_TEMP] mkstemp uses 0600 as the mode and is safe */
tmpfd = mkstemp(tmpvol);
if (tmpfd < 0) {
gf_msg(this->name, GF_LOG_WARNING, errno, GD_MSG_FILE_OP_FAILED,
diff --git a/xlators/mgmt/glusterd/src/glusterd-syncop.c b/xlators/mgmt/glusterd/src/glusterd-syncop.c
index 618d8bc..9e47d14 100644
--- a/xlators/mgmt/glusterd/src/glusterd-syncop.c
+++ b/xlators/mgmt/glusterd/src/glusterd-syncop.c
@@ -1752,6 +1752,7 @@ gd_brick_op_phase(glusterd_op_t op, dict_t *op_ctx, dict_t *req_dict,
if (dict_get(op_ctx, "client-count"))
break;
}
+ /* coverity[MIXED_ENUMS] */
} else if (cmd == GF_OP_CMD_DETACH_START) {
op = GD_OP_REMOVE_BRICK;
dict_del(req_dict, "rebalance-command");
diff --git a/xlators/mgmt/glusterd/src/glusterd-tierd-svc-helper.c b/xlators/mgmt/glusterd/src/glusterd-tierd-svc-helper.c
index 922eae7..59843a0 100644
--- a/xlators/mgmt/glusterd/src/glusterd-tierd-svc-helper.c
+++ b/xlators/mgmt/glusterd/src/glusterd-tierd-svc-helper.c
@@ -116,7 +116,7 @@ glusterd_svc_check_tier_volfile_identical(char *svc_name,
goto out;
}
- /* coverity[secure_temp] mkstemp uses 0600 as the mode and is safe */
+ /* coverity[SECURE_TEMP] mkstemp uses 0600 as the mode and is safe */
tmp_fd = mkstemp(tmpvol);
if (tmp_fd < 0) {
gf_msg(this->name, GF_LOG_WARNING, errno, GD_MSG_FILE_OP_FAILED,
@@ -177,7 +177,7 @@ glusterd_svc_check_tier_topology_identical(char *svc_name,
goto out;
}
- /* coverity[secure_temp] mkstemp uses 0600 as the mode and is safe */
+ /* coverity[SECURE_TEMP] mkstemp uses 0600 as the mode and is safe */
tmpfd = mkstemp(tmpvol);
if (tmpfd < 0) {
gf_msg(this->name, GF_LOG_WARNING, errno, GD_MSG_FILE_OP_FAILED,
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index 14e23d1..8b0fc9a 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -908,6 +908,7 @@ glusterd_create_sub_tier_volinfo(glusterd_volinfo_t *volinfo,
(*dup_volinfo)->brick_count = tier_info->cold_brick_count;
}
out:
+ /* coverity[REVERSE_NULL] */
if (ret && *dup_volinfo) {
glusterd_volinfo_delete(*dup_volinfo);
*dup_volinfo = NULL;
@@ -2738,6 +2739,7 @@ glusterd_readin_file(const char *filepath, int *line_count)
/* Reduce allocation to minimal size. */
p = GF_REALLOC(lines, (counter + 1) * sizeof(char *));
if (!p) {
+ /* coverity[TAINTED_SCALAR] */
free_lines(lines, counter);
lines = NULL;
goto out;
@@ -6782,6 +6784,7 @@ glusterd_restart_bricks(void *opaque)
if (!brickinfo->start_triggered) {
pthread_mutex_lock(&brickinfo->restart_mutex);
{
+ /* coverity[SLEEP] */
glusterd_brick_start(volinfo, brickinfo, _gf_false,
_gf_false);
}
@@ -8886,7 +8889,7 @@ glusterd_nfs_statedump(char *options, int option_cnt, char **op_errstr)
kill(pid, SIGUSR1);
sleep(1);
-
+ /* coverity[TAINTED_STRING] */
sys_unlink(dumpoptions_path);
ret = 0;
out:
@@ -9012,6 +9015,7 @@ glusterd_quotad_statedump(char *options, int option_cnt, char **op_errstr)
sleep(1);
+ /* coverity[TAINTED_STRING] */
sys_unlink(dumpoptions_path);
ret = 0;
out:
@@ -13423,7 +13427,7 @@ glusterd_get_global_options_for_all_vols(rpcsvc_request_t *req, dict_t *ctx,
if (key_fixed)
key = key_fixed;
}
-
+ /* coverity[CONSTANT_EXPRESSION_RESULT] */
ALL_VOLUME_OPTION_CHECK("all", _gf_true, key, ret, op_errstr, out);
for (i = 0; valid_all_vol_opts[i].option; i++) {
@@ -14153,6 +14157,7 @@ glusterd_disallow_op_for_tier(glusterd_volinfo_t *volinfo, glusterd_op_t op,
break;
case GD_OP_REMOVE_BRICK:
switch (cmd) {
+ /* coverity[MIXED_ENUMS] */
case GF_DEFRAG_CMD_DETACH_START:
case GF_OP_CMD_DETACH_COMMIT_FORCE:
case GF_OP_CMD_DETACH_COMMIT:
diff --git a/xlators/mgmt/glusterd/src/glusterd-volgen.c b/xlators/mgmt/glusterd/src/glusterd-volgen.c
index 539e8a5..6852f8e 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volgen.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volgen.c
@@ -322,7 +322,7 @@ volopt_trie_cbk(char *word, void *param)
}
static int
-process_nodevec(struct trienodevec *nodevec, char **hint)
+process_nodevec(struct trienodevec *nodevec, char **outputhint, char *inputhint)
{
int ret = 0;
char *hint1 = NULL;
@@ -331,14 +331,14 @@ process_nodevec(struct trienodevec *nodevec, char **hint)
trienode_t **nodes = nodevec->nodes;
if (!nodes[0]) {
- *hint = NULL;
+ *outputhint = NULL;
return 0;
}
#if 0
/* Limit as in git */
if (trienode_get_dist (nodes[0]) >= 6) {
- *hint = NULL;
+ *outputhint = NULL;
return 0;
}
#endif
@@ -347,23 +347,30 @@ process_nodevec(struct trienodevec *nodevec, char **hint)
return -1;
if (nodevec->cnt < 2 || !nodes[1]) {
- *hint = hint1;
+ *outputhint = hint1;
return 0;
}
- if (trienode_get_word(nodes[1], &hint2))
+ if (trienode_get_word(nodes[1], &hint2)) {
+ GF_FREE(hint1);
return -1;
+ }
- if (*hint)
- hintinfx = *hint;
- ret = gf_asprintf(hint, "%s or %s%s", hint1, hintinfx, hint2);
+ if (inputhint)
+ hintinfx = inputhint;
+ ret = gf_asprintf(outputhint, "%s or %s%s", hint1, hintinfx, hint2);
if (ret > 0)
ret = 0;
+ if (hint1)
+ GF_FREE(hint1);
+ if (hint2)
+ GF_FREE(hint2);
return ret;
}
static int
-volopt_trie_section(int lvl, char **patt, char *word, char **hint, int hints)
+volopt_trie_section(int lvl, char **patt, char *word, char **outputhint,
+ char *inputhint, int hints)
{
trienode_t *nodes[] = {NULL, NULL};
struct trienodevec nodevec = {nodes, 2};
@@ -384,7 +391,7 @@ volopt_trie_section(int lvl, char **patt, char *word, char **hint, int hints)
nodevec.cnt = hints;
ret = trie_measure_vec(trie, word, &nodevec);
if (!ret && nodevec.nodes[0])
- ret = process_nodevec(&nodevec, hint);
+ ret = process_nodevec(&nodevec, outputhint, inputhint);
trie_destroy(trie);
@@ -396,6 +403,7 @@ volopt_trie(char *key, char **hint)
{
char *patt[] = {NULL};
char *fullhint = NULL;
+ char *inputhint = NULL;
char *dot = NULL;
char *dom = NULL;
int len = 0;
@@ -405,7 +413,7 @@ volopt_trie(char *key, char **hint)
dot = strchr(key, '.');
if (!dot)
- return volopt_trie_section(1, patt, key, hint, 2);
+ return volopt_trie_section(1, patt, key, hint, inputhint, 2);
len = dot - key;
dom = gf_strdup(key);
@@ -413,7 +421,7 @@ volopt_trie(char *key, char **hint)
return -1;
dom[len] = '\0';
- ret = volopt_trie_section(0, NULL, dom, patt, 1);
+ ret = volopt_trie_section(0, NULL, dom, patt, inputhint, 1);
GF_FREE(dom);
if (ret) {
patt[0] = NULL;
@@ -422,8 +430,8 @@ volopt_trie(char *key, char **hint)
if (!patt[0])
goto out;
- *hint = "...";
- ret = volopt_trie_section(1, patt, dot + 1, hint, 2);
+ inputhint = "...";
+ ret = volopt_trie_section(1, patt, dot + 1, hint, inputhint, 2);
if (ret)
goto out;
if (*hint) {
--
1.8.3.1

8
SOURCES/0389-socket-Resolve-ssl_ctx-leak-for-a-brick-while-only-m.patch → SOURCES/0433-socket-Resolve-ssl_ctx-leak-for-a-brick-while-only-m.patch
View File

@ -1,7 +1,7 @@
From d3558cfbded7e973fae45ce2196767611336e351 Mon Sep 17 00:00:00 2001
From 0e3871a57b7a621444dc5cfd49935a1e412f6436 Mon Sep 17 00:00:00 2001
From: Mohit Agrawal <moagrawa@redhat.com>
Date: Mon, 8 Jun 2020 13:27:50 +0530
Subject: [PATCH 389/392] socket: Resolve ssl_ctx leak for a brick while only
Subject: [PATCH 433/449] socket: Resolve ssl_ctx leak for a brick while only
mgmt SSL is enabled
Problem: While only mgmt SSL is enabled for a brick process use_ssl flag
@ -17,10 +17,10 @@ Solution: To avoid a leak check only ssl_ctx, if it is valid cleanup
> (Cherry pick from commit 9873baee34afdf0c20f5fc98a7dbf2a9f07447e2)
> (Reviewed on upstream link https://review.gluster.org/#/c/glusterfs/+/24366/)
BUG: 1848894
BUG: 1810924
Change-Id: I2f4295478f4149dcb7d608ea78ee5104f28812c3
Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/203705
Reviewed-on: https://code.engineering.redhat.com/gerrit/202625
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---

39
SOURCES/0434-glusterd-ganesha-fix-Coverity-CID-1405785.patch Normal file
View File

@ -0,0 +1,39 @@
From 2ea3fc203671429d0aa9994e5bbd57f6a604523d Mon Sep 17 00:00:00 2001
From: Xie Changlong <xiechanglong@cmss.chinamobile.com>
Date: Mon, 28 Oct 2019 17:43:28 +0800
Subject: [PATCH 434/449] glusterd/ganesha: fix Coverity CID 1405785
To avoid double free
> upstream patch link: https://review.gluster.org/#/c/glusterfs/+/23630/
> updates: bz#789278
> Change-Id: I15ae54ed696295d4cb015668722e77983b062ccb
> Signed-off-by: Xie Changlong <xiechanglong@cmss.chinamobile.com>
BUG: 1787310
Change-Id: I15ae54ed696295d4cb015668722e77983b062ccb
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202623
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/mgmt/glusterd/src/glusterd-ganesha.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/xlators/mgmt/glusterd/src/glusterd-ganesha.c b/xlators/mgmt/glusterd/src/glusterd-ganesha.c
index 06f028f..caba34f 100644
--- a/xlators/mgmt/glusterd/src/glusterd-ganesha.c
+++ b/xlators/mgmt/glusterd/src/glusterd-ganesha.c
@@ -659,8 +659,8 @@ tear_down_cluster(gf_boolean_t run_teardown)
"Failed to close dir %s. Reason :"
" %s",
CONFDIR, strerror(errno));
- goto exit;
}
+ goto exit;
}
out:
--
1.8.3.1

38
SOURCES/0435-glusterd-coverity-fix.patch Normal file
View File

@ -0,0 +1,38 @@
From 9425fd5a49a17a8f91c13632ae055a6510b0b44c Mon Sep 17 00:00:00 2001
From: Sanju Rakonde <srakonde@redhat.com>
Date: Fri, 17 May 2019 14:27:58 +0530
Subject: [PATCH 435/449] glusterd: coverity fix
CID: 1401345 - Unused value
> upstream patch link: https://review.gluster.org/#/c/glusterfs/+/22737/
> updates: bz#789278
> Change-Id: I6b8f2611151ce0174042384b7632019c312ebae3
> Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
BUG: 1787310
Change-Id: I6b8f2611151ce0174042384b7632019c312ebae3
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202622
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/mgmt/glusterd/src/glusterd-utils.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index 8b0fc9a..2eb2a76 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -3740,7 +3740,7 @@ glusterd_compare_friend_volume(dict_t *peer_data, int32_t count,
* deleting a volume, so we shouldn't be trying to create a
* fresh volume here which would lead to a stale entry
*/
- if (stage_deleted == 0)
+ if (!ret && stage_deleted == 0)
*status = GLUSTERD_VOL_COMP_UPDATE_REQ;
ret = 0;
goto out;
--
1.8.3.1

187
SOURCES/0436-glusterd-coverity-fixes.patch Normal file
View File

@ -0,0 +1,187 @@
From 179213798496448316547506da65dbd9fd741dfa Mon Sep 17 00:00:00 2001
From: Atin Mukherjee <amukherj@redhat.com>
Date: Wed, 24 Apr 2019 22:02:51 +0530
Subject: [PATCH 436/449] glusterd: coverity fixes
Addresses the following:
* CID 1124776: Resource leaks (RESOURCE_LEAK) - Variable "aa" going out
of scope leaks the storage it points to in glusterd-volgen.c
* Bunch of CHECKED_RETURN defects in the callers of synctask_barrier_init
* CID 1400742: Program hangs (LOCK) - adding annotation to fix this
false positive
> upstream patch link: https://review.gluster.org/#/c/glusterfs/+/22615
> Updates: bz#789278
> Change-Id: I02f16e7eeb8c5cf72f7d0b29d00df4f03b3718b3
> Signed-off-by: Atin Mukherjee <amukherj@redhat.com>
BUG: 1787310
Change-Id: I02f16e7eeb8c5cf72f7d0b29d00df4f03b3718b3
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202626
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/mgmt/glusterd/src/glusterd-handler.c | 6 ++++++
xlators/mgmt/glusterd/src/glusterd-mgmt.c | 24 +++++++++++++++++++-----
xlators/mgmt/glusterd/src/glusterd-syncop.c | 22 ++++++++++++++++++----
xlators/mgmt/glusterd/src/glusterd-volgen.c | 5 +++--
4 files changed, 46 insertions(+), 11 deletions(-)
diff --git a/xlators/mgmt/glusterd/src/glusterd-handler.c b/xlators/mgmt/glusterd/src/glusterd-handler.c
index 1f31e72..b8799ab 100644
--- a/xlators/mgmt/glusterd/src/glusterd-handler.c
+++ b/xlators/mgmt/glusterd/src/glusterd-handler.c
@@ -3458,6 +3458,12 @@ glusterd_friend_remove(uuid_t uuid, char *hostname)
ret = glusterd_peerinfo_cleanup(peerinfo);
out:
gf_msg_debug(THIS->name, 0, "returning %d", ret);
+ /* We don't need to do a mutex unlock of peerinfo->delete_lock as the same
+ * will be anyway destroyed within glusterd_peerinfo_cleanup, coverity
+ * though cries about it
+ */
+ /* coverity[LOCK] */
+
return ret;
}
diff --git a/xlators/mgmt/glusterd/src/glusterd-mgmt.c b/xlators/mgmt/glusterd/src/glusterd-mgmt.c
index a4915f3..1e185d7 100644
--- a/xlators/mgmt/glusterd/src/glusterd-mgmt.c
+++ b/xlators/mgmt/glusterd/src/glusterd-mgmt.c
@@ -757,7 +757,10 @@ glusterd_mgmt_v3_initiate_lockdown(glusterd_op_t op, dict_t *dict,
/* Sending mgmt_v3 lock req to other nodes in the cluster */
gd_syncargs_init(&args, NULL);
- synctask_barrier_init((&args));
+ ret = synctask_barrier_init((&args));
+ if (ret)
+ goto out;
+
peer_cnt = 0;
RCU_READ_LOCK;
@@ -1108,7 +1111,10 @@ glusterd_mgmt_v3_pre_validate(glusterd_op_t op, dict_t *req_dict,
/* Sending Pre Validation req to other nodes in the cluster */
gd_syncargs_init(&args, req_dict);
- synctask_barrier_init((&args));
+ ret = synctask_barrier_init((&args));
+ if (ret)
+ goto out;
+
peer_cnt = 0;
RCU_READ_LOCK;
@@ -1458,7 +1464,10 @@ glusterd_mgmt_v3_brick_op(glusterd_op_t op, dict_t *op_ctx, dict_t *req_dict,
/* Sending brick op req to other nodes in the cluster */
gd_syncargs_init(&args, op_ctx);
- synctask_barrier_init((&args));
+ ret = synctask_barrier_init((&args));
+ if (ret)
+ goto out;
+
peer_cnt = 0;
RCU_READ_LOCK;
@@ -1722,7 +1731,9 @@ glusterd_mgmt_v3_commit(glusterd_op_t op, dict_t *op_ctx, dict_t *req_dict,
/* Sending commit req to other nodes in the cluster */
gd_syncargs_init(&args, op_ctx);
- synctask_barrier_init((&args));
+ ret = synctask_barrier_init((&args));
+ if (ret)
+ goto out;
peer_cnt = 0;
RCU_READ_LOCK;
@@ -1963,7 +1974,10 @@ glusterd_mgmt_v3_post_validate(glusterd_op_t op, int32_t op_ret, dict_t *dict,
/* Sending Post Validation req to other nodes in the cluster */
gd_syncargs_init(&args, req_dict);
- synctask_barrier_init((&args));
+ ret = synctask_barrier_init((&args));
+ if (ret)
+ goto out;
+
peer_cnt = 0;
RCU_READ_LOCK;
diff --git a/xlators/mgmt/glusterd/src/glusterd-syncop.c b/xlators/mgmt/glusterd/src/glusterd-syncop.c
index 9e47d14..c78983a 100644
--- a/xlators/mgmt/glusterd/src/glusterd-syncop.c
+++ b/xlators/mgmt/glusterd/src/glusterd-syncop.c
@@ -1191,7 +1191,12 @@ gd_lock_op_phase(glusterd_conf_t *conf, glusterd_op_t op, dict_t *op_ctx,
struct syncargs args = {0};
this = THIS;
- synctask_barrier_init((&args));
+ GF_VALIDATE_OR_GOTO("glusterd", this, out);
+
+ ret = synctask_barrier_init((&args));
+ if (ret)
+ goto out;
+
peer_cnt = 0;
RCU_READ_LOCK;
@@ -1321,7 +1326,10 @@ stage_done:
}
gd_syncargs_init(&args, aggr_dict);
- synctask_barrier_init((&args));
+ ret = synctask_barrier_init((&args));
+ if (ret)
+ goto out;
+
peer_cnt = 0;
RCU_READ_LOCK;
@@ -1449,7 +1457,10 @@ commit_done:
}
gd_syncargs_init(&args, op_ctx);
- synctask_barrier_init((&args));
+ ret = synctask_barrier_init((&args));
+ if (ret)
+ goto out;
+
peer_cnt = 0;
origin_glusterd = is_origin_glusterd(req_dict);
@@ -1541,7 +1552,10 @@ gd_unlock_op_phase(glusterd_conf_t *conf, glusterd_op_t op, int *op_ret,
goto out;
}
- synctask_barrier_init((&args));
+ ret = synctask_barrier_init((&args));
+ if (ret)
+ goto out;
+
peer_cnt = 0;
if (cluster_lock) {
diff --git a/xlators/mgmt/glusterd/src/glusterd-volgen.c b/xlators/mgmt/glusterd/src/glusterd-volgen.c
index 6852f8e..16346e7 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volgen.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volgen.c
@@ -4808,9 +4808,10 @@ nfs_option_handler(volgen_graph_t *graph, struct volopt_map_entry *vme,
if (ret != -1) {
ret = gf_canonicalize_path(vme->value);
- if (ret)
+ if (ret) {
+ GF_FREE(aa);
return -1;
-
+ }
ret = xlator_set_option(xl, aa, ret, vme->value);
GF_FREE(aa);
}
--
1.8.3.1

48
SOURCES/0437-glusterd-prevent-use-after-free-in-glusterd_op_ac_se.patch Normal file
View File

@ -0,0 +1,48 @@
From ffd428d07036531b7ed98c7393b87490aaa223ec Mon Sep 17 00:00:00 2001
From: Niels de Vos <ndevos@redhat.com>
Date: Fri, 3 May 2019 09:18:31 +0200
Subject: [PATCH 437/449] glusterd: prevent use-after-free in
glusterd_op_ac_send_brick_op()
Coverity reported that GF_FREE(req_ctx) could be called 2x on req_ctx.
> upstream patch link: https://review.gluster.org/#/c/glusterfs/+/22656/
> Change-Id: I9120686e5920de8c27688e10de0db6aa26292064
> CID: 1401115
> Updates: bz#789278
> Signed-off-by: Niels de Vos <ndevos@redhat.com>
BUG: 1787310
Change-Id: I9120686e5920de8c27688e10de0db6aa26292064
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202619
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/mgmt/glusterd/src/glusterd-op-sm.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/xlators/mgmt/glusterd/src/glusterd-op-sm.c b/xlators/mgmt/glusterd/src/glusterd-op-sm.c
index 46fc607..1e84f5f 100644
--- a/xlators/mgmt/glusterd/src/glusterd-op-sm.c
+++ b/xlators/mgmt/glusterd/src/glusterd-op-sm.c
@@ -7575,7 +7575,6 @@ glusterd_op_ac_send_brick_op(glusterd_op_sm_event_t *event, void *ctx)
if (op_errstr == NULL)
gf_asprintf(&op_errstr, OPERRSTR_BUILD_PAYLOAD);
opinfo.op_errstr = op_errstr;
- GF_FREE(req_ctx);
goto out;
}
}
@@ -7594,7 +7593,7 @@ glusterd_op_ac_send_brick_op(glusterd_op_sm_event_t *event, void *ctx)
}
out:
- if (ret && req_ctx && free_req_ctx)
+ if (ret && free_req_ctx)
GF_FREE(req_ctx);
gf_msg_debug(this->name, 0, "Returning with %d", ret);
--
1.8.3.1

324
SOURCES/0438-dht-sparse-files-rebalance-enhancements.patch Normal file
View File

@ -0,0 +1,324 @@
From 7b2f1bd4e5a57ea3abd5f14a7d81b120735faecd Mon Sep 17 00:00:00 2001
From: Barak Sason Rofman <bsasonro@redhat.com>
Date: Wed, 6 May 2020 13:28:40 +0300
Subject: [PATCH 438/449] dht - sparse files rebalance enhancements
Currently data migration in rebalance reads sparse file sequentially,
disregarding which segments are holes and which are data. This can lead
to extremely long migration time for large sparse file.
Data migration mechanism needs to be enhanced so only data segments are
read and migrated. This can be achieved using lseek to seek for holes
and data in the file.
This enhancement is a consequence of
https://bugzilla.redhat.com/show_bug.cgi?id=1823703
> fixes: #1222
> Change-Id: If5f448a0c532926464e1f34f504c5c94749b08c3
> Signed-off-by: Barak Sason Rofman <bsasonro@redhat.com>
> (Cherry pick from commit 7b7559733ca0c25c63f9d56cb7f4650dbd694c40)
> (Reviewed on upstream link https://review.gluster.org/#/c/glusterfs/+/24409/)
BUG: 1836099
Change-Id: If5f448a0c532926464e1f34f504c5c94749b08c3
Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202647
Reviewed-by: Barak Sason Rofman <bsasonro@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/basic/distribute/spare_file_rebalance.t | 51 ++++++++
xlators/cluster/dht/src/dht-rebalance.c | 172 ++++++++++++--------------
2 files changed, 130 insertions(+), 93 deletions(-)
create mode 100644 tests/basic/distribute/spare_file_rebalance.t
diff --git a/tests/basic/distribute/spare_file_rebalance.t b/tests/basic/distribute/spare_file_rebalance.t
new file mode 100644
index 0000000..061c02f
--- /dev/null
+++ b/tests/basic/distribute/spare_file_rebalance.t
@@ -0,0 +1,51 @@
+#!/bin/bash
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../dht.rc
+
+# Initialize
+#------------------------------------------------------------
+cleanup;
+
+# Start glusterd
+TEST glusterd;
+TEST pidof glusterd;
+TEST $CLI volume info;
+
+# Create a volume
+TEST $CLI volume create $V0 $H0:$B0/${V0}{1,2};
+
+# Verify volume creation
+EXPECT "$V0" volinfo_field $V0 'Volume Name';
+EXPECT 'Created' volinfo_field $V0 'Status';
+
+# Start volume and verify successful start
+TEST $CLI volume start $V0;
+EXPECT 'Started' volinfo_field $V0 'Status';
+TEST glusterfs --volfile-id=$V0 --volfile-server=$H0 --entry-timeout=0 $M0;
+
+#------------------------------------------------------------
+
+# Test case - Create sparse files on MP and verify
+# file info after rebalance
+#------------------------------------------------------------
+
+# Create some sparse files and get their size
+TEST cd $M0;
+dd if=/dev/urandom of=sparse_file bs=10k count=1 seek=2M
+cp --sparse=always sparse_file sparse_file_3;
+
+# Add a 3rd brick
+TEST $CLI volume add-brick $V0 $H0:$B0/${V0}3;
+
+# Trigger rebalance
+TEST $CLI volume rebalance $V0 start force;
+EXPECT_WITHIN $REBALANCE_TIMEOUT "0" rebalance_completed;
+
+# Compare original and rebalanced files
+TEST cd $B0/${V0}2
+TEST cmp sparse_file $B0/${V0}3/sparse_file_3
+EXPECT_WITHIN 30 "";
+
+cleanup;
diff --git a/xlators/cluster/dht/src/dht-rebalance.c b/xlators/cluster/dht/src/dht-rebalance.c
index 88b6b54..d0c21b4 100644
--- a/xlators/cluster/dht/src/dht-rebalance.c
+++ b/xlators/cluster/dht/src/dht-rebalance.c
@@ -18,8 +18,8 @@
#include <glusterfs/events.h>
#define GF_DISK_SECTOR_SIZE 512
-#define DHT_REBALANCE_PID 4242 /* Change it if required */
-#define DHT_REBALANCE_BLKSIZE (1024 * 1024) /* 1 MB */
+#define DHT_REBALANCE_PID 4242 /* Change it if required */
+#define DHT_REBALANCE_BLKSIZE 1048576 /* 1 MB */
#define MAX_MIGRATE_QUEUE_COUNT 500
#define MIN_MIGRATE_QUEUE_COUNT 200
#define MAX_REBAL_TYPE_SIZE 16
@@ -178,75 +178,6 @@ dht_strip_out_acls(dict_t *dict)
}
}
-static int
-dht_write_with_holes(xlator_t *to, fd_t *fd, struct iovec *vec, int count,
- int32_t size, off_t offset, struct iobref *iobref,
- int *fop_errno)
-{
- int i = 0;
- int ret = -1;
- int start_idx = 0;
- int tmp_offset = 0;
- int write_needed = 0;
- int buf_len = 0;
- int size_pending = 0;
- char *buf = NULL;
-
- /* loop through each vector */
- for (i = 0; i < count; i++) {
- buf = vec[i].iov_base;
- buf_len = vec[i].iov_len;
-
- for (start_idx = 0; (start_idx + GF_DISK_SECTOR_SIZE) <= buf_len;
- start_idx += GF_DISK_SECTOR_SIZE) {
- if (mem_0filled(buf + start_idx, GF_DISK_SECTOR_SIZE) != 0) {
- write_needed = 1;
- continue;
- }
-
- if (write_needed) {
- ret = syncop_write(
- to, fd, (buf + tmp_offset), (start_idx - tmp_offset),
- (offset + tmp_offset), iobref, 0, NULL, NULL);
- /* 'path' will be logged in calling function */
- if (ret < 0) {
- gf_log(THIS->name, GF_LOG_WARNING, "failed to write (%s)",
- strerror(-ret));
- *fop_errno = -ret;
- ret = -1;
- goto out;
- }
-
- write_needed = 0;
- }
- tmp_offset = start_idx + GF_DISK_SECTOR_SIZE;
- }
-
- if ((start_idx < buf_len) || write_needed) {
- /* This means, last chunk is not yet written.. write it */
- ret = syncop_write(to, fd, (buf + tmp_offset),
- (buf_len - tmp_offset), (offset + tmp_offset),
- iobref, 0, NULL, NULL);
- if (ret < 0) {
- /* 'path' will be logged in calling function */
- gf_log(THIS->name, GF_LOG_WARNING, "failed to write (%s)",
- strerror(-ret));
- *fop_errno = -ret;
- ret = -1;
- goto out;
- }
- }
-
- size_pending = (size - buf_len);
- if (!size_pending)
- break;
- }
-
- ret = size;
-out:
- return ret;
-}
-
/*
return values:
-1 : failure
@@ -1101,32 +1032,97 @@ __dht_rebalance_migrate_data(xlator_t *this, gf_defrag_info_t *defrag,
int ret = 0;
int count = 0;
off_t offset = 0;
+ off_t data_offset = 0;
+ off_t hole_offset = 0;
struct iovec *vector = NULL;
struct iobref *iobref = NULL;
uint64_t total = 0;
size_t read_size = 0;
+ size_t data_block_size = 0;
dict_t *xdata = NULL;
dht_conf_t *conf = NULL;
conf = this->private;
+
/* if file size is '0', no need to enter this loop */
while (total < ia_size) {
- read_size = (((ia_size - total) > DHT_REBALANCE_BLKSIZE)
- ? DHT_REBALANCE_BLKSIZE
- : (ia_size - total));
+ /* This is a regular file - read it sequentially */
+ if (!hole_exists) {
+ read_size = (((ia_size - total) > DHT_REBALANCE_BLKSIZE)
+ ? DHT_REBALANCE_BLKSIZE
+ : (ia_size - total));
+ } else {
+ /* This is a sparse file - read only the data segments in the file
+ */
+
+ /* If the previous data block is fully copied, find the next data
+ * segment
+ * starting at the offset of the last read and written byte, */
+ if (data_block_size <= 0) {
+ ret = syncop_seek(from, src, offset, GF_SEEK_DATA, NULL,
+ &data_offset);
+ if (ret) {
+ if (ret == -ENXIO)
+ ret = 0; /* No more data segments */
+ else
+ *fop_errno = -ret; /* Error occurred */
+
+ break;
+ }
+
+ /* If the position of the current data segment is greater than
+ * the position of the next hole, find the next hole in order to
+ * calculate the length of the new data segment */
+ if (data_offset > hole_offset) {
+ /* Starting at the offset of the last data segment, find the
+ * next hole */
+ ret = syncop_seek(from, src, data_offset, GF_SEEK_HOLE,
+ NULL, &hole_offset);
+ if (ret) {
+ /* If an error occurred here it's a real error because
+ * if the seek for a data segment was successful then
+ * necessarily another hole must exist (EOF is a hole)
+ */
+ *fop_errno = -ret;
+ break;
+ }
+
+ /* Calculate the total size of the current data block */
+ data_block_size = hole_offset - data_offset;
+ }
+ } else {
+ /* There is still data in the current segment, move the
+ * data_offset to the position of the last written byte */
+ data_offset = offset;
+ }
+
+ /* Calculate how much data needs to be read and written. If the data
+ * segment's length is bigger than DHT_REBALANCE_BLKSIZE, read and
+ * write DHT_REBALANCE_BLKSIZE data length and the rest in the
+ * next iteration(s) */
+ read_size = ((data_block_size > DHT_REBALANCE_BLKSIZE)
+ ? DHT_REBALANCE_BLKSIZE
+ : data_block_size);
+
+ /* Calculate the remaining size of the data block - maybe there's no
+ * need to seek for data in the next iteration */
+ data_block_size -= read_size;
+
+ /* Set offset to the offset of the data segment so read and write
+ * will have the correct position */
+ offset = data_offset;
+ }
ret = syncop_readv(from, src, read_size, offset, 0, &vector, &count,
&iobref, NULL, NULL, NULL);
+
if (!ret || (ret < 0)) {
*fop_errno = -ret;
break;
}
- if (hole_exists) {
- ret = dht_write_with_holes(to, dst, vector, count, ret, offset,
- iobref, fop_errno);
- } else {
- if (!conf->force_migration && !dht_is_tier_xlator(this)) {
+ if (!conf->force_migration && !dht_is_tier_xlator(this)) {
+ if (!xdata) {
xdata = dict_new();
if (!xdata) {
gf_msg("dht", GF_LOG_ERROR, 0, DHT_MSG_MIGRATE_FILE_FAILED,
@@ -1146,7 +1142,7 @@ __dht_rebalance_migrate_data(xlator_t *this, gf_defrag_info_t *defrag,
* https://github.com/gluster/glusterfs/issues/308
* for more details.
*/
- ret = dict_set_int32(xdata, GF_AVOID_OVERWRITE, 1);
+ ret = dict_set_int32_sizen(xdata, GF_AVOID_OVERWRITE, 1);
if (ret) {
gf_msg("dht", GF_LOG_ERROR, 0, ENOMEM,
"failed to set dict");
@@ -1155,22 +1151,12 @@ __dht_rebalance_migrate_data(xlator_t *this, gf_defrag_info_t *defrag,
break;
}
}
-
- ret = syncop_writev(to, dst, vector, count, offset, iobref, 0, NULL,
- NULL, xdata, NULL);
- if (ret < 0) {
- *fop_errno = -ret;
- }
- }
-
- if ((defrag && defrag->cmd == GF_DEFRAG_CMD_START_TIER) &&
- (gf_defrag_get_pause_state(&defrag->tier_conf) != TIER_RUNNING)) {
- gf_msg("tier", GF_LOG_INFO, 0, DHT_MSG_TIER_PAUSED,
- "Migrate file paused");
- ret = -1;
}
+ ret = syncop_writev(to, dst, vector, count, offset, iobref, 0, NULL,
+ NULL, xdata, NULL);
if (ret < 0) {
+ *fop_errno = -ret;
break;
}
--
1.8.3.1

28
SOURCES/0387-cluster-afr-Delay-post-op-for-fsync.patch → SOURCES/0439-cluster-afr-Delay-post-op-for-fsync.patch
View File

@ -1,7 +1,7 @@
From 399fad1ac0f9273483270e8af06a5b2d28927533 Mon Sep 17 00:00:00 2001
From 3ed98fc9dcb39223032e343fd5b0ad17fa3cae14 Mon Sep 17 00:00:00 2001
From: Pranith Kumar K <pkarampu@redhat.com>
Date: Fri, 29 May 2020 14:24:53 +0530
Subject: [PATCH 387/392] cluster/afr: Delay post-op for fsync
Subject: [PATCH 439/449] cluster/afr: Delay post-op for fsync
Problem:
AFR doesn't delay post-op for fsync fop. For fsync heavy workloads
@ -18,13 +18,11 @@ happens before delayed-post-op completes.
> Upstream-patch: https://review.gluster.org/c/glusterfs/+/24473
> Fixes: #1253
BUG: 1848896
BUG: 1838479
Change-Id: I531940d13269a111c49e0510d49514dc169f4577
Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
(cherry picked from commit 3ed98fc9dcb39223032e343fd5b0ad17fa3cae14)
Reviewed-on: https://code.engineering.redhat.com/gerrit/203694
Reviewed-on: https://code.engineering.redhat.com/gerrit/202676
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Tested-by: Karthik Subrahmanya <ksubrahm@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
api/src/glfs-resolve.c | 14 ++-
@ -329,7 +327,7 @@ index 8e65ae2..ffd0ab8 100644
goto out;
}
diff --git a/xlators/cluster/afr/src/afr.h b/xlators/cluster/afr/src/afr.h
index e731cfa..6bc4721 100644
index 18f1a6a..ff96246 100644
--- a/xlators/cluster/afr/src/afr.h
+++ b/xlators/cluster/afr/src/afr.h
@@ -854,7 +854,7 @@ typedef struct _afr_local {
@ -342,10 +340,10 @@ index e731cfa..6bc4721 100644
syncbarrier_t barrier;
diff --git a/xlators/cluster/dht/src/dht-rebalance.c b/xlators/cluster/dht/src/dht-rebalance.c
index 8f31dca..145e616 100644
index d0c21b4..e9974cd 100644
--- a/xlators/cluster/dht/src/dht-rebalance.c
+++ b/xlators/cluster/dht/src/dht-rebalance.c
@@ -1564,6 +1564,7 @@ dht_migrate_file(xlator_t *this, loc_t *loc, xlator_t *from, xlator_t *to,
@@ -1550,6 +1550,7 @@ dht_migrate_file(xlator_t *this, loc_t *loc, xlator_t *from, xlator_t *to,
xlator_t *old_target = NULL;
xlator_t *hashed_subvol = NULL;
fd_t *linkto_fd = NULL;
@ -353,7 +351,7 @@ index 8f31dca..145e616 100644
if (from == to) {
gf_msg_debug(this->name, 0,
@@ -1882,7 +1883,15 @@ dht_migrate_file(xlator_t *this, loc_t *loc, xlator_t *from, xlator_t *to,
@@ -1868,7 +1869,15 @@ dht_migrate_file(xlator_t *this, loc_t *loc, xlator_t *from, xlator_t *to,
/* TODO: Sync the locks */
@ -370,7 +368,7 @@ index 8f31dca..145e616 100644
if (ret) {
gf_log(this->name, GF_LOG_WARNING, "%s: failed to fsync on %s (%s)",
loc->path, to->name, strerror(-ret));
@@ -2356,11 +2365,15 @@ out:
@@ -2342,11 +2351,15 @@ out:
if (dst_fd)
syncop_close(dst_fd);
@ -387,10 +385,10 @@ index 8f31dca..145e616 100644
loc_wipe(&parent_loc);
diff --git a/xlators/mount/fuse/src/fuse-bridge.c b/xlators/mount/fuse/src/fuse-bridge.c
index 6e99053..1592067 100644
index fdeec49..4264fad 100644
--- a/xlators/mount/fuse/src/fuse-bridge.c
+++ b/xlators/mount/fuse/src/fuse-bridge.c
@@ -5551,6 +5551,7 @@ fuse_migrate_fd(xlator_t *this, fd_t *basefd, xlator_t *old_subvol,
@@ -5559,6 +5559,7 @@ fuse_migrate_fd(xlator_t *this, fd_t *basefd, xlator_t *old_subvol,
char create_in_progress = 0;
fuse_fd_ctx_t *basefd_ctx = NULL;
fd_t *oldfd = NULL;
@ -398,7 +396,7 @@ index 6e99053..1592067 100644
basefd_ctx = fuse_fd_ctx_get(this, basefd);
GF_VALIDATE_OR_GOTO("glusterfs-fuse", basefd_ctx, out);
@@ -5587,10 +5588,23 @@ fuse_migrate_fd(xlator_t *this, fd_t *basefd, xlator_t *old_subvol,
@@ -5595,10 +5596,23 @@ fuse_migrate_fd(xlator_t *this, fd_t *basefd, xlator_t *old_subvol,
}
if (oldfd->inode->table->xl == old_subvol) {
@ -425,7 +423,7 @@ index 6e99053..1592067 100644
if (ret < 0) {
gf_log("glusterfs-fuse", GF_LOG_WARNING,
@@ -5645,6 +5659,9 @@ out:
@@ -5653,6 +5667,9 @@ out:
fd_unref(oldfd);

62
SOURCES/0440-glusterd-snapshot-Improve-log-message-during-snapsho.patch Normal file
View File

@ -0,0 +1,62 @@
From 9cbab9110523cfafe23d6c6b3080d0d744062b85 Mon Sep 17 00:00:00 2001
From: Mohammed Rafi KC <rkavunga@redhat.com>
Date: Thu, 21 May 2020 16:04:33 +0530
Subject: [PATCH 440/449] glusterd/snapshot: Improve log message during
snapshot clone
While taking a snapshot clone, if the snapshot is not activated,
the cli was returning that the bricks are down.
This patch clearly print tha the error is due to the snapshot
state.
>Change-Id: Ia840e6e071342e061ad38bf15e2e2ff2b0dacdfa
>Fixes: #1255
>Signed-off-by: Mohammed Rafi KC <rkavunga@redhat.com>
Upstream Patch: https://review.gluster.org/#/c/glusterfs/+/24478/
BUG: 1837926
Change-Id: Ia840e6e071342e061ad38bf15e2e2ff2b0dacdfa
Signed-off-by: Srijan Sivakumar <ssivakum@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202707
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/mgmt/glusterd/src/glusterd-snapshot.c | 15 ++++++++++++---
1 file changed, 12 insertions(+), 3 deletions(-)
diff --git a/xlators/mgmt/glusterd/src/glusterd-snapshot.c b/xlators/mgmt/glusterd/src/glusterd-snapshot.c
index c56be91..5b8ae97 100644
--- a/xlators/mgmt/glusterd/src/glusterd-snapshot.c
+++ b/xlators/mgmt/glusterd/src/glusterd-snapshot.c
@@ -2238,7 +2238,6 @@ glusterd_snapshot_clone_prevalidate(dict_t *dict, char **op_errstr,
char *clonename = NULL;
char *snapname = NULL;
char device_name[64] = "";
- char key[PATH_MAX] = "";
glusterd_snap_t *snap = NULL;
char err_str[PATH_MAX] = "";
int ret = -1;
@@ -2299,8 +2298,18 @@ glusterd_snapshot_clone_prevalidate(dict_t *dict, char **op_errstr,
goto out;
}
- snprintf(key, sizeof(key) - 1, "vol1_volid");
- ret = dict_get_bin(dict, key, (void **)&snap_volid);
+
+ if (!glusterd_is_volume_started(snap_vol)) {
+ snprintf(err_str, sizeof(err_str),
+ "Snapshot %s is "
+ "not activated",
+ snap->snapname);
+ loglevel = GF_LOG_WARNING;
+ *op_errno = EG_VOLSTP;
+ goto out;
+ }
+
+ ret = dict_get_bin(dict, "vol1_volid", (void **)&snap_volid);
if (ret) {
gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_GET_FAILED,
"Unable to fetch snap_volid");
--
1.8.3.1

195
SOURCES/0441-fuse-occasional-logging-for-fuse-device-weird-write-.patch Normal file
View File

@ -0,0 +1,195 @@
From 1bde083cbd1e06be66d00e4ca52075687cee0d60 Mon Sep 17 00:00:00 2001
From: Csaba Henk <csaba@redhat.com>
Date: Fri, 8 May 2020 23:01:04 +0200
Subject: [PATCH 441/449] fuse: occasional logging for fuse device 'weird'
write errors
This change is a followup to
I510158843e4b1d482bdc496c2e97b1860dc1ba93.
In referred change we pushed log messages about 'weird'
write errors to fuse device out of sight, by reporting
them at Debug loglevel instead of Error (where
'weird' means errno is not POSIX compliant but having
meaningful semantics for FUSE protocol).
This solved the issue of spurious error reporting.
And so far so good: these messages don't indicate
an error condition by themselves. However, when they
come in high repetitions, that indicates a suboptimal
condition which should be reported.[1]
Therefore now we shall emit a Warning if a certain
errno occurs a certain number of times[2] as the
outcome of a write to the fuse device.
___
[1] typically ENOENTs and ENOTDIRs accumulate
when glusterfs' inode invalidation lags behind
the kernel's internal inode garbage collection
(in this case above errnos mean that the inode
which we requested to be invalidated is not found
in kernel). This can be mitigated with the
invalidate-limit command line / mount option,
cf. bz#1732717.
[2] 256, as of the current implementation.
Upstream on https://review.gluster.org/24415
> Change-Id: I8cc7fe104da43a88875f93b0db49d5677cc16045
> Updates: #1000
> Signed-off-by: Csaba Henk <csaba@redhat.com>
BUG: 1839137
Change-Id: I8448d6d328d47cb01d560cd99a2f43cd8dab312d
Signed-off-by: Csaba Henk <csaba@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202646
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/mount/fuse/src/fuse-bridge.c | 36 +++++++++++++++++++++++++++++++++++-
xlators/mount/fuse/src/fuse-bridge.h | 18 ++++++++++++++++++
2 files changed, 53 insertions(+), 1 deletion(-)
diff --git a/xlators/mount/fuse/src/fuse-bridge.c b/xlators/mount/fuse/src/fuse-bridge.c
index 4264fad..2e7584c 100644
--- a/xlators/mount/fuse/src/fuse-bridge.c
+++ b/xlators/mount/fuse/src/fuse-bridge.c
@@ -218,14 +218,30 @@ check_and_dump_fuse_W(fuse_private_t *priv, struct iovec *iov_out, int count,
if (res == -1) {
const char *errdesc = NULL;
gf_loglevel_t loglevel = GF_LOG_ERROR;
+ gf_boolean_t errno_degraded = _gf_false;
+ gf_boolean_t errno_promoted = _gf_false;
+
+#define ACCOUNT_ERRNO(eno) \
+ do { \
+ if (errno_degraded) { \
+ pthread_mutex_lock(&priv->fusedev_errno_cnt_mutex); \
+ { \
+ if (!++priv->fusedev_errno_cnt[FUSEDEV_##eno]) \
+ errno_promoted = _gf_true; \
+ } \
+ pthread_mutex_unlock(&priv->fusedev_errno_cnt_mutex); \
+ } \
+ } while (0)
/* If caller masked the errno, then it
* does not indicate an error at the application
* level, so we degrade the log severity to DEBUG.
*/
if (errnomask && errno < ERRNOMASK_MAX &&
- GET_ERRNO_MASK(errnomask, errno))
+ GET_ERRNO_MASK(errnomask, errno)) {
loglevel = GF_LOG_DEBUG;
+ errno_degraded = _gf_true;
+ }
switch (errno) {
/* The listed errnos are FUSE status indicators,
@@ -235,33 +251,43 @@ check_and_dump_fuse_W(fuse_private_t *priv, struct iovec *iov_out, int count,
*/
case ENOENT:
errdesc = "ENOENT";
+ ACCOUNT_ERRNO(ENOENT);
break;
case ENOTDIR:
errdesc = "ENOTDIR";
+ ACCOUNT_ERRNO(ENOTDIR);
break;
case ENODEV:
errdesc = "ENODEV";
+ ACCOUNT_ERRNO(ENODEV);
break;
case EPERM:
errdesc = "EPERM";
+ ACCOUNT_ERRNO(EPERM);
break;
case ENOMEM:
errdesc = "ENOMEM";
+ ACCOUNT_ERRNO(ENOMEM);
break;
case ENOTCONN:
errdesc = "ENOTCONN";
+ ACCOUNT_ERRNO(ENOTCONN);
break;
case ECONNREFUSED:
errdesc = "ECONNREFUSED";
+ ACCOUNT_ERRNO(ECONNREFUSED);
break;
case EOVERFLOW:
errdesc = "EOVERFLOW";
+ ACCOUNT_ERRNO(EOVERFLOW);
break;
case EBUSY:
errdesc = "EBUSY";
+ ACCOUNT_ERRNO(EBUSY);
break;
case ENOTEMPTY:
errdesc = "ENOTEMPTY";
+ ACCOUNT_ERRNO(ENOTEMPTY);
break;
default:
errdesc = strerror(errno);
@@ -269,7 +295,13 @@ check_and_dump_fuse_W(fuse_private_t *priv, struct iovec *iov_out, int count,
gf_log_callingfn("glusterfs-fuse", loglevel,
"writing to fuse device failed: %s", errdesc);
+ if (errno_promoted)
+ gf_log("glusterfs-fuse", GF_LOG_WARNING,
+ "writing to fuse device yielded %s %d times", errdesc,
+ UINT8_MAX + 1);
return errno;
+
+#undef ACCOUNT_ERRNO
}
fouh = iov_out[0].iov_base;
@@ -6584,6 +6616,8 @@ init(xlator_t *this_xl)
INIT_LIST_HEAD(&priv->interrupt_list);
pthread_mutex_init(&priv->interrupt_mutex, NULL);
+ pthread_mutex_init(&priv->fusedev_errno_cnt_mutex, NULL);
+
/* get options from option dictionary */
ret = dict_get_str(options, ZR_MOUNTPOINT_OPT, &value_string);
if (ret == -1 || value_string == NULL) {
diff --git a/xlators/mount/fuse/src/fuse-bridge.h b/xlators/mount/fuse/src/fuse-bridge.h
index d2d462c..2fb15a6 100644
--- a/xlators/mount/fuse/src/fuse-bridge.h
+++ b/xlators/mount/fuse/src/fuse-bridge.h
@@ -78,6 +78,20 @@ typedef struct fuse_in_header fuse_in_header_t;
typedef void(fuse_handler_t)(xlator_t *this, fuse_in_header_t *finh, void *msg,
struct iobuf *iobuf);
+enum fusedev_errno {
+ FUSEDEV_ENOENT,
+ FUSEDEV_ENOTDIR,
+ FUSEDEV_ENODEV,
+ FUSEDEV_EPERM,
+ FUSEDEV_ENOMEM,
+ FUSEDEV_ENOTCONN,
+ FUSEDEV_ECONNREFUSED,
+ FUSEDEV_EOVERFLOW,
+ FUSEDEV_EBUSY,
+ FUSEDEV_ENOTEMPTY,
+ FUSEDEV_EMAXPLUS
+};
+
struct fuse_private {
int fd;
uint32_t proto_minor;
@@ -192,6 +206,10 @@ struct fuse_private {
/* LRU Limit, if not set, default is 64k for now */
uint32_t lru_limit;
uint32_t invalidate_limit;
+
+ /* counters for fusdev errnos */
+ uint8_t fusedev_errno_cnt[FUSEDEV_EMAXPLUS];
+ pthread_mutex_t fusedev_errno_cnt_mutex;
};
typedef struct fuse_private fuse_private_t;
--
1.8.3.1

139
SOURCES/0442-fuse-correctly-handle-setxattr-values.patch Normal file
View File

@ -0,0 +1,139 @@
From 56c8ef4a64506c64aeb95d5a2c38d7107f90ac3a Mon Sep 17 00:00:00 2001
From: Xavi Hernandez <xhernandez@redhat.com>
Date: Tue, 5 Feb 2019 16:57:52 +0100
Subject: [PATCH 442/449] fuse: correctly handle setxattr values
The setxattr function receives a pointer to raw data, which may not be
null-terminated. When this data needs to be interpreted as a string, an
explicit null termination needs to be added before using the value.
Upstream patch https://review.gluster.org/#/c/glusterfs/+/22157
> Change-Id: Id110f9b215b22786da5782adec9449ce38d0d563
> updates: bz#1193929
> Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
Note: this change is not addressing the issue of bz 1787310,
indeed it is prerequisite for other changes that do.
BUG: 1787310
Change-Id: I56417b130eb2a1f388108456c905a577eb658793
Signed-off-by: Csaba Henk <csaba@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202758
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
libglusterfs/src/glusterfs/xlator.h | 2 +-
libglusterfs/src/xlator.c | 28 +++++++++++++++++++++++++---
xlators/mount/fuse/src/fuse-bridge.c | 20 ++++++++++++++++----
3 files changed, 42 insertions(+), 8 deletions(-)
diff --git a/libglusterfs/src/glusterfs/xlator.h b/libglusterfs/src/glusterfs/xlator.h
index db04c4d..8650ccc 100644
--- a/libglusterfs/src/glusterfs/xlator.h
+++ b/libglusterfs/src/glusterfs/xlator.h
@@ -1043,7 +1043,7 @@ xlator_mem_acct_init(xlator_t *xl, int num_types);
void
xlator_mem_acct_unref(struct mem_acct *mem_acct);
int
-is_gf_log_command(xlator_t *trans, const char *name, char *value);
+is_gf_log_command(xlator_t *trans, const char *name, char *value, size_t size);
int
glusterd_check_log_level(const char *value);
int
diff --git a/libglusterfs/src/xlator.c b/libglusterfs/src/xlator.c
index 6bd4f09..108b96a 100644
--- a/libglusterfs/src/xlator.c
+++ b/libglusterfs/src/xlator.c
@@ -1278,8 +1278,21 @@ xlator_destroy(xlator_t *xl)
return 0;
}
+static int32_t
+gf_bin_to_string(char *dst, size_t size, void *src, size_t len)
+{
+ if (len >= size) {
+ return EINVAL;
+ }
+
+ memcpy(dst, src, len);
+ dst[len] = 0;
+
+ return 0;
+}
+
int
-is_gf_log_command(xlator_t *this, const char *name, char *value)
+is_gf_log_command(xlator_t *this, const char *name, char *value, size_t size)
{
xlator_t *trav = NULL;
char key[1024] = {
@@ -1291,7 +1304,11 @@ is_gf_log_command(xlator_t *this, const char *name, char *value)
glusterfs_ctx_t *ctx = NULL;
if (!strcmp("trusted.glusterfs.syslog", name)) {
- ret = gf_string2boolean(value, &syslog_flag);
+ ret = gf_bin_to_string(key, sizeof(key), value, size);
+ if (ret != 0) {
+ goto out;
+ }
+ ret = gf_string2boolean(key, &syslog_flag);
if (ret) {
ret = EOPNOTSUPP;
goto out;
@@ -1307,7 +1324,12 @@ is_gf_log_command(xlator_t *this, const char *name, char *value)
if (fnmatch("trusted.glusterfs*set-log-level", name, FNM_NOESCAPE))
goto out;
- log_level = glusterd_check_log_level(value);
+ ret = gf_bin_to_string(key, sizeof(key), value, size);
+ if (ret != 0) {
+ goto out;
+ }
+
+ log_level = glusterd_check_log_level(key);
if (log_level == -1) {
ret = EOPNOTSUPP;
goto out;
diff --git a/xlators/mount/fuse/src/fuse-bridge.c b/xlators/mount/fuse/src/fuse-bridge.c
index 2e7584c..cfad2b4 100644
--- a/xlators/mount/fuse/src/fuse-bridge.c
+++ b/xlators/mount/fuse/src/fuse-bridge.c
@@ -4112,7 +4112,7 @@ fuse_setxattr(xlator_t *this, fuse_in_header_t *finh, void *msg,
/* Check if the command is for changing the log
level of process or specific xlator */
- ret = is_gf_log_command(this, name, value);
+ ret = is_gf_log_command(this, name, value, fsi->size);
if (ret >= 0) {
op_errno = ret;
goto done;
@@ -4159,11 +4159,23 @@ fuse_setxattr(xlator_t *this, fuse_in_header_t *finh, void *msg,
* fixups to make sure that's the case. To avoid nasty
* surprises, allocate an extra byte and add a NUL here.
*/
- dict_value = memdup(value, fsi->size + 1);
+ dict_value = GF_MALLOC(fsi->size + 1, gf_common_mt_char);
+ if (dict_value == NULL) {
+ gf_log("glusterfs-fuse", GF_LOG_ERROR,
+ "%" PRIu64 ": SETXATTR value allocation failed",
+ finh->unique);
+ op_errno = ENOMEM;
+ goto done;
+ }
+ memcpy(dict_value, value, fsi->size);
dict_value[fsi->size] = '\0';
}
- dict_set(state->xattr, newkey,
- data_from_dynptr((void *)dict_value, fsi->size));
+ ret = dict_set_dynptr(state->xattr, newkey, dict_value, fsi->size);
+ if (ret < 0) {
+ op_errno = -ret;
+ GF_FREE(dict_value);
+ goto done;
+ }
state->flags = fsi->flags;
state->name = newkey;
--
1.8.3.1

55
SOURCES/0443-fuse-fix-high-sev-coverity-issue.patch Normal file
View File

@ -0,0 +1,55 @@
From 3ac3312d63b9dc3c15cd8765ab8b7c601b007500 Mon Sep 17 00:00:00 2001
From: Sunny Kumar <sunkumar@redhat.com>
Date: Tue, 19 Mar 2019 22:51:14 +0530
Subject: [PATCH 443/449] fuse : fix high sev coverity issue
This patch fixed coverity issue in fuse-bridge.c.
CID : 1398630 : Resource leak
CID : 1399757 : Uninitialized pointer read
Upstream patch https://review.gluster.org/c/glusterfs/+/22382
> updates: bz#789278
>
> Change-Id: I69f8591400ee56a5d215eeac443a8e3d7777db27
> Signed-off-by: Sunny Kumar <sunkumar@redhat.com>
BUG: 1787310
Change-Id: Ib2c9af25019ee57131b3d384fc4b557437e75d3e
Signed-off-by: Csaba Henk <csaba@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202759
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/mount/fuse/src/fuse-bridge.c | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/xlators/mount/fuse/src/fuse-bridge.c b/xlators/mount/fuse/src/fuse-bridge.c
index cfad2b4..d17320b 100644
--- a/xlators/mount/fuse/src/fuse-bridge.c
+++ b/xlators/mount/fuse/src/fuse-bridge.c
@@ -4174,6 +4174,7 @@ fuse_setxattr(xlator_t *this, fuse_in_header_t *finh, void *msg,
if (ret < 0) {
op_errno = -ret;
GF_FREE(dict_value);
+ GF_FREE(newkey);
goto done;
}
@@ -5963,7 +5964,12 @@ fuse_thread_proc(void *data)
ssize_t res = 0;
struct iobuf *iobuf = NULL;
fuse_in_header_t *finh = NULL;
- struct iovec iov_in[2];
+ struct iovec iov_in[2] = {
+ {
+ 0,
+ },
+ };
+
void *msg = NULL;
/* we need 512 extra buffer size for BATCH_FORGET fop. By tests, it is
found to be reduces 'REALLOC()' in the loop */
--
1.8.3.1

40
SOURCES/0444-mount-fuse-Fixing-a-coverity-issue.patch Normal file
View File

@ -0,0 +1,40 @@
From 53a6aed98aad73ff51f884bf815bccfa337eb524 Mon Sep 17 00:00:00 2001
From: Barak Sason <bsasonro@redhat.com>
Date: Sun, 18 Aug 2019 17:38:09 +0300
Subject: [PATCH 444/449] mount/fuse - Fixing a coverity issue
Fixed resource leak of dict_value and newkey variables
CID: 1398630
Upstream patch https://review.gluster.org/c/glusterfs/+/23260
> Updates: bz#789278
>
> Change-Id: I589fdc0aecaeb4f446cd29f95bad36ccd7a35beb
> Signed-off-by: Barak Sason <bsasonro@redhat.com>
BUG: 1787310
Change-Id: Id191face7b082e2e8d6e62f60b56248688d396f6
Signed-off-by: Csaba Henk <csaba@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202760
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/mount/fuse/src/fuse-bridge.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/xlators/mount/fuse/src/fuse-bridge.c b/xlators/mount/fuse/src/fuse-bridge.c
index d17320b..f61fa39 100644
--- a/xlators/mount/fuse/src/fuse-bridge.c
+++ b/xlators/mount/fuse/src/fuse-bridge.c
@@ -4165,6 +4165,7 @@ fuse_setxattr(xlator_t *this, fuse_in_header_t *finh, void *msg,
"%" PRIu64 ": SETXATTR value allocation failed",
finh->unique);
op_errno = ENOMEM;
+ GF_FREE(newkey);
goto done;
}
memcpy(dict_value, value, fsi->size);
--
1.8.3.1

481
SOURCES/0445-feature-changelog-Avoid-thread-creation-if-xlator-is.patch Normal file
View File

@ -0,0 +1,481 @@
From dc03340654d921916ac3890d713fc84ef4bb1e28 Mon Sep 17 00:00:00 2001
From: Mohit Agrawal <moagrawal@redhat.com>
Date: Sat, 29 Sep 2018 13:15:35 +0530
Subject: [PATCH 445/449] feature/changelog: Avoid thread creation if xlator is
not enabled
Problem:
Changelog creates threads even if the changelog is not enabled
Background:
Changelog xlator broadly does two things
1. Journalling - Cosumers are geo-rep and glusterfind
2. Event Notification for registered events like (open, release etc) -
Consumers are bitrot, geo-rep
The existing option "changelog.changelog" controls journalling and
there is no option to control event notification and is enabled by
default. So when bitrot/geo-rep is not enabled on the volume, threads
and resources(rpc and rbuf) related to event notifications consumes
resources and cpu cycle which is unnecessary.
Solution:
The solution is to have two different options as below.
1. changelog-notification : Event notifications
2. changelog : Journalling
This patch introduces the option "changelog-notification" which is
not exposed to user. When either bitrot or changelog (journalling)
is enabled, it internally enbales 'changelog-notification'. But
once the 'changelog-notification' is enabled, it will not be disabled
for the life time of the brick process even after bitrot and changelog
is disabled. As of now, rpc resource cleanup has lot of races and is
difficult to cleanup cleanly. If allowed, it leads to memory leaks
and crashes on enable/disable of bitrot or changelog (journal) in a
loop. Hence to be safer, the event notification is not disabled within
lifetime of process once enabled.
> Change-Id: Ifd00286e0966049e8eb9f21567fe407cf11bb02a
> Updates: #475
> Signed-off-by: Mohit Agrawal <moagrawal@redhat.com>
> (Cherry pick from commit 6de80bcd6366778ac34ce58ec496fa08cc02bd0b)
> (Reviewed on upstream link https://review.gluster.org/#/c/glusterfs/+/21896/)
BUG: 1790336
Change-Id: Ifd00286e0966049e8eb9f21567fe407cf11bb02a
Signed-off-by: Mohit Agrawal <moagrawal@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202778
Tested-by: Mohit Agrawal <moagrawa@redhat.com>
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
rpc/rpc-lib/src/rpcsvc.c | 26 ++--
tests/basic/changelog/changelog-history.t | 12 +-
tests/bugs/bitrot/bug-1227996.t | 1 -
tests/bugs/bitrot/bug-1245981.t | 4 +-
xlators/features/changelog/src/changelog-helpers.h | 4 +
.../features/changelog/src/changelog-rpc-common.c | 3 +
xlators/features/changelog/src/changelog.c | 149 +++++++++++++++------
xlators/mgmt/glusterd/src/glusterd-volgen.c | 13 ++
8 files changed, 154 insertions(+), 58 deletions(-)
diff --git a/rpc/rpc-lib/src/rpcsvc.c b/rpc/rpc-lib/src/rpcsvc.c
index b058932..3f184bf 100644
--- a/rpc/rpc-lib/src/rpcsvc.c
+++ b/rpc/rpc-lib/src/rpcsvc.c
@@ -1865,6 +1865,18 @@ rpcsvc_program_unregister(rpcsvc_t *svc, rpcsvc_program_t *program)
goto out;
}
+ pthread_rwlock_rdlock(&svc->rpclock);
+ {
+ list_for_each_entry(prog, &svc->programs, program)
+ {
+ if ((prog->prognum == program->prognum) &&
+ (prog->progver == program->progver)) {
+ break;
+ }
+ }
+ }
+ pthread_rwlock_unlock(&svc->rpclock);
+
ret = rpcsvc_program_unregister_portmap(program);
if (ret == -1) {
gf_log(GF_RPCSVC, GF_LOG_ERROR,
@@ -1881,17 +1893,6 @@ rpcsvc_program_unregister(rpcsvc_t *svc, rpcsvc_program_t *program)
goto out;
}
#endif
- pthread_rwlock_rdlock(&svc->rpclock);
- {
- list_for_each_entry(prog, &svc->programs, program)
- {
- if ((prog->prognum == program->prognum) &&
- (prog->progver == program->progver)) {
- break;
- }
- }
- }
- pthread_rwlock_unlock(&svc->rpclock);
gf_log(GF_RPCSVC, GF_LOG_DEBUG,
"Program unregistered: %s, Num: %d,"
@@ -1912,6 +1913,9 @@ rpcsvc_program_unregister(rpcsvc_t *svc, rpcsvc_program_t *program)
ret = 0;
out:
+ if (prog)
+ GF_FREE(prog);
+
if (ret == -1) {
if (program) {
gf_log(GF_RPCSVC, GF_LOG_ERROR,
diff --git a/tests/basic/changelog/changelog-history.t b/tests/basic/changelog/changelog-history.t
index 3ce4098..b56e247 100644
--- a/tests/basic/changelog/changelog-history.t
+++ b/tests/basic/changelog/changelog-history.t
@@ -5,6 +5,7 @@
cleanup;
+SCRIPT_TIMEOUT=300
HISTORY_BIN_PATH=$(dirname $0)/../../utils/changelog
build_tester $HISTORY_BIN_PATH/get-history.c -lgfchangelog
@@ -68,18 +69,21 @@ TEST $CLI volume set $V0 changelog.changelog off
sleep 3
time_after_disable=$(date '+%s')
+TEST $CLI volume set $V0 changelog.changelog on
+sleep 5
+
#Passes, gives the changelogs till continuous changelogs are available
# but returns 1
-EXPECT "1" $HISTORY_BIN_PATH/get-history $time_after_enable1 $time_in_sec_htime2
+EXPECT_WITHIN 10 "1" $HISTORY_BIN_PATH/get-history $time_after_enable1 $time_in_sec_htime2
#Fails as start falls between htime files
-EXPECT "-3" $HISTORY_BIN_PATH/get-history $time_between_htime $time_in_sec_htime1
+EXPECT_WITHIN 10 "-3" $HISTORY_BIN_PATH/get-history $time_between_htime $time_in_sec_htime1
#Passes as start and end falls in same htime file
-EXPECT "0" $HISTORY_BIN_PATH/get-history $time_in_sec_htime1 $time_in_sec_htime2
+EXPECT_WITHIN 10 "0" $HISTORY_BIN_PATH/get-history $time_in_sec_htime1 $time_in_sec_htime2
#Passes, gives the changelogs till continuous changelogs are available
-EXPECT "0" $HISTORY_BIN_PATH/get-history $time_in_sec_htime2 $time_after_disable
+EXPECT_WITHIN 10 "0" $HISTORY_BIN_PATH/get-history $time_in_sec_htime2 $time_after_disable
TEST rm $HISTORY_BIN_PATH/get-history
diff --git a/tests/bugs/bitrot/bug-1227996.t b/tests/bugs/bitrot/bug-1227996.t
index 47ebc42..121c7b5 100644
--- a/tests/bugs/bitrot/bug-1227996.t
+++ b/tests/bugs/bitrot/bug-1227996.t
@@ -17,7 +17,6 @@ TEST pidof glusterd;
## Lets create and start the volume
TEST $CLI volume create $V0 $H0:$B0/${V0}0 $H0:$B0/${V0}1
TEST $CLI volume start $V0
-
## Enable bitrot on volume $V0
TEST $CLI volume bitrot $V0 enable
diff --git a/tests/bugs/bitrot/bug-1245981.t b/tests/bugs/bitrot/bug-1245981.t
index 2bed4d9..f395525 100644
--- a/tests/bugs/bitrot/bug-1245981.t
+++ b/tests/bugs/bitrot/bug-1245981.t
@@ -47,9 +47,9 @@ touch $M0/5
sleep `expr $SLEEP_TIME \* 2`
backpath=$(get_backend_paths $fname)
-TEST getfattr -m . -n trusted.bit-rot.signature $backpath
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT 'trusted.bit-rot.signature' check_for_xattr 'trusted.bit-rot.signature' $backpath
backpath=$(get_backend_paths $M0/new_file)
-TEST getfattr -m . -n trusted.bit-rot.signature $backpath
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT 'trusted.bit-rot.signature' check_for_xattr 'trusted.bit-rot.signature' $backpath
cleanup;
diff --git a/xlators/features/changelog/src/changelog-helpers.h b/xlators/features/changelog/src/changelog-helpers.h
index 517c4dc..3afacc9 100644
--- a/xlators/features/changelog/src/changelog-helpers.h
+++ b/xlators/features/changelog/src/changelog-helpers.h
@@ -190,8 +190,12 @@ typedef struct changelog_ev_selector {
/* changelog's private structure */
struct changelog_priv {
+ /* changelog journalling */
gf_boolean_t active;
+ /* changelog live notifications */
+ gf_boolean_t rpc_active;
+
/* to generate unique socket file per brick */
char *changelog_brick;
diff --git a/xlators/features/changelog/src/changelog-rpc-common.c b/xlators/features/changelog/src/changelog-rpc-common.c
index dcdcfb1..f2d1853 100644
--- a/xlators/features/changelog/src/changelog-rpc-common.c
+++ b/xlators/features/changelog/src/changelog-rpc-common.c
@@ -263,6 +263,9 @@ changelog_rpc_server_destroy(xlator_t *this, rpcsvc_t *rpc, char *sockfile,
struct rpcsvc_program *prog = NULL;
rpc_transport_t *trans = NULL;
+ if (!rpc)
+ return;
+
while (*progs) {
prog = *progs;
(void)rpcsvc_program_unregister(rpc, prog);
diff --git a/xlators/features/changelog/src/changelog.c b/xlators/features/changelog/src/changelog.c
index d9025f3..ff06c09 100644
--- a/xlators/features/changelog/src/changelog.c
+++ b/xlators/features/changelog/src/changelog.c
@@ -34,6 +34,12 @@ static struct changelog_bootstrap cb_bootstrap[] = {
},
};
+static int
+changelog_init_rpc(xlator_t *this, changelog_priv_t *priv);
+
+static int
+changelog_init(xlator_t *this, changelog_priv_t *priv);
+
/* Entry operations - TYPE III */
/**
@@ -2008,6 +2014,11 @@ notify(xlator_t *this, int event, void *data, ...)
uint64_t clntcnt = 0;
changelog_clnt_t *conn = NULL;
gf_boolean_t cleanup_notify = _gf_false;
+ char sockfile[UNIX_PATH_MAX] = {
+ 0,
+ };
+ rpcsvc_listener_t *listener = NULL;
+ rpcsvc_listener_t *next = NULL;
INIT_LIST_HEAD(&queue);
@@ -2021,23 +2032,40 @@ notify(xlator_t *this, int event, void *data, ...)
"cleanup changelog rpc connection of brick %s",
priv->victim->name);
- this->cleanup_starting = 1;
- changelog_destroy_rpc_listner(this, priv);
- conn = &priv->connections;
- if (conn)
- changelog_ev_cleanup_connections(this, conn);
- xprtcnt = GF_ATOMIC_GET(priv->xprtcnt);
- clntcnt = GF_ATOMIC_GET(priv->clntcnt);
-
- if (!xprtcnt && !clntcnt) {
- LOCK(&priv->lock);
- {
- cleanup_notify = priv->notify_down;
- priv->notify_down = _gf_true;
+ if (priv->rpc_active) {
+ this->cleanup_starting = 1;
+ changelog_destroy_rpc_listner(this, priv);
+ conn = &priv->connections;
+ if (conn)
+ changelog_ev_cleanup_connections(this, conn);
+ xprtcnt = GF_ATOMIC_GET(priv->xprtcnt);
+ clntcnt = GF_ATOMIC_GET(priv->clntcnt);
+ if (!xprtcnt && !clntcnt) {
+ LOCK(&priv->lock);
+ {
+ cleanup_notify = priv->notify_down;
+ priv->notify_down = _gf_true;
+ }
+ UNLOCK(&priv->lock);
+ list_for_each_entry_safe(listener, next, &priv->rpc->listeners,
+ list)
+ {
+ if (listener->trans) {
+ rpc_transport_unref(listener->trans);
+ }
+ }
+ CHANGELOG_MAKE_SOCKET_PATH(priv->changelog_brick, sockfile,
+ UNIX_PATH_MAX);
+ sys_unlink(sockfile);
+ if (priv->rpc) {
+ rpcsvc_destroy(priv->rpc);
+ priv->rpc = NULL;
+ }
+ if (!cleanup_notify)
+ default_notify(this, GF_EVENT_PARENT_DOWN, data);
}
- UNLOCK(&priv->lock);
- if (!cleanup_notify)
- default_notify(this, GF_EVENT_PARENT_DOWN, data);
+ } else {
+ default_notify(this, GF_EVENT_PARENT_DOWN, data);
}
goto out;
}
@@ -2425,6 +2453,22 @@ changelog_barrier_pthread_destroy(changelog_priv_t *priv)
LOCK_DESTROY(&priv->bflags.lock);
}
+static void
+changelog_cleanup_rpc(xlator_t *this, changelog_priv_t *priv)
+{
+ /* terminate rpc server */
+ if (!this->cleanup_starting)
+ changelog_destroy_rpc_listner(this, priv);
+
+ (void)changelog_cleanup_rpc_threads(this, priv);
+ /* cleanup rot buffs */
+ rbuf_dtor(priv->rbuf);
+
+ /* cleanup poller thread */
+ if (priv->poller)
+ (void)changelog_thread_cleanup(this, priv->poller);
+}
+
int
reconfigure(xlator_t *this, dict_t *options)
{
@@ -2433,6 +2477,9 @@ reconfigure(xlator_t *this, dict_t *options)
changelog_priv_t *priv = NULL;
gf_boolean_t active_earlier = _gf_true;
gf_boolean_t active_now = _gf_true;
+ gf_boolean_t rpc_active_earlier = _gf_true;
+ gf_boolean_t rpc_active_now = _gf_true;
+ gf_boolean_t iniate_rpc = _gf_false;
changelog_time_slice_t *slice = NULL;
changelog_log_data_t cld = {
0,
@@ -2454,6 +2501,7 @@ reconfigure(xlator_t *this, dict_t *options)
ret = -1;
active_earlier = priv->active;
+ rpc_active_earlier = priv->rpc_active;
/* first stop the rollover and the fsync thread */
changelog_cleanup_helper_threads(this, priv);
@@ -2487,6 +2535,29 @@ reconfigure(xlator_t *this, dict_t *options)
goto out;
GF_OPTION_RECONF("changelog", active_now, options, bool, out);
+ GF_OPTION_RECONF("changelog-notification", rpc_active_now, options, bool,
+ out);
+
+ /* If journalling is enabled, enable rpc notifications */
+ if (active_now && !active_earlier) {
+ if (!rpc_active_earlier)
+ iniate_rpc = _gf_true;
+ }
+
+ if (rpc_active_now && !rpc_active_earlier) {
+ iniate_rpc = _gf_true;
+ }
+
+ /* TODO: Disable of changelog-notifications is not supported for now
+ * as there is no clean way of cleaning up of rpc resources
+ */
+
+ if (iniate_rpc) {
+ ret = changelog_init_rpc(this, priv);
+ if (ret)
+ goto out;
+ priv->rpc_active = _gf_true;
+ }
/**
* changelog_handle_change() handles changes that could possibly
@@ -2618,6 +2689,7 @@ changelog_init_options(xlator_t *this, changelog_priv_t *priv)
goto dealloc_2;
GF_OPTION_INIT("changelog", priv->active, bool, dealloc_2);
+ GF_OPTION_INIT("changelog-notification", priv->rpc_active, bool, dealloc_2);
GF_OPTION_INIT("capture-del-path", priv->capture_del_path, bool, dealloc_2);
GF_OPTION_INIT("op-mode", tmp, str, dealloc_2);
@@ -2656,22 +2728,6 @@ error_return:
return -1;
}
-static void
-changelog_cleanup_rpc(xlator_t *this, changelog_priv_t *priv)
-{
- /* terminate rpc server */
- if (!this->cleanup_starting)
- changelog_destroy_rpc_listner(this, priv);
-
- (void)changelog_cleanup_rpc_threads(this, priv);
- /* cleanup rot buffs */
- rbuf_dtor(priv->rbuf);
-
- /* cleanup poller thread */
- if (priv->poller)
- (void)changelog_thread_cleanup(this, priv->poller);
-}
-
static int
changelog_init_rpc(xlator_t *this, changelog_priv_t *priv)
{
@@ -2768,10 +2824,13 @@ init(xlator_t *this)
INIT_LIST_HEAD(&priv->queue);
priv->barrier_enabled = _gf_false;
- /* RPC ball rolling.. */
- ret = changelog_init_rpc(this, priv);
- if (ret)
- goto cleanup_barrier;
+ if (priv->rpc_active || priv->active) {
+ /* RPC ball rolling.. */
+ ret = changelog_init_rpc(this, priv);
+ if (ret)
+ goto cleanup_barrier;
+ priv->rpc_active = _gf_true;
+ }
ret = changelog_init(this, priv);
if (ret)
@@ -2783,7 +2842,9 @@ init(xlator_t *this)
return 0;
cleanup_rpc:
- changelog_cleanup_rpc(this, priv);
+ if (priv->rpc_active) {
+ changelog_cleanup_rpc(this, priv);
+ }
cleanup_barrier:
changelog_barrier_pthread_destroy(priv);
cleanup_options:
@@ -2808,9 +2869,10 @@ fini(xlator_t *this)
priv = this->private;
if (priv) {
- /* terminate RPC server/threads */
- changelog_cleanup_rpc(this, priv);
-
+ if (priv->active || priv->rpc_active) {
+ /* terminate RPC server/threads */
+ changelog_cleanup_rpc(this, priv);
+ }
/* call barrier_disable to cancel timer */
if (priv->barrier_enabled)
__chlog_barrier_disable(this, &queue);
@@ -2879,6 +2941,13 @@ struct volume_options options[] = {
.flags = OPT_FLAG_SETTABLE,
.level = OPT_STATUS_BASIC,
.tags = {"journal", "georep", "glusterfind"}},
+ {.key = {"changelog-notification"},
+ .type = GF_OPTION_TYPE_BOOL,
+ .default_value = "off",
+ .description = "enable/disable changelog live notification",
+ .op_version = {3},
+ .level = OPT_STATUS_BASIC,
+ .tags = {"bitrot", "georep"}},
{.key = {"changelog-brick"},
.type = GF_OPTION_TYPE_PATH,
.description = "brick path to generate unique socket file name."
diff --git a/xlators/mgmt/glusterd/src/glusterd-volgen.c b/xlators/mgmt/glusterd/src/glusterd-volgen.c
index 16346e7..13f84ea 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volgen.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volgen.c
@@ -1876,6 +1876,19 @@ brick_graph_add_changelog(volgen_graph_t *graph, glusterd_volinfo_t *volinfo,
ret = xlator_set_fixed_option(xl, "changelog-dir", changelog_basepath);
if (ret)
goto out;
+
+ ret = glusterd_is_bitrot_enabled(volinfo);
+ if (ret == -1) {
+ goto out;
+ } else if (ret) {
+ ret = xlator_set_fixed_option(xl, "changelog-notification", "on");
+ if (ret)
+ goto out;
+ } else {
+ ret = xlator_set_fixed_option(xl, "changelog-notification", "off");
+ if (ret)
+ goto out;
+ }
out:
return ret;
}
--
1.8.3.1

594
SOURCES/0446-bitrot-Make-number-of-signer-threads-configurable.patch Normal file
View File

@ -0,0 +1,594 @@
From 866a4c49ad9c5a9125814a9f843d4c7fd967ab2b Mon Sep 17 00:00:00 2001
From: Kotresh HR <khiremat@redhat.com>
Date: Mon, 3 Feb 2020 18:10:17 +0530
Subject: [PATCH 446/449] bitrot: Make number of signer threads configurable
The number of signing process threads (glfs_brpobj)
is set to 4 by default. The recommendation is to set
it to number of cores available. This patch makes it
configurable as follows
gluster vol bitrot <volname> signer-threads <count>
> fixes: bz#1797869
> Change-Id: Ia883b3e5e34e0bc8d095243508d320c9c9c58adc
> Signed-off-by: Kotresh HR <khiremat@redhat.com>
> (Cherry pick from commit 8fad76650bd85463708f59d2518f5b764ae4c702)
> (Reviewed on upstream link https://review.gluster.org/#/c/glusterfs/+/24091/)
BUG: 1790336
Change-Id: Ia883b3e5e34e0bc8d095243508d320c9c9c58adc
Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202780
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
cli/src/cli-cmd-parser.c | 29 +++++++-
cli/src/cli-cmd-volume.c | 12 +++
doc/gluster.8 | 6 ++
libglusterfs/src/glusterfs/common-utils.h | 1 +
rpc/xdr/src/cli1-xdr.x | 1 +
tests/bitrot/br-signer-threads-config-1797869.t | 73 +++++++++++++++++++
xlators/features/bit-rot/src/bitd/bit-rot.c | 45 +++++++++---
xlators/features/bit-rot/src/bitd/bit-rot.h | 20 ++---
.../bit-rot/src/stub/bit-rot-stub-mem-types.h | 1 +
xlators/mgmt/glusterd/src/glusterd-bitrot.c | 85 ++++++++++++++++++++++
xlators/mgmt/glusterd/src/glusterd-volgen.c | 16 ++--
xlators/mgmt/glusterd/src/glusterd-volume-set.c | 9 +++
12 files changed, 270 insertions(+), 28 deletions(-)
create mode 100644 tests/bitrot/br-signer-threads-config-1797869.t
diff --git a/cli/src/cli-cmd-parser.c b/cli/src/cli-cmd-parser.c
index 7446b95..5fd05f4 100644
--- a/cli/src/cli-cmd-parser.c
+++ b/cli/src/cli-cmd-parser.c
@@ -5661,7 +5661,7 @@ cli_cmd_bitrot_parse(const char **words, int wordcount, dict_t **options)
char *volname = NULL;
char *opwords[] = {
"enable", "disable", "scrub-throttle", "scrub-frequency", "scrub",
- "signing-time", NULL};
+ "signing-time", "signer-threads", NULL};
char *scrub_throt_values[] = {"lazy", "normal", "aggressive", NULL};
char *scrub_freq_values[] = {"hourly", "daily", "weekly", "biweekly",
"monthly", "minute", NULL};
@@ -5669,6 +5669,7 @@ cli_cmd_bitrot_parse(const char **words, int wordcount, dict_t **options)
dict_t *dict = NULL;
gf_bitrot_type type = GF_BITROT_OPTION_TYPE_NONE;
int32_t expiry_time = 0;
+ int32_t signer_th_count = 0;
GF_ASSERT(words);
GF_ASSERT(options);
@@ -5849,6 +5850,31 @@ cli_cmd_bitrot_parse(const char **words, int wordcount, dict_t **options)
}
goto set_type;
}
+ } else if (!strcmp(words[3], "signer-threads")) {
+ if (!words[4]) {
+ cli_err(
+ "Missing signer-thread value for bitrot "
+ "option");
+ ret = -1;
+ goto out;
+ } else {
+ type = GF_BITROT_OPTION_TYPE_SIGNER_THREADS;
+
+ signer_th_count = strtol(words[4], NULL, 0);
+ if (signer_th_count < 1) {
+ cli_err("signer-thread count should not be less than 1");
+ ret = -1;
+ goto out;
+ }
+
+ ret = dict_set_uint32(dict, "signer-threads",
+ (unsigned int)signer_th_count);
+ if (ret) {
+ cli_out("Failed to set dict for bitrot");
+ goto out;
+ }
+ goto set_type;
+ }
} else {
cli_err(
"Invalid option %s for bitrot. Please enter valid "
@@ -5857,7 +5883,6 @@ cli_cmd_bitrot_parse(const char **words, int wordcount, dict_t **options)
ret = -1;
goto out;
}
-
set_type:
ret = dict_set_int32(dict, "type", type);
if (ret < 0)
diff --git a/cli/src/cli-cmd-volume.c b/cli/src/cli-cmd-volume.c
index f33fc99..72504ca 100644
--- a/cli/src/cli-cmd-volume.c
+++ b/cli/src/cli-cmd-volume.c
@@ -3236,6 +3236,16 @@ struct cli_cmd bitrot_cmds[] = {
{"volume bitrot <VOLNAME> {enable|disable}", NULL, /*cli_cmd_bitrot_cbk,*/
"Enable/disable bitrot for volume <VOLNAME>"},
+ {"volume bitrot <VOLNAME> signing-time <time-in-secs>",
+ NULL, /*cli_cmd_bitrot_cbk,*/
+ "Waiting time for an object after last fd is closed to start signing "
+ "process"},
+
+ {"volume bitrot <VOLNAME> signer-threads <count>",
+ NULL, /*cli_cmd_bitrot_cbk,*/
+ "Number of signing process threads. Usually set to number of available "
+ "cores"},
+
{"volume bitrot <VOLNAME> scrub-throttle {lazy|normal|aggressive}",
NULL, /*cli_cmd_bitrot_cbk,*/
"Set the speed of the scrubber for volume <VOLNAME>"},
@@ -3251,6 +3261,8 @@ struct cli_cmd bitrot_cmds[] = {
"the scrubber. ondemand starts the scrubber immediately."},
{"volume bitrot <VOLNAME> {enable|disable}\n"
+ "volume bitrot <VOLNAME> signing-time <time-in-secs>\n"
+ "volume bitrot <VOLNAME> signer-threads <count>\n"
"volume bitrot <volname> scrub-throttle {lazy|normal|aggressive}\n"
"volume bitrot <volname> scrub-frequency {hourly|daily|weekly|biweekly"
"|monthly}\n"
diff --git a/doc/gluster.8 b/doc/gluster.8
index 66bdb48..084346d 100644
--- a/doc/gluster.8
+++ b/doc/gluster.8
@@ -244,6 +244,12 @@ Use "!<OPTION>" to reset option <OPTION> to default value.
\fB\ volume bitrot <VOLNAME> {enable|disable} \fR
Enable/disable bitrot for volume <VOLNAME>
.TP
+\fB\ volume bitrot <VOLNAME> signing-time <time-in-secs> \fR
+Waiting time for an object after last fd is closed to start signing process.
+.TP
+\fB\ volume bitrot <VOLNAME> signer-threads <count> \fR
+Number of signing process threads. Usually set to number of available cores.
+.TP
\fB\ volume bitrot <VOLNAME> scrub-throttle {lazy|normal|aggressive} \fR
Scrub-throttle value is a measure of how fast or slow the scrubber scrubs the filesystem for volume <VOLNAME>
.TP
diff --git a/libglusterfs/src/glusterfs/common-utils.h b/libglusterfs/src/glusterfs/common-utils.h
index 0e2ecc7..f0a0a41 100644
--- a/libglusterfs/src/glusterfs/common-utils.h
+++ b/libglusterfs/src/glusterfs/common-utils.h
@@ -126,6 +126,7 @@ trap(void);
/* Default value of signing waiting time to sign a file for bitrot */
#define SIGNING_TIMEOUT "120"
+#define BR_WORKERS "4"
/* xxhash */
#define GF_XXH64_DIGEST_LENGTH 8
diff --git a/rpc/xdr/src/cli1-xdr.x b/rpc/xdr/src/cli1-xdr.x
index a32c864..777cb00 100644
--- a/rpc/xdr/src/cli1-xdr.x
+++ b/rpc/xdr/src/cli1-xdr.x
@@ -68,6 +68,7 @@ enum gf_bitrot_type {
GF_BITROT_OPTION_TYPE_EXPIRY_TIME,
GF_BITROT_CMD_SCRUB_STATUS,
GF_BITROT_CMD_SCRUB_ONDEMAND,
+ GF_BITROT_OPTION_TYPE_SIGNER_THREADS,
GF_BITROT_OPTION_TYPE_MAX
};
diff --git a/tests/bitrot/br-signer-threads-config-1797869.t b/tests/bitrot/br-signer-threads-config-1797869.t
new file mode 100644
index 0000000..657ef3e
--- /dev/null
+++ b/tests/bitrot/br-signer-threads-config-1797869.t
@@ -0,0 +1,73 @@
+#!/bin/bash
+
+. $(dirname $0)/../include.rc
+. $(dirname $0)/../volume.rc
+. $(dirname $0)/../cluster.rc
+
+function get_bitd_count_1 {
+ ps auxww | grep glusterfs | grep bitd.pid | grep -v grep | grep $H1 | wc -l
+}
+
+function get_bitd_count_2 {
+ ps auxww | grep glusterfs | grep bitd.pid | grep -v grep | grep $H2 | wc -l
+}
+
+function get_bitd_pid_1 {
+ ps auxww | grep glusterfs | grep bitd.pid | grep -v grep | grep $H1 | awk '{print $2}'
+}
+
+function get_bitd_pid_2 {
+ ps auxww | grep glusterfs | grep bitd.pid | grep -v grep | grep $H2 | awk '{print $2}'
+}
+
+function get_signer_th_count_1 {
+ ps -eL | grep $(get_bitd_pid_1) | grep glfs_brpobj | wc -l
+}
+
+function get_signer_th_count_2 {
+ ps -eL | grep $(get_bitd_pid_2) | grep glfs_brpobj | wc -l
+}
+
+cleanup;
+
+TEST launch_cluster 2
+
+TEST $CLI_1 peer probe $H2;
+EXPECT_WITHIN $PROBE_TIMEOUT 1 peer_count;
+
+TEST $CLI_1 volume create $V0 $H1:$B1
+TEST $CLI_1 volume create $V1 $H2:$B2
+EXPECT 'Created' volinfo_field_1 $V0 'Status';
+EXPECT 'Created' volinfo_field_1 $V1 'Status';
+
+TEST $CLI_1 volume start $V0
+TEST $CLI_1 volume start $V1
+EXPECT 'Started' volinfo_field_1 $V0 'Status';
+EXPECT 'Started' volinfo_field_1 $V1 'Status';
+
+#Enable bitrot
+TEST $CLI_1 volume bitrot $V0 enable
+TEST $CLI_1 volume bitrot $V1 enable
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" get_bitd_count_1
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" get_bitd_count_2
+
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "4" get_signer_th_count_1
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "4" get_signer_th_count_2
+
+old_bitd_pid_1=$(get_bitd_pid_1)
+old_bitd_pid_2=$(get_bitd_pid_2)
+TEST $CLI_1 volume bitrot $V0 signer-threads 1
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" get_signer_th_count_1
+EXPECT_NOT "$old_bitd_pid_1" get_bitd_pid_1;
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "4" get_signer_th_count_2
+EXPECT "$old_bitd_pid_2" get_bitd_pid_2;
+
+old_bitd_pid_1=$(get_bitd_pid_1)
+old_bitd_pid_2=$(get_bitd_pid_2)
+TEST $CLI_1 volume bitrot $V1 signer-threads 2
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "2" get_signer_th_count_2
+EXPECT_NOT "$old_bitd_pid_2" get_bitd_pid_2;
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" get_signer_th_count_1
+EXPECT "$old_bitd_pid_1" get_bitd_pid_1;
+
+cleanup;
diff --git a/xlators/features/bit-rot/src/bitd/bit-rot.c b/xlators/features/bit-rot/src/bitd/bit-rot.c
index 7b1c5dc..b8feef7 100644
--- a/xlators/features/bit-rot/src/bitd/bit-rot.c
+++ b/xlators/features/bit-rot/src/bitd/bit-rot.c
@@ -1734,22 +1734,26 @@ out:
return 0;
}
-/**
- * Initialize signer specific structures, spawn worker threads.
- */
-
static void
br_fini_signer(xlator_t *this, br_private_t *priv)
{
int i = 0;
- for (; i < BR_WORKERS; i++) {
+ if (priv == NULL)
+ return;
+
+ for (; i < priv->signer_th_count; i++) {
(void)gf_thread_cleanup_xint(priv->obj_queue->workers[i]);
}
+ GF_FREE(priv->obj_queue->workers);
pthread_cond_destroy(&priv->object_cond);
}
+/**
+ * Initialize signer specific structures, spawn worker threads.
+ */
+
static int32_t
br_init_signer(xlator_t *this, br_private_t *priv)
{
@@ -1769,7 +1773,12 @@ br_init_signer(xlator_t *this, br_private_t *priv)
goto cleanup_cond;
INIT_LIST_HEAD(&priv->obj_queue->objects);
- for (i = 0; i < BR_WORKERS; i++) {
+ priv->obj_queue->workers = GF_CALLOC(
+ priv->signer_th_count, sizeof(pthread_t), gf_br_mt_br_worker_t);
+ if (!priv->obj_queue->workers)
+ goto cleanup_obj_queue;
+
+ for (i = 0; i < priv->signer_th_count; i++) {
ret = gf_thread_create(&priv->obj_queue->workers[i], NULL,
br_process_object, this, "brpobj");
if (ret != 0) {
@@ -1787,7 +1796,9 @@ cleanup_threads:
for (i--; i >= 0; i--) {
(void)gf_thread_cleanup_xint(priv->obj_queue->workers[i]);
}
+ GF_FREE(priv->obj_queue->workers);
+cleanup_obj_queue:
GF_FREE(priv->obj_queue);
cleanup_cond:
@@ -1840,7 +1851,7 @@ br_rate_limit_signer(xlator_t *this, int child_count, int numbricks)
if (contribution == 0)
contribution = 1;
spec.rate = BR_HASH_CALC_READ_SIZE * contribution;
- spec.maxlimit = BR_WORKERS * BR_HASH_CALC_READ_SIZE;
+ spec.maxlimit = priv->signer_th_count * BR_HASH_CALC_READ_SIZE;
#endif
@@ -1860,11 +1871,16 @@ br_rate_limit_signer(xlator_t *this, int child_count, int numbricks)
static int32_t
br_signer_handle_options(xlator_t *this, br_private_t *priv, dict_t *options)
{
- if (options)
+ if (options) {
GF_OPTION_RECONF("expiry-time", priv->expiry_time, options, uint32,
error_return);
- else
+ GF_OPTION_RECONF("signer-threads", priv->signer_th_count, options,
+ uint32, error_return);
+ } else {
GF_OPTION_INIT("expiry-time", priv->expiry_time, uint32, error_return);
+ GF_OPTION_INIT("signer-threads", priv->signer_th_count, uint32,
+ error_return);
+ }
return 0;
@@ -1880,6 +1896,8 @@ br_signer_init(xlator_t *this, br_private_t *priv)
GF_OPTION_INIT("expiry-time", priv->expiry_time, uint32, error_return);
GF_OPTION_INIT("brick-count", numbricks, int32, error_return);
+ GF_OPTION_INIT("signer-threads", priv->signer_th_count, uint32,
+ error_return);
ret = br_rate_limit_signer(this, priv->child_count, numbricks);
if (ret)
@@ -2210,6 +2228,15 @@ struct volume_options options[] = {
.description = "Pause/Resume scrub. Upon resume, scrubber "
"continues from where it left off.",
},
+ {
+ .key = {"signer-threads"},
+ .type = GF_OPTION_TYPE_INT,
+ .default_value = BR_WORKERS,
+ .op_version = {GD_OP_VERSION_7_0},
+ .flags = OPT_FLAG_SETTABLE,
+ .description = "Number of signing process threads. As a best "
+ "practice, set this to the number of processor cores",
+ },
{.key = {NULL}},
};
diff --git a/xlators/features/bit-rot/src/bitd/bit-rot.h b/xlators/features/bit-rot/src/bitd/bit-rot.h
index a4d4fd7..8ac7dcd 100644
--- a/xlators/features/bit-rot/src/bitd/bit-rot.h
+++ b/xlators/features/bit-rot/src/bitd/bit-rot.h
@@ -30,12 +30,6 @@
#include <openssl/sha.h>
-/**
- * TODO: make this configurable. As a best practice, set this to the
- * number of processor cores.
- */
-#define BR_WORKERS 4
-
typedef enum scrub_throttle {
BR_SCRUB_THROTTLE_VOID = -1,
BR_SCRUB_THROTTLE_LAZY = 0,
@@ -108,12 +102,12 @@ struct br_child {
typedef struct br_child br_child_t;
struct br_obj_n_workers {
- struct list_head objects; /* queue of objects expired from the
- timer wheel and ready to be picked
- up for signing */
- pthread_t workers[BR_WORKERS]; /* Threads which pick up the objects
- from the above queue and start
- signing each object */
+ struct list_head objects; /* queue of objects expired from the
+ timer wheel and ready to be picked
+ up for signing */
+ pthread_t *workers; /* Threads which pick up the objects
+ from the above queue and start
+ signing each object */
};
struct br_scrubber {
@@ -209,6 +203,8 @@ struct br_private {
uint32_t expiry_time; /* objects "wait" time */
+ uint32_t signer_th_count; /* Number of signing process threads */
+
tbf_t *tbf; /* token bucket filter */
gf_boolean_t iamscrubber; /* function as a fs scrubber */
diff --git a/xlators/features/bit-rot/src/stub/bit-rot-stub-mem-types.h b/xlators/features/bit-rot/src/stub/bit-rot-stub-mem-types.h
index 40bcda1..9d93caf 100644
--- a/xlators/features/bit-rot/src/stub/bit-rot-stub-mem-types.h
+++ b/xlators/features/bit-rot/src/stub/bit-rot-stub-mem-types.h
@@ -29,6 +29,7 @@ enum br_mem_types {
gf_br_stub_mt_sigstub_t,
gf_br_mt_br_child_event_t,
gf_br_stub_mt_misc,
+ gf_br_mt_br_worker_t,
gf_br_stub_mt_end,
};
diff --git a/xlators/mgmt/glusterd/src/glusterd-bitrot.c b/xlators/mgmt/glusterd/src/glusterd-bitrot.c
index c653249..f79af2d 100644
--- a/xlators/mgmt/glusterd/src/glusterd-bitrot.c
+++ b/xlators/mgmt/glusterd/src/glusterd-bitrot.c
@@ -34,6 +34,7 @@ const char *gd_bitrot_op_list[GF_BITROT_OPTION_TYPE_MAX] = {
[GF_BITROT_OPTION_TYPE_SCRUB_FREQ] = "scrub-frequency",
[GF_BITROT_OPTION_TYPE_SCRUB] = "scrub",
[GF_BITROT_OPTION_TYPE_EXPIRY_TIME] = "expiry-time",
+ [GF_BITROT_OPTION_TYPE_SIGNER_THREADS] = "signer-threads",
};
int
@@ -354,6 +355,81 @@ out:
return ret;
}
+static gf_boolean_t
+is_bitd_configure_noop(xlator_t *this, glusterd_volinfo_t *volinfo)
+{
+ gf_boolean_t noop = _gf_true;
+ glusterd_brickinfo_t *brickinfo = NULL;
+
+ if (!glusterd_is_bitrot_enabled(volinfo))
+ goto out;
+ else if (volinfo->status != GLUSTERD_STATUS_STARTED)
+ goto out;
+ else {
+ cds_list_for_each_entry(brickinfo, &volinfo->bricks, brick_list)
+ {
+ if (!glusterd_is_local_brick(this, volinfo, brickinfo))
+ continue;
+ noop = _gf_false;
+ return noop;
+ }
+ }
+out:
+ return noop;
+}
+
+static int
+glusterd_bitrot_signer_threads(glusterd_volinfo_t *volinfo, dict_t *dict,
+ char *key, char **op_errstr)
+{
+ int32_t ret = -1;
+ uint32_t signer_th_count = 0;
+ uint32_t existing_th_count = 0;
+ xlator_t *this = NULL;
+ glusterd_conf_t *priv = NULL;
+ char dkey[32] = {
+ 0,
+ };
+
+ this = THIS;
+ GF_ASSERT(this);
+
+ priv = this->private;
+ GF_VALIDATE_OR_GOTO(this->name, priv, out);
+
+ ret = dict_get_uint32(dict, "signer-threads", &signer_th_count);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, errno, GD_MSG_DICT_GET_FAILED,
+ "Unable to get bitrot signer thread count.");
+ goto out;
+ }
+
+ ret = dict_get_uint32(volinfo->dict, key, &existing_th_count);
+ if (ret == 0 && signer_th_count == existing_th_count) {
+ goto out;
+ }
+
+ snprintf(dkey, sizeof(dkey), "%d", signer_th_count);
+ ret = dict_set_dynstr_with_alloc(volinfo->dict, key, dkey);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, errno, GD_MSG_DICT_SET_FAILED,
+ "Failed to set option %s", key);
+ goto out;
+ }
+
+ if (!is_bitd_configure_noop(this, volinfo)) {
+ ret = priv->bitd_svc.manager(&(priv->bitd_svc), NULL,
+ PROC_START_NO_WAIT);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_BITDSVC_RECONF_FAIL,
+ "Failed to reconfigure bitrot services");
+ goto out;
+ }
+ }
+out:
+ return ret;
+}
+
static int
glusterd_bitrot_enable(glusterd_volinfo_t *volinfo, char **op_errstr)
{
@@ -594,6 +670,15 @@ glusterd_op_bitrot(dict_t *dict, char **op_errstr, dict_t *rsp_dict)
volinfo, dict, "features.expiry-time", op_errstr);
if (ret)
goto out;
+ break;
+
+ case GF_BITROT_OPTION_TYPE_SIGNER_THREADS:
+ ret = glusterd_bitrot_signer_threads(
+ volinfo, dict, "features.signer-threads", op_errstr);
+ if (ret)
+ goto out;
+ break;
+
case GF_BITROT_CMD_SCRUB_STATUS:
case GF_BITROT_CMD_SCRUB_ONDEMAND:
break;
diff --git a/xlators/mgmt/glusterd/src/glusterd-volgen.c b/xlators/mgmt/glusterd/src/glusterd-volgen.c
index 13f84ea..094a71f 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volgen.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volgen.c
@@ -4658,6 +4658,12 @@ bitrot_option_handler(volgen_graph_t *graph, struct volopt_map_entry *vme,
return -1;
}
+ if (!strcmp(vme->option, "signer-threads")) {
+ ret = xlator_set_fixed_option(xl, "signer-threads", vme->value);
+ if (ret)
+ return -1;
+ }
+
return ret;
}
@@ -4940,18 +4946,18 @@ glusterd_prepare_shd_volume_options_for_tier(glusterd_volinfo_t *volinfo,
dict_t *set_dict)
{
int ret = -1;
- char *key = NULL;
+ char *key = NULL;
- key = volgen_get_shd_key (volinfo->tier_info.cold_type);
+ key = volgen_get_shd_key(volinfo->tier_info.cold_type);
if (key) {
- ret = dict_set_str (set_dict, key, "enable");
+ ret = dict_set_str(set_dict, key, "enable");
if (ret)
goto out;
}
- key = volgen_get_shd_key (volinfo->tier_info.hot_type);
+ key = volgen_get_shd_key(volinfo->tier_info.hot_type);
if (key) {
- ret = dict_set_str (set_dict, key, "enable");
+ ret = dict_set_str(set_dict, key, "enable");
if (ret)
goto out;
}
diff --git a/xlators/mgmt/glusterd/src/glusterd-volume-set.c b/xlators/mgmt/glusterd/src/glusterd-volume-set.c
index 9001b88..62acadf 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volume-set.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volume-set.c
@@ -3379,6 +3379,15 @@ struct volopt_map_entry glusterd_volopt_map[] = {
.op_version = GD_OP_VERSION_3_7_0,
.type = NO_DOC,
},
+ {
+ .key = "features.signer-threads",
+ .voltype = "features/bit-rot",
+ .value = BR_WORKERS,
+ .option = "signer-threads",
+ .op_version = GD_OP_VERSION_7_0,
+ .type = NO_DOC,
+ },
+ /* Upcall translator options */
/* Upcall translator options */
{
.key = "features.cache-invalidation",
--
1.8.3.1

359
SOURCES/0447-core-brick_mux-brick-crashed-when-creating-and-delet.patch Normal file
View File

@ -0,0 +1,359 @@
From 51090a4b3cb000d601083f12d1875547819fc03f Mon Sep 17 00:00:00 2001
From: Mohit Agrawal <moagrawal@redhat.com>
Date: Wed, 4 Mar 2020 09:17:26 +0530
Subject: [PATCH 447/449] core[brick_mux]: brick crashed when creating and
deleting volumes over time
Problem: In brick_mux environment, while volumes are created/stopped in a loop
after running a long time the main brick is crashed.The brick is crashed
because the main brick process was not cleaned up memory for all objects
at the time of detaching a volume.
Below are the objects that are missed at the time of detaching a volume
1) xlator object for a brick graph
2) local_pool for posix_lock xlator
3) rpc object cleanup at quota xlator
4) inode leak at brick xlator
Solution: To avoid the crash resolve all leak at the time of detaching a brick
> Change-Id: Ibb6e46c5fba22b9441a88cbaf6b3278823235913
> updates: #977
> Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
> (Cherry pick from commit e589d8de66d3325da8fbbbe44d1a5bd6335e08ab)
> (Reviewed on upstream link https://review.gluster.org/#/c/glusterfs/+/24209/)
BUG: 1790336
Change-Id: Ibb6e46c5fba22b9441a88cbaf6b3278823235913
Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202782
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Xavi Hernandez Juan <xhernandez@redhat.com>
---
libglusterfs/src/glusterfs/glusterfs.h | 1 +
libglusterfs/src/graph.c | 1 +
libglusterfs/src/graph.y | 2 +-
libglusterfs/src/xlator.c | 29 ++++++++----
xlators/features/changelog/src/changelog.c | 1 +
xlators/features/locks/src/posix.c | 4 ++
xlators/features/quota/src/quota-enforcer-client.c | 14 +++++-
xlators/features/quota/src/quota.c | 54 ++++++++++++++++++++--
xlators/features/quota/src/quota.h | 3 ++
xlators/protocol/server/src/server.c | 12 +++--
10 files changed, 103 insertions(+), 18 deletions(-)
diff --git a/libglusterfs/src/glusterfs/glusterfs.h b/libglusterfs/src/glusterfs/glusterfs.h
index 177a020..584846e 100644
--- a/libglusterfs/src/glusterfs/glusterfs.h
+++ b/libglusterfs/src/glusterfs/glusterfs.h
@@ -603,6 +603,7 @@ struct _glusterfs_graph {
int used; /* Should be set when fuse gets
first CHILD_UP */
uint32_t volfile_checksum;
+ pthread_mutex_t mutex;
};
typedef struct _glusterfs_graph glusterfs_graph_t;
diff --git a/libglusterfs/src/graph.c b/libglusterfs/src/graph.c
index bb5e67a..1cd92db 100644
--- a/libglusterfs/src/graph.c
+++ b/libglusterfs/src/graph.c
@@ -1092,6 +1092,7 @@ glusterfs_graph_destroy_residual(glusterfs_graph_t *graph)
ret = xlator_tree_free_memacct(graph->first);
list_del_init(&graph->list);
+ pthread_mutex_destroy(&graph->mutex);
GF_FREE(graph);
return ret;
diff --git a/libglusterfs/src/graph.y b/libglusterfs/src/graph.y
index 5b92985..5733515 100644
--- a/libglusterfs/src/graph.y
+++ b/libglusterfs/src/graph.y
@@ -541,7 +541,7 @@ glusterfs_graph_new ()
return NULL;
INIT_LIST_HEAD (&graph->list);
-
+ pthread_mutex_init(&graph->mutex, NULL);
gettimeofday (&graph->dob, NULL);
return graph;
diff --git a/libglusterfs/src/xlator.c b/libglusterfs/src/xlator.c
index 108b96a..36cc32c 100644
--- a/libglusterfs/src/xlator.c
+++ b/libglusterfs/src/xlator.c
@@ -938,6 +938,8 @@ xlator_mem_cleanup(xlator_t *this)
xlator_list_t **trav_p = NULL;
xlator_t *top = NULL;
xlator_t *victim = NULL;
+ glusterfs_graph_t *graph = NULL;
+ gf_boolean_t graph_cleanup = _gf_false;
if (this->call_cleanup || !this->ctx)
return;
@@ -945,6 +947,12 @@ xlator_mem_cleanup(xlator_t *this)
this->call_cleanup = 1;
ctx = this->ctx;
+ inode_table = this->itable;
+ if (inode_table) {
+ inode_table_destroy(inode_table);
+ this->itable = NULL;
+ }
+
xlator_call_fini(trav);
while (prev) {
@@ -953,12 +961,6 @@ xlator_mem_cleanup(xlator_t *this)
prev = trav;
}
- inode_table = this->itable;
- if (inode_table) {
- inode_table_destroy(inode_table);
- this->itable = NULL;
- }
-
if (this->fini) {
this->fini(this);
}
@@ -968,17 +970,28 @@ xlator_mem_cleanup(xlator_t *this)
if (ctx->active) {
top = ctx->active->first;
LOCK(&ctx->volfile_lock);
- /* TODO here we have leak for xlator node in a graph */
- /* Need to move only top xlator from a graph */
for (trav_p = &top->children; *trav_p; trav_p = &(*trav_p)->next) {
victim = (*trav_p)->xlator;
if (victim->call_cleanup && !strcmp(victim->name, this->name)) {
+ graph_cleanup = _gf_true;
(*trav_p) = (*trav_p)->next;
break;
}
}
UNLOCK(&ctx->volfile_lock);
}
+
+ if (graph_cleanup) {
+ prev = this;
+ graph = ctx->active;
+ pthread_mutex_lock(&graph->mutex);
+ while (prev) {
+ trav = prev->next;
+ GF_FREE(prev);
+ prev = trav;
+ }
+ pthread_mutex_unlock(&graph->mutex);
+ }
}
void
diff --git a/xlators/features/changelog/src/changelog.c b/xlators/features/changelog/src/changelog.c
index ff06c09..b54112c 100644
--- a/xlators/features/changelog/src/changelog.c
+++ b/xlators/features/changelog/src/changelog.c
@@ -2872,6 +2872,7 @@ fini(xlator_t *this)
if (priv->active || priv->rpc_active) {
/* terminate RPC server/threads */
changelog_cleanup_rpc(this, priv);
+ GF_FREE(priv->ev_dispatcher);
}
/* call barrier_disable to cancel timer */
if (priv->barrier_enabled)
diff --git a/xlators/features/locks/src/posix.c b/xlators/features/locks/src/posix.c
index 9a14c64..50f1265 100644
--- a/xlators/features/locks/src/posix.c
+++ b/xlators/features/locks/src/posix.c
@@ -4102,6 +4102,10 @@ fini(xlator_t *this)
if (!priv)
return;
this->private = NULL;
+ if (this->local_pool) {
+ mem_pool_destroy(this->local_pool);
+ this->local_pool = NULL;
+ }
GF_FREE(priv->brickname);
GF_FREE(priv);
diff --git a/xlators/features/quota/src/quota-enforcer-client.c b/xlators/features/quota/src/quota-enforcer-client.c
index 1a4c2e3..097439d 100644
--- a/xlators/features/quota/src/quota-enforcer-client.c
+++ b/xlators/features/quota/src/quota-enforcer-client.c
@@ -362,16 +362,28 @@ quota_enforcer_notify(struct rpc_clnt *rpc, void *mydata,
{
xlator_t *this = NULL;
int ret = 0;
+ quota_priv_t *priv = NULL;
this = mydata;
-
+ priv = this->private;
switch (event) {
case RPC_CLNT_CONNECT: {
+ pthread_mutex_lock(&priv->conn_mutex);
+ {
+ priv->conn_status = _gf_true;
+ }
+ pthread_mutex_unlock(&priv->conn_mutex);
gf_msg_trace(this->name, 0, "got RPC_CLNT_CONNECT");
break;
}
case RPC_CLNT_DISCONNECT: {
+ pthread_mutex_lock(&priv->conn_mutex);
+ {
+ priv->conn_status = _gf_false;
+ pthread_cond_signal(&priv->conn_cond);
+ }
+ pthread_mutex_unlock(&priv->conn_mutex);
gf_msg_trace(this->name, 0, "got RPC_CLNT_DISCONNECT");
break;
}
diff --git a/xlators/features/quota/src/quota.c b/xlators/features/quota/src/quota.c
index a0c236d..d1123ce 100644
--- a/xlators/features/quota/src/quota.c
+++ b/xlators/features/quota/src/quota.c
@@ -5014,6 +5014,43 @@ quota_forget(xlator_t *this, inode_t *inode)
return 0;
}
+int
+notify(xlator_t *this, int event, void *data, ...)
+{
+ quota_priv_t *priv = NULL;
+ int ret = 0;
+ rpc_clnt_t *rpc = NULL;
+ gf_boolean_t conn_status = _gf_true;
+ xlator_t *victim = data;
+
+ priv = this->private;
+ if (!priv || !priv->is_quota_on)
+ goto out;
+
+ if (event == GF_EVENT_PARENT_DOWN) {
+ rpc = priv->rpc_clnt;
+ if (rpc) {
+ rpc_clnt_disable(rpc);
+ pthread_mutex_lock(&priv->conn_mutex);
+ {
+ conn_status = priv->conn_status;
+ while (conn_status) {
+ (void)pthread_cond_wait(&priv->conn_cond,
+ &priv->conn_mutex);
+ conn_status = priv->conn_status;
+ }
+ }
+ pthread_mutex_unlock(&priv->conn_mutex);
+ gf_log(this->name, GF_LOG_INFO,
+ "Notify GF_EVENT_PARENT_DOWN for brick %s", victim->name);
+ }
+ }
+
+out:
+ ret = default_notify(this, event, data);
+ return ret;
+}
+
int32_t
init(xlator_t *this)
{
@@ -5056,6 +5093,10 @@ init(xlator_t *this)
goto err;
}
+ pthread_mutex_init(&priv->conn_mutex, NULL);
+ pthread_cond_init(&priv->conn_cond, NULL);
+ priv->conn_status = _gf_false;
+
if (priv->is_quota_on) {
rpc = quota_enforcer_init(this, this->options);
if (rpc == NULL) {
@@ -5169,20 +5210,22 @@ fini(xlator_t *this)
{
quota_priv_t *priv = NULL;
rpc_clnt_t *rpc = NULL;
- int i = 0, cnt = 0;
priv = this->private;
if (!priv)
return;
rpc = priv->rpc_clnt;
priv->rpc_clnt = NULL;
- this->private = NULL;
if (rpc) {
- cnt = GF_ATOMIC_GET(rpc->refcount);
- for (i = 0; i < cnt; i++)
- rpc_clnt_unref(rpc);
+ rpc_clnt_connection_cleanup(&rpc->conn);
+ rpc_clnt_unref(rpc);
}
+
+ this->private = NULL;
LOCK_DESTROY(&priv->lock);
+ pthread_mutex_destroy(&priv->conn_mutex);
+ pthread_cond_destroy(&priv->conn_cond);
+
GF_FREE(priv);
if (this->local_pool) {
mem_pool_destroy(this->local_pool);
@@ -5314,6 +5357,7 @@ struct volume_options options[] = {
xlator_api_t xlator_api = {
.init = init,
.fini = fini,
+ .notify = notify,
.reconfigure = reconfigure,
.mem_acct_init = mem_acct_init,
.op_version = {1}, /* Present from the initial version */
diff --git a/xlators/features/quota/src/quota.h b/xlators/features/quota/src/quota.h
index a5a99ca..e51ffd4 100644
--- a/xlators/features/quota/src/quota.h
+++ b/xlators/features/quota/src/quota.h
@@ -217,6 +217,9 @@ struct quota_priv {
char *volume_uuid;
uint64_t validation_count;
int32_t quotad_conn_status;
+ pthread_mutex_t conn_mutex;
+ pthread_cond_t conn_cond;
+ gf_boolean_t conn_status;
};
typedef struct quota_priv quota_priv_t;
diff --git a/xlators/protocol/server/src/server.c b/xlators/protocol/server/src/server.c
index a5f09fe..54d9c0f 100644
--- a/xlators/protocol/server/src/server.c
+++ b/xlators/protocol/server/src/server.c
@@ -409,7 +409,13 @@ server_call_xlator_mem_cleanup(xlator_t *this, char *victim_name)
arg = calloc(1, sizeof(*arg));
arg->this = this;
- arg->victim_name = gf_strdup(victim_name);
+ arg->victim_name = strdup(victim_name);
+ if (!arg->victim_name) {
+ gf_smsg(this->name, GF_LOG_CRITICAL, ENOMEM, LG_MSG_NO_MEMORY,
+ "Memory allocation is failed");
+ return;
+ }
+
th_ret = gf_thread_create_detached(&th_id, server_graph_janitor_threads,
arg, "graphjanitor");
if (th_ret) {
@@ -417,7 +423,7 @@ server_call_xlator_mem_cleanup(xlator_t *this, char *victim_name)
"graph janitor Thread"
" creation is failed for brick %s",
victim_name);
- GF_FREE(arg->victim_name);
+ free(arg->victim_name);
free(arg);
}
}
@@ -628,7 +634,7 @@ server_graph_janitor_threads(void *data)
}
out:
- GF_FREE(arg->victim_name);
+ free(arg->victim_name);
free(arg);
return NULL;
}
--
1.8.3.1

341
SOURCES/0448-Posix-Use-simple-approach-to-close-fd.patch Normal file
View File

@ -0,0 +1,341 @@
From 175c99dccc47d2b4267a8819404e5cbeb8cfba11 Mon Sep 17 00:00:00 2001
From: Mohit Agrawal <moagrawal@redhat.com>
Date: Thu, 12 Mar 2020 21:12:13 +0530
Subject: [PATCH 448/449] Posix: Use simple approach to close fd
Problem: posix_release(dir) functions add the fd's into a ctx->janitor_fds
and janitor thread closes the fd's.In brick_mux environment it is
difficult to handle race condition in janitor threads because brick
spawns a single janitor thread for all bricks.
Solution: Use synctask to execute posix_release(dir) functions instead of
using background a thread to close fds.
> Credits: Pranith Karampuri <pkarampu@redhat.com>
> Change-Id: Iffb031f0695a7da83d5a2f6bac8863dad225317e
> Fixes: bz#1811631
> Signed-off-by: Mohit Agrawal <moagrawal@redhat.com>
> (Cherry pick from commit fb20713b380e1df8d7f9e9df96563be2f9144fd6)
> (Reviewed on upstream link https://review.gluster.org/#/c/glusterfs/+/24221/)
BUG: 1790336
Change-Id: Iffb031f0695a7da83d5a2f6bac8863dad225317e
Signed-off-by: Mohit Agrawal <moagrawal@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202791
Tested-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
libglusterfs/src/glusterfs/glusterfs.h | 6 +-
libglusterfs/src/glusterfs/syncop.h | 7 +-
rpc/rpc-lib/src/rpcsvc.c | 6 ++
run-tests.sh | 2 +-
tests/features/ssl-authz.t | 7 +-
xlators/storage/posix/src/posix-common.c | 4 --
xlators/storage/posix/src/posix-helpers.c | 98 --------------------------
xlators/storage/posix/src/posix-inode-fd-ops.c | 28 ++------
xlators/storage/posix/src/posix.h | 3 -
9 files changed, 20 insertions(+), 141 deletions(-)
diff --git a/libglusterfs/src/glusterfs/glusterfs.h b/libglusterfs/src/glusterfs/glusterfs.h
index 584846e..495a4d7 100644
--- a/libglusterfs/src/glusterfs/glusterfs.h
+++ b/libglusterfs/src/glusterfs/glusterfs.h
@@ -734,11 +734,7 @@ struct _glusterfs_ctx {
struct list_head volfile_list;
- /* Add members to manage janitor threads for cleanup fd */
- struct list_head janitor_fds;
- pthread_cond_t janitor_cond;
- pthread_mutex_t janitor_lock;
- pthread_t janitor;
+ char volume_id[GF_UUID_BUF_SIZE]; /* Used only in protocol/client */
};
typedef struct _glusterfs_ctx glusterfs_ctx_t;
diff --git a/libglusterfs/src/glusterfs/syncop.h b/libglusterfs/src/glusterfs/syncop.h
index 3011b4c..1e4c73b 100644
--- a/libglusterfs/src/glusterfs/syncop.h
+++ b/libglusterfs/src/glusterfs/syncop.h
@@ -254,7 +254,7 @@ struct syncopctx {
task = synctask_get(); \
stb->task = task; \
if (task) \
- frame = task->opframe; \
+ frame = copy_frame(task->opframe); \
else \
frame = syncop_create_frame(THIS); \
\
@@ -269,10 +269,7 @@ struct syncopctx {
STACK_WIND_COOKIE(frame, cbk, (void *)stb, subvol, fn_op, params); \
\
__yield(stb); \
- if (task) \
- STACK_RESET(frame->root); \
- else \
- STACK_DESTROY(frame->root); \
+ STACK_DESTROY(frame->root); \
} while (0)
/*
diff --git a/rpc/rpc-lib/src/rpcsvc.c b/rpc/rpc-lib/src/rpcsvc.c
index 3f184bf..23ca1fd 100644
--- a/rpc/rpc-lib/src/rpcsvc.c
+++ b/rpc/rpc-lib/src/rpcsvc.c
@@ -375,6 +375,12 @@ rpcsvc_program_actor(rpcsvc_request_t *req)
req->ownthread = program->ownthread;
req->synctask = program->synctask;
+ if (((req->procnum == GFS3_OP_RELEASE) ||
+ (req->procnum == GFS3_OP_RELEASEDIR)) &&
+ (program->prognum == GLUSTER_FOP_PROGRAM)) {
+ req->ownthread = _gf_false;
+ req->synctask = _gf_true;
+ }
err = SUCCESS;
gf_log(GF_RPCSVC, GF_LOG_TRACE, "Actor found: %s - %s for %s",
diff --git a/run-tests.sh b/run-tests.sh
index 5683b21..c835d93 100755
--- a/run-tests.sh
+++ b/run-tests.sh
@@ -356,7 +356,7 @@ function run_tests()
selected_tests=$((selected_tests+1))
echo
echo $section_separator$section_separator
- if [[ $(get_test_status $t) == "BAD_TEST" ]] && \
+ if [[ $(get_test_status $t) =~ "BAD_TEST" ]] && \
[[ $skip_bad_tests == "yes" ]]
then
skipped_bad_tests=$((skipped_bad_tests+1))
diff --git a/tests/features/ssl-authz.t b/tests/features/ssl-authz.t
index 132b598..497083e 100755
--- a/tests/features/ssl-authz.t
+++ b/tests/features/ssl-authz.t
@@ -67,13 +67,14 @@ echo "Memory consumption for glusterfsd process"
for i in $(seq 1 100); do
gluster v heal $V0 info >/dev/null
done
-
+#Wait to cleanup memory
+sleep 10
end=`pmap -x $glusterfsd_pid | grep total | awk -F " " '{print $4}'`
diff=$((end-start))
-# If memory consumption is more than 5M some leak in SSL code path
+# If memory consumption is more than 15M some leak in SSL code path
-TEST [ $diff -lt 5000 ]
+TEST [ $diff -lt 15000 ]
# Set ssl-allow to a wildcard that includes our identity.
diff --git a/xlators/storage/posix/src/posix-common.c b/xlators/storage/posix/src/posix-common.c
index 2cb58ba..ac53796 100644
--- a/xlators/storage/posix/src/posix-common.c
+++ b/xlators/storage/posix/src/posix-common.c
@@ -1041,10 +1041,6 @@ posix_init(xlator_t *this)
pthread_mutex_init(&_private->janitor_mutex, NULL);
pthread_cond_init(&_private->janitor_cond, NULL);
INIT_LIST_HEAD(&_private->fsyncs);
- ret = posix_spawn_ctx_janitor_thread(this);
- if (ret)
- goto out;
-
ret = gf_thread_create(&_private->fsyncer, NULL, posix_fsyncer, this,
"posixfsy");
if (ret) {
diff --git a/xlators/storage/posix/src/posix-helpers.c b/xlators/storage/posix/src/posix-helpers.c
index 2336add..39dbcce 100644
--- a/xlators/storage/posix/src/posix-helpers.c
+++ b/xlators/storage/posix/src/posix-helpers.c
@@ -1582,104 +1582,6 @@ unlock:
return;
}
-static struct posix_fd *
-janitor_get_next_fd(glusterfs_ctx_t *ctx, int32_t janitor_sleep)
-{
- struct posix_fd *pfd = NULL;
-
- struct timespec timeout;
-
- pthread_mutex_lock(&ctx->janitor_lock);
- {
- if (list_empty(&ctx->janitor_fds)) {
- time(&timeout.tv_sec);
- timeout.tv_sec += janitor_sleep;
- timeout.tv_nsec = 0;
-
- pthread_cond_timedwait(&ctx->janitor_cond, &ctx->janitor_lock,
- &timeout);
- goto unlock;
- }
-
- pfd = list_entry(ctx->janitor_fds.next, struct posix_fd, list);
-
- list_del(ctx->janitor_fds.next);
- }
-unlock:
- pthread_mutex_unlock(&ctx->janitor_lock);
-
- return pfd;
-}
-
-static void *
-posix_ctx_janitor_thread_proc(void *data)
-{
- xlator_t *this = NULL;
- struct posix_fd *pfd;
- glusterfs_ctx_t *ctx = NULL;
- struct posix_private *priv = NULL;
- int32_t sleep_duration = 0;
-
- this = data;
- ctx = THIS->ctx;
- THIS = this;
-
- priv = this->private;
- sleep_duration = priv->janitor_sleep_duration;
- while (1) {
- pfd = janitor_get_next_fd(ctx, sleep_duration);
- if (pfd) {
- if (pfd->dir == NULL) {
- gf_msg_trace(this->name, 0, "janitor: closing file fd=%d",
- pfd->fd);
- sys_close(pfd->fd);
- } else {
- gf_msg_debug(this->name, 0, "janitor: closing dir fd=%p",
- pfd->dir);
- sys_closedir(pfd->dir);
- }
-
- GF_FREE(pfd);
- }
- }
-
- return NULL;
-}
-
-int
-posix_spawn_ctx_janitor_thread(xlator_t *this)
-{
- struct posix_private *priv = NULL;
- int ret = 0;
- glusterfs_ctx_t *ctx = NULL;
-
- priv = this->private;
- ctx = THIS->ctx;
-
- LOCK(&priv->lock);
- {
- if (!ctx->janitor) {
- pthread_mutex_init(&ctx->janitor_lock, NULL);
- pthread_cond_init(&ctx->janitor_cond, NULL);
- INIT_LIST_HEAD(&ctx->janitor_fds);
-
- ret = gf_thread_create(&ctx->janitor, NULL,
- posix_ctx_janitor_thread_proc, this,
- "posixctxjan");
-
- if (ret) {
- gf_msg(this->name, GF_LOG_ERROR, errno, P_MSG_THREAD_FAILED,
- "spawning janitor "
- "thread failed");
- goto unlock;
- }
- }
- }
-unlock:
- UNLOCK(&priv->lock);
- return ret;
-}
-
static int
is_fresh_file(int64_t ctime_sec)
{
diff --git a/xlators/storage/posix/src/posix-inode-fd-ops.c b/xlators/storage/posix/src/posix-inode-fd-ops.c
index 5748b9f..d135d8b 100644
--- a/xlators/storage/posix/src/posix-inode-fd-ops.c
+++ b/xlators/storage/posix/src/posix-inode-fd-ops.c
@@ -1358,7 +1358,6 @@ posix_releasedir(xlator_t *this, fd_t *fd)
struct posix_fd *pfd = NULL;
uint64_t tmp_pfd = 0;
int ret = 0;
- glusterfs_ctx_t *ctx = NULL;
VALIDATE_OR_GOTO(this, out);
VALIDATE_OR_GOTO(fd, out);
@@ -1376,21 +1375,11 @@ posix_releasedir(xlator_t *this, fd_t *fd)
goto out;
}
- ctx = THIS->ctx;
-
- pthread_mutex_lock(&ctx->janitor_lock);
- {
- INIT_LIST_HEAD(&pfd->list);
- list_add_tail(&pfd->list, &ctx->janitor_fds);
- pthread_cond_signal(&ctx->janitor_cond);
- }
- pthread_mutex_unlock(&ctx->janitor_lock);
-
- /*gf_msg_debug(this->name, 0, "janitor: closing dir fd=%p", pfd->dir);
+ gf_msg_debug(this->name, 0, "janitor: closing dir fd=%p", pfd->dir);
sys_closedir(pfd->dir);
GF_FREE(pfd);
- */
+
out:
return 0;
}
@@ -2510,13 +2499,11 @@ posix_release(xlator_t *this, fd_t *fd)
struct posix_fd *pfd = NULL;
int ret = -1;
uint64_t tmp_pfd = 0;
- glusterfs_ctx_t *ctx = NULL;
VALIDATE_OR_GOTO(this, out);
VALIDATE_OR_GOTO(fd, out);
priv = this->private;
- ctx = THIS->ctx;
ret = fd_ctx_del(fd, this, &tmp_pfd);
if (ret < 0) {
@@ -2531,13 +2518,10 @@ posix_release(xlator_t *this, fd_t *fd)
"pfd->dir is %p (not NULL) for file fd=%p", pfd->dir, fd);
}
- pthread_mutex_lock(&ctx->janitor_lock);
- {
- INIT_LIST_HEAD(&pfd->list);
- list_add_tail(&pfd->list, &ctx->janitor_fds);
- pthread_cond_signal(&ctx->janitor_cond);
- }
- pthread_mutex_unlock(&ctx->janitor_lock);
+ gf_msg_debug(this->name, 0, "janitor: closing dir fd=%p", pfd->dir);
+
+ sys_close(pfd->fd);
+ GF_FREE(pfd);
if (!priv)
goto out;
diff --git a/xlators/storage/posix/src/posix.h b/xlators/storage/posix/src/posix.h
index ac9d83c..61495a7 100644
--- a/xlators/storage/posix/src/posix.h
+++ b/xlators/storage/posix/src/posix.h
@@ -666,9 +666,6 @@ posix_cs_maintenance(xlator_t *this, fd_t *fd, loc_t *loc, int *pfd,
int
posix_check_dev_file(xlator_t *this, inode_t *inode, char *fop, int *op_errno);
-int
-posix_spawn_ctx_janitor_thread(xlator_t *this);
-
void
posix_update_iatt_buf(struct iatt *buf, int fd, char *loc, dict_t *xdata);
--
1.8.3.1

107
SOURCES/0449-test-Test-case-brick-mux-validation-in-cluster.t-is-.patch Normal file
View File

@ -0,0 +1,107 @@
From 6e15fca1621b06270983f57ac146f0f8e52f0797 Mon Sep 17 00:00:00 2001
From: Mohit Agrawal <moagrawal@redhat.com>
Date: Tue, 9 Jun 2020 15:38:12 +0530
Subject: [PATCH 449/449] test: Test case brick-mux-validation-in-cluster.t is
failing on RHEL-8
Brick process are not properly attached on any cluster node while
some volume options are changed on peer node and glusterd is down on
that specific node.
Solution: At the time of restart glusterd it got a friend update request
from a peer node if peer node having some changes on volume.If the brick
process is started before received a friend update request in that case
brick_mux behavior is not workingproperly. All bricks are attached to
the same process even volumes options are not the same. To avoid the
issue introduce an atomic flag volpeerupdate and update the value while
glusterd has received a friend update request from peer for a specific
volume.If volpeerupdate flag is 1 volume is started by
glusterd_import_friend_volume synctask
> Change-Id: I4c026f1e7807ded249153670e6967a2be8d22cb7
> Credit: Sanju Rakaonde <srakonde@redhat.com>
> fixes: #1290
> Signed-off-by: Mohit Agrawal <moagrawal@redhat.com>
> (Reviewed on upstream link https://review.gluster.org/#/c/glusterfs/+/24540/)
> (Cherry pick from commit 955bfd567329cf7fe63e9c3b89d333a55e5e9a20)
BUG: 1844359
Change-Id: I4c026f1e7807ded249153670e6967a2be8d22cb7
Signed-off-by: Mohit Agrawal <moagrawal@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202812
Tested-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-by: Sanju Rakonde <srakonde@redhat.com>
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/bugs/glusterd/brick-mux-validation-in-cluster.t | 4 +---
xlators/mgmt/glusterd/src/glusterd-utils.c | 7 +++++--
xlators/mgmt/glusterd/src/glusterd.h | 4 ++++
3 files changed, 10 insertions(+), 5 deletions(-)
diff --git a/tests/bugs/glusterd/brick-mux-validation-in-cluster.t b/tests/bugs/glusterd/brick-mux-validation-in-cluster.t
index f088dbb..b6af487 100644
--- a/tests/bugs/glusterd/brick-mux-validation-in-cluster.t
+++ b/tests/bugs/glusterd/brick-mux-validation-in-cluster.t
@@ -100,10 +100,8 @@ $CLI_2 volume set $V0 performance.readdir-ahead on
$CLI_2 volume set $V1 performance.readdir-ahead on
TEST $glusterd_1;
+EXPECT_WITHIN $PROBE_TIMEOUT 2 peer_count
-sleep 10
-
-EXPECT 4 count_brick_processes
EXPECT_WITHIN $PROCESS_UP_TIMEOUT 4 count_brick_pids
EXPECT_WITHIN $PROCESS_UP_TIMEOUT 0 count_N/A_brick_pids
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index 2eb2a76..6f904ae 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -3758,6 +3758,7 @@ glusterd_compare_friend_volume(dict_t *peer_data, int32_t count,
"Version of volume %s differ. local version = %d, "
"remote version = %d on peer %s",
volinfo->volname, volinfo->version, version, hostname);
+ GF_ATOMIC_INIT(volinfo->volpeerupdate, 1);
*status = GLUSTERD_VOL_COMP_UPDATE_REQ;
goto out;
} else if (version < volinfo->version) {
@@ -4784,7 +4785,8 @@ glusterd_volinfo_stop_stale_bricks(glusterd_volinfo_t *new_volinfo,
* or if it's part of the new volume and is pending a snap,
* then stop the brick process
*/
- if (ret || (new_brickinfo->snap_status == -1)) {
+ if (ret || (new_brickinfo->snap_status == -1) ||
+ GF_ATOMIC_GET(old_volinfo->volpeerupdate)) {
/*TODO: may need to switch to 'atomic' flavour of
* brick_stop, once we make peer rpc program also
* synctask enabled*/
@@ -6490,7 +6492,8 @@ glusterd_brick_start(glusterd_volinfo_t *volinfo,
* three different triggers for an attempt to start the brick process
* due to the quorum handling code in glusterd_friend_sm.
*/
- if (brickinfo->status == GF_BRICK_STARTING || brickinfo->start_triggered) {
+ if (brickinfo->status == GF_BRICK_STARTING || brickinfo->start_triggered ||
+ GF_ATOMIC_GET(volinfo->volpeerupdate)) {
gf_msg_debug(this->name, 0,
"brick %s is already in starting "
"phase",
diff --git a/xlators/mgmt/glusterd/src/glusterd.h b/xlators/mgmt/glusterd/src/glusterd.h
index 1c6c3b1..f739b5d 100644
--- a/xlators/mgmt/glusterd/src/glusterd.h
+++ b/xlators/mgmt/glusterd/src/glusterd.h
@@ -523,6 +523,10 @@ struct glusterd_volinfo_ {
pthread_mutex_t store_volinfo_lock; /* acquire lock for
* updating the volinfo
*/
+ gf_atomic_t volpeerupdate;
+ /* Flag to check about volume has received updates
+ from peer
+ */
};
typedef enum gd_snap_status_ {
--
1.8.3.1

35
SOURCES/0450-tests-basic-ctime-enable-ctime-before-testing.patch Normal file
View File

@ -0,0 +1,35 @@
From 09dce9ce8e946a86209b6f057bf14323036fa12a Mon Sep 17 00:00:00 2001
From: Shwetha K Acharya <sacharya@redhat.com>
Date: Wed, 10 Jun 2020 11:44:56 +0530
Subject: [PATCH 450/451] tests/basic/ctime: enable ctime before testing
This is to ensure that this test successfully runs, even if
ctime is disabled by default (which is the case in downstream.)
Label: DOWNSTREAM ONLY
BUG: 1844359
Change-Id: I91e80b3d8a56fc089aeb58b0254812111d394842
Signed-off-by: Shwetha K Acharya <sacharya@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/202874
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/basic/ctime/ctime-utimesat.t | 1 +
1 file changed, 1 insertion(+)
diff --git a/tests/basic/ctime/ctime-utimesat.t b/tests/basic/ctime/ctime-utimesat.t
index 540e57a..da12fbe 100644
--- a/tests/basic/ctime/ctime-utimesat.t
+++ b/tests/basic/ctime/ctime-utimesat.t
@@ -14,6 +14,7 @@ TEST $CLI volume set $V0 performance.read-after-open off
TEST $CLI volume set $V0 performance.open-behind off
TEST $CLI volume set $V0 performance.write-behind off
TEST $CLI volume set $V0 performance.io-cache off
+TEST $CLI volume set $V0 ctime on
TEST $CLI volume start $V0
--
1.8.3.1

8
SOURCES/0384-extras-Modify-group-virt-to-include-network-related-.patch → SOURCES/0451-extras-Modify-group-virt-to-include-network-related-.patch
View File

@ -1,7 +1,7 @@
From 2948ee521316d40384130138233178ba940b175f Mon Sep 17 00:00:00 2001
From 96d9b659fd0367abe1666a5ac6203208e0dc056d Mon Sep 17 00:00:00 2001
From: Krutika Dhananjay <kdhananj@redhat.com>
Date: Mon, 4 May 2020 14:30:57 +0530
Subject: [PATCH 384/392] extras: Modify group 'virt' to include
Subject: [PATCH 451/451] extras: Modify group 'virt' to include
network-related options
This is needed to work around an issue seen where vms running on
@ -17,9 +17,9 @@ Upstream patch:
> Signed-off-by: Krutika Dhananjay <kdhananj@redhat.com>
Change-Id: Ic25b5f50144ad42458e5c847e1e7e191032396c1
BUG: 1848899
BUG: 1845064
Signed-off-by: Krutika Dhananjay <kdhananj@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/203685
Reviewed-on: https://code.engineering.redhat.com/gerrit/203291
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---

48
SOURCES/0452-Tier-DHT-Handle-the-pause-case-missed-out.patch Normal file
View File

@ -0,0 +1,48 @@
From c184943bdf38de5b4cbf165fd1cd98ce7bd9e976 Mon Sep 17 00:00:00 2001
From: hari gowtham <hgowtham@redhat.com>
Date: Tue, 16 Jun 2020 14:47:53 +0530
Subject: [PATCH 452/456] Tier/DHT: Handle the pause case missed out
Problem: While backporting a change from master
the changes related to tier were removed. This started affecting
the tier pause functionality. Backporting it
to downstream left this usecase messed up as we still support tier.
patch that caused this: https://code.engineering.redhat.com/gerrit/#/c/202647/2
Fix: add the condition back for tier pause to work.
Label: DOWNSTREAM ONLY
BUG: 1844359
Change-Id: I46c6c179b09c7e1a729be9fd257fa4a490f0287e
Signed-off-by: hari gowtham <hgowtham@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/203560
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/cluster/dht/src/dht-rebalance.c | 9 +++++++++
1 file changed, 9 insertions(+)
diff --git a/xlators/cluster/dht/src/dht-rebalance.c b/xlators/cluster/dht/src/dht-rebalance.c
index e9974cd..abc10fc 100644
--- a/xlators/cluster/dht/src/dht-rebalance.c
+++ b/xlators/cluster/dht/src/dht-rebalance.c
@@ -1160,6 +1160,15 @@ __dht_rebalance_migrate_data(xlator_t *this, gf_defrag_info_t *defrag,
break;
}
+ if ((defrag && defrag->cmd == GF_DEFRAG_CMD_START_TIER) &&
+ (gf_defrag_get_pause_state(&defrag->tier_conf) != TIER_RUNNING)) {
+ gf_msg("tier", GF_LOG_INFO, 0, DHT_MSG_TIER_PAUSED,
+ "Migrate file paused");
+ ret = -1;
+ break;
+ }
+
+
offset += ret;
total += ret;
--
1.8.3.1

300
SOURCES/0453-glusterd-add-brick-command-failure.patch Normal file
View File

@ -0,0 +1,300 @@
From a04592cce9aaa6ccb8a038bc3b4e31bc125d1d10 Mon Sep 17 00:00:00 2001
From: Sanju Rakonde <srakonde@redhat.com>
Date: Tue, 16 Jun 2020 18:03:21 +0530
Subject: [PATCH 453/456] glusterd: add-brick command failure
Problem: add-brick operation is failing when replica or disperse
count is not mentioned in the add-brick command.
Reason: with commit a113d93 we are checking brick order while
doing add-brick operation for replica and disperse volumes. If
replica count or disperse count is not mentioned in the command,
the dict get is failing and resulting add-brick operation failure.
> upstream patch: https://review.gluster.org/#/c/glusterfs/+/24581/
> fixes: #1306
> Change-Id: Ie957540e303bfb5f2d69015661a60d7e72557353
> Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
BUG: 1847081
Change-Id: Ie957540e303bfb5f2d69015661a60d7e72557353
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/203867
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/bugs/glusterd/brick-order-check-add-brick.t | 40 ++++++++++++++++++++++
tests/cluster.rc | 11 ++++--
xlators/mgmt/glusterd/src/glusterd-brick-ops.c | 39 ++++++++++++++-------
xlators/mgmt/glusterd/src/glusterd-utils.c | 30 ++---------------
xlators/mgmt/glusterd/src/glusterd-utils.h | 3 +-
xlators/mgmt/glusterd/src/glusterd-volume-ops.c | 41 +++++++++++++++++++----
6 files changed, 115 insertions(+), 49 deletions(-)
create mode 100644 tests/bugs/glusterd/brick-order-check-add-brick.t
diff --git a/tests/bugs/glusterd/brick-order-check-add-brick.t b/tests/bugs/glusterd/brick-order-check-add-brick.t
new file mode 100644
index 0000000..29f0ed1
--- /dev/null
+++ b/tests/bugs/glusterd/brick-order-check-add-brick.t
@@ -0,0 +1,40 @@
+#!/bin/bash
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../cluster.rc
+. $(dirname $0)/../../snapshot.rc
+
+cleanup;
+
+TEST verify_lvm_version;
+#Create cluster with 3 nodes
+TEST launch_cluster 3 -NO_DEBUG -NO_FORCE
+TEST setup_lvm 3
+
+TEST $CLI_1 peer probe $H2
+TEST $CLI_1 peer probe $H3
+EXPECT_WITHIN $PROBE_TIMEOUT 2 peer_count
+
+TEST $CLI_1 volume create $V0 replica 3 $H1:$L1/$V0 $H2:$L2/$V0 $H3:$L3/$V0
+EXPECT '1 x 3 = 3' volinfo_field $V0 'Number of Bricks'
+EXPECT 'Created' volinfo_field $V0 'Status'
+
+TEST $CLI_1 volume start $V0
+EXPECT 'Started' volinfo_field $V0 'Status'
+
+#add-brick with or without mentioning the replica count should not fail
+TEST $CLI_1 volume add-brick $V0 replica 3 $H1:$L1/${V0}_1 $H2:$L2/${V0}_1 $H3:$L3/${V0}_1
+EXPECT '2 x 3 = 6' volinfo_field $V0 'Number of Bricks'
+
+TEST $CLI_1 volume add-brick $V0 $H1:$L1/${V0}_2 $H2:$L2/${V0}_2 $H3:$L3/${V0}_2
+EXPECT '3 x 3 = 9' volinfo_field $V0 'Number of Bricks'
+
+#adding bricks from same host should fail the brick order check
+TEST ! $CLI_1 volume add-brick $V0 $H1:$L1/${V0}_3 $H1:$L1/${V0}_4 $H1:$L1/${V0}_5
+EXPECT '3 x 3 = 9' volinfo_field $V0 'Number of Bricks'
+
+#adding bricks from same host with force should succeed
+TEST $CLI_1 volume add-brick $V0 $H1:$L1/${V0}_3 $H1:$L1/${V0}_4 $H1:$L1/${V0}_5 force
+EXPECT '4 x 3 = 12' volinfo_field $V0 'Number of Bricks'
+
+cleanup
diff --git a/tests/cluster.rc b/tests/cluster.rc
index 99be8e7..8b73153 100644
--- a/tests/cluster.rc
+++ b/tests/cluster.rc
@@ -11,7 +11,7 @@ function launch_cluster() {
define_backends $count;
define_hosts $count;
define_glusterds $count $2;
- define_clis $count;
+ define_clis $count $3;
start_glusterds;
}
@@ -133,8 +133,13 @@ function define_clis() {
lopt1="--log-file=$logdir/$logfile1"
- eval "CLI_$i='$CLI --glusterd-sock=${!b}/glusterd/gd.sock $lopt'";
- eval "CLI$i='$CLI --glusterd-sock=${!b}/glusterd/gd.sock $lopt1'";
+ if [ "$2" == "-NO_FORCE" ]; then
+ eval "CLI_$i='$CLI_NO_FORCE --glusterd-sock=${!b}/glusterd/gd.sock $lopt'";
+ eval "CLI$i='$CLI_NO_FORCE --glusterd-sock=${!b}/glusterd/gd.sock $lopt1'";
+ else
+ eval "CLI_$i='$CLI --glusterd-sock=${!b}/glusterd/gd.sock $lopt'";
+ eval "CLI$i='$CLI --glusterd-sock=${!b}/glusterd/gd.sock $lopt1'";
+ fi
done
}
diff --git a/xlators/mgmt/glusterd/src/glusterd-brick-ops.c b/xlators/mgmt/glusterd/src/glusterd-brick-ops.c
index 121346c..5ae577a 100644
--- a/xlators/mgmt/glusterd/src/glusterd-brick-ops.c
+++ b/xlators/mgmt/glusterd/src/glusterd-brick-ops.c
@@ -1576,20 +1576,35 @@ glusterd_op_stage_add_brick(dict_t *dict, char **op_errstr, dict_t *rsp_dict)
/* Check brick order if the volume type is replicate or disperse. If
* force at the end of command not given then check brick order.
+ * doing this check at the originator node is sufficient.
*/
- if (!is_force) {
- if ((volinfo->type == GF_CLUSTER_TYPE_REPLICATE) ||
- (volinfo->type == GF_CLUSTER_TYPE_DISPERSE)) {
- ret = glusterd_check_brick_order(dict, msg, volinfo->type);
- if (ret) {
- gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_BAD_BRKORDER,
- "Not adding brick because of "
- "bad brick order. %s",
- msg);
- *op_errstr = gf_strdup(msg);
- goto out;
- }
+ if (is_origin_glusterd(dict) && !is_force) {
+ ret = 0;
+ if (volinfo->type == GF_CLUSTER_TYPE_REPLICATE) {
+ gf_msg_debug(this->name, 0,
+ "Replicate cluster type "
+ "found. Checking brick order.");
+ if (replica_count)
+ ret = glusterd_check_brick_order(dict, msg, volinfo->type,
+ replica_count);
+ else
+ ret = glusterd_check_brick_order(dict, msg, volinfo->type,
+ volinfo->replica_count);
+ } else if (volinfo->type == GF_CLUSTER_TYPE_DISPERSE) {
+ gf_msg_debug(this->name, 0,
+ "Disperse cluster type"
+ " found. Checking brick order.");
+ ret = glusterd_check_brick_order(dict, msg, volinfo->type,
+ volinfo->disperse_count);
+ }
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_BAD_BRKORDER,
+ "Not adding brick because of "
+ "bad brick order. %s",
+ msg);
+ *op_errstr = gf_strdup(msg);
+ goto out;
}
}
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index 6f904ae..545e688 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -14802,7 +14802,8 @@ glusterd_compare_addrinfo(struct addrinfo *first, struct addrinfo *next)
* volume are present on the same server
*/
int32_t
-glusterd_check_brick_order(dict_t *dict, char *err_str, int32_t type)
+glusterd_check_brick_order(dict_t *dict, char *err_str, int32_t type,
+ int32_t sub_count)
{
int ret = -1;
int i = 0;
@@ -14819,7 +14820,6 @@ glusterd_check_brick_order(dict_t *dict, char *err_str, int32_t type)
char *tmpptr = NULL;
char *volname = NULL;
int32_t brick_count = 0;
- int32_t sub_count = 0;
struct addrinfo *ai_info = NULL;
char brick_addr[128] = {
0,
@@ -14870,31 +14870,6 @@ glusterd_check_brick_order(dict_t *dict, char *err_str, int32_t type)
goto out;
}
- if (type != GF_CLUSTER_TYPE_DISPERSE) {
- ret = dict_get_int32n(dict, "replica-count", SLEN("replica-count"),
- &sub_count);
- if (ret) {
- gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_GET_FAILED,
- "Bricks check : Could"
- " not retrieve replica count");
- goto out;
- }
- gf_msg_debug(this->name, 0,
- "Replicate cluster type "
- "found. Checking brick order.");
- } else {
- ret = dict_get_int32n(dict, "disperse-count", SLEN("disperse-count"),
- &sub_count);
- if (ret) {
- gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_GET_FAILED,
- "Bricks check : Could"
- " not retrieve disperse count");
- goto out;
- }
- gf_msg(this->name, GF_LOG_INFO, 0, GD_MSG_DISPERSE_CLUSTER_FOUND,
- "Disperse cluster type"
- " found. Checking brick order.");
- }
brick_list_dup = brick_list_ptr = gf_strdup(brick_list);
/* Resolve hostnames and get addrinfo */
while (i < brick_count) {
@@ -14989,5 +14964,6 @@ out:
ai_list_tmp2 = ai_list_tmp1;
}
free(ai_list_tmp2);
+ gf_msg_debug("glusterd", 0, "Returning %d", ret);
return ret;
}
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.h b/xlators/mgmt/glusterd/src/glusterd-utils.h
index e2e2454..5f5de82 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.h
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.h
@@ -883,6 +883,7 @@ char *
search_brick_path_from_proc(pid_t brick_pid, char *brickpath);
int32_t
-glusterd_check_brick_order(dict_t *dict, char *err_str, int32_t type);
+glusterd_check_brick_order(dict_t *dict, char *err_str, int32_t type,
+ int32_t sub_count);
#endif
diff --git a/xlators/mgmt/glusterd/src/glusterd-volume-ops.c b/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
index 8da2ff3..134b04c 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
@@ -1024,6 +1024,8 @@ glusterd_op_stage_create_volume(dict_t *dict, char **op_errstr,
int32_t local_brick_count = 0;
int32_t i = 0;
int32_t type = 0;
+ int32_t replica_count = 0;
+ int32_t disperse_count = 0;
char *brick = NULL;
char *tmpptr = NULL;
xlator_t *this = NULL;
@@ -1119,15 +1121,42 @@ glusterd_op_stage_create_volume(dict_t *dict, char **op_errstr,
}
if (!is_force) {
- if ((type == GF_CLUSTER_TYPE_REPLICATE) ||
- (type == GF_CLUSTER_TYPE_DISPERSE)) {
- ret = glusterd_check_brick_order(dict, msg, type);
+ if (type == GF_CLUSTER_TYPE_REPLICATE) {
+ ret = dict_get_int32n(dict, "replica-count",
+ SLEN("replica-count"), &replica_count);
if (ret) {
- gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_BAD_BRKORDER,
- "Not creating volume because of "
- "bad brick order");
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_GET_FAILED,
+ "Bricks check : Could"
+ " not retrieve replica count");
+ goto out;
+ }
+ gf_msg_debug(this->name, 0,
+ "Replicate cluster type "
+ "found. Checking brick order.");
+ ret = glusterd_check_brick_order(dict, msg, type,
+ replica_count);
+ } else if (type == GF_CLUSTER_TYPE_DISPERSE) {
+ ret = dict_get_int32n(dict, "disperse-count",
+ SLEN("disperse-count"), &disperse_count);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_GET_FAILED,
+ "Bricks check : Could"
+ " not retrieve disperse count");
goto out;
}
+ gf_msg_debug(this->name, 0,
+ "Disperse cluster type"
+ " found. Checking brick order.");
+ ret = glusterd_check_brick_order(dict, msg, type,
+ disperse_count);
+ }
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_BAD_BRKORDER,
+ "Not creating the volume because of "
+ "bad brick order. %s",
+ msg);
+ *op_errstr = gf_strdup(msg);
+ goto out;
}
}
}
--
1.8.3.1

152
SOURCES/0454-features-locks-avoid-use-after-freed-of-frame-for-bl.patch Normal file
View File

@ -0,0 +1,152 @@
From cddd253c5e3f0a7c3b91c35cea8ad1921cb43b98 Mon Sep 17 00:00:00 2001
From: Kinglong Mee <kinglongmee@gmail.com>
Date: Thu, 18 Jul 2019 11:43:01 +0800
Subject: [PATCH 454/456] features/locks: avoid use after freed of frame for
blocked lock
The fop contains blocked lock may use freed frame info when other
unlock fop has unwind the blocked lock.
Because the blocked lock is added to block list in inode lock(or
other lock), after that, when out of the inode lock, the fop
contains the blocked lock should not use it.
Upstream Patch - https://review.gluster.org/#/c/glusterfs/+/23155/
>Change-Id: Icb309a1cc78380dc982b26d50c18d67e4f2c8915
>fixes: bz#1737291
>Signed-off-by: Kinglong Mee <mijinlong@horiscale.com>
Change-Id: Icb309a1cc78380dc982b26d50c18d67e4f2c8915
BUG: 1812789
Reviewed-on: https://code.engineering.redhat.com/gerrit/206465
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Xavi Hernandez Juan <xhernandez@redhat.com>
---
xlators/features/locks/src/common.c | 4 ++++
xlators/features/locks/src/entrylk.c | 4 ++--
xlators/features/locks/src/inodelk.c | 7 +++++--
xlators/features/locks/src/posix.c | 5 +++--
xlators/features/locks/src/reservelk.c | 2 --
5 files changed, 14 insertions(+), 8 deletions(-)
diff --git a/xlators/features/locks/src/common.c b/xlators/features/locks/src/common.c
index 6e7fb4b..1406e70 100644
--- a/xlators/features/locks/src/common.c
+++ b/xlators/features/locks/src/common.c
@@ -1080,6 +1080,10 @@ pl_setlk(xlator_t *this, pl_inode_t *pl_inode, posix_lock_t *lock,
lock->fl_type == F_UNLCK ? "Unlock" : "Lock",
lock->client_pid, lkowner_utoa(&lock->owner),
lock->user_flock.l_start, lock->user_flock.l_len);
+
+ pl_trace_block(this, lock->frame, NULL, NULL, F_SETLKW,
+ &lock->user_flock, NULL);
+
lock->blocked = 1;
__insert_lock(pl_inode, lock);
ret = -1;
diff --git a/xlators/features/locks/src/entrylk.c b/xlators/features/locks/src/entrylk.c
index ced5eca..93c649c 100644
--- a/xlators/features/locks/src/entrylk.c
+++ b/xlators/features/locks/src/entrylk.c
@@ -552,6 +552,8 @@ __lock_blocked_add(xlator_t *this, pl_inode_t *pinode, pl_dom_list_t *dom,
gf_msg_trace(this->name, 0, "Blocking lock: {pinode=%p, basename=%s}",
pinode, lock->basename);
+ entrylk_trace_block(this, lock->frame, NULL, NULL, NULL, lock->basename,
+ ENTRYLK_LOCK, lock->type);
out:
return -EAGAIN;
}
@@ -932,8 +934,6 @@ out:
op_ret, op_errno);
unwind:
STACK_UNWIND_STRICT(entrylk, frame, op_ret, op_errno, NULL);
- } else {
- entrylk_trace_block(this, frame, volume, fd, loc, basename, cmd, type);
}
if (pcontend != NULL) {
diff --git a/xlators/features/locks/src/inodelk.c b/xlators/features/locks/src/inodelk.c
index a9c42f1..24dee49 100644
--- a/xlators/features/locks/src/inodelk.c
+++ b/xlators/features/locks/src/inodelk.c
@@ -420,6 +420,8 @@ __lock_blocked_add(xlator_t *this, pl_dom_list_t *dom, pl_inode_lock_t *lock,
lkowner_utoa(&lock->owner), lock->user_flock.l_start,
lock->user_flock.l_len);
+ pl_trace_block(this, lock->frame, NULL, NULL, F_SETLKW, &lock->user_flock,
+ lock->volume);
out:
return -EAGAIN;
}
@@ -959,6 +961,7 @@ pl_common_inodelk(call_frame_t *frame, xlator_t *this, const char *volume,
int ret = -1;
GF_UNUSED int dict_ret = -1;
int can_block = 0;
+ short lock_type = 0;
pl_inode_t *pinode = NULL;
pl_inode_lock_t *reqlock = NULL;
pl_dom_list_t *dom = NULL;
@@ -1024,13 +1027,13 @@ pl_common_inodelk(call_frame_t *frame, xlator_t *this, const char *volume,
/* fall through */
case F_SETLK:
+ lock_type = flock->l_type;
memcpy(&reqlock->user_flock, flock, sizeof(struct gf_flock));
ret = pl_inode_setlk(this, ctx, pinode, reqlock, can_block, dom,
inode);
if (ret < 0) {
- if ((can_block) && (F_UNLCK != flock->l_type)) {
- pl_trace_block(this, frame, fd, loc, cmd, flock, volume);
+ if ((can_block) && (F_UNLCK != lock_type)) {
goto out;
}
gf_log(this->name, GF_LOG_TRACE, "returning EAGAIN");
diff --git a/xlators/features/locks/src/posix.c b/xlators/features/locks/src/posix.c
index 50f1265..7887b82 100644
--- a/xlators/features/locks/src/posix.c
+++ b/xlators/features/locks/src/posix.c
@@ -2557,6 +2557,7 @@ pl_lk(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t cmd,
uint32_t lk_flags = 0;
posix_locks_private_t *priv = this->private;
pl_local_t *local = NULL;
+ short lock_type = 0;
int ret = dict_get_uint32(xdata, GF_LOCK_MODE, &lk_flags);
if (ret == 0) {
@@ -2701,6 +2702,7 @@ pl_lk(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t cmd,
case F_SETLK:
reqlock->frame = frame;
reqlock->this = this;
+ lock_type = flock->l_type;
pthread_mutex_lock(&pl_inode->mutex);
{
@@ -2738,8 +2740,7 @@ pl_lk(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t cmd,
ret = pl_setlk(this, pl_inode, reqlock, can_block);
if (ret == -1) {
- if ((can_block) && (F_UNLCK != flock->l_type)) {
- pl_trace_block(this, frame, fd, NULL, cmd, flock, NULL);
+ if ((can_block) && (F_UNLCK != lock_type)) {
goto out;
}
gf_log(this->name, GF_LOG_DEBUG, "returning EAGAIN");
diff --git a/xlators/features/locks/src/reservelk.c b/xlators/features/locks/src/reservelk.c
index 51076d7..604691f 100644
--- a/xlators/features/locks/src/reservelk.c
+++ b/xlators/features/locks/src/reservelk.c
@@ -312,8 +312,6 @@ grant_blocked_lock_calls(xlator_t *this, pl_inode_t *pl_inode)
ret = pl_setlk(this, pl_inode, lock, can_block);
if (ret == -1) {
if (can_block) {
- pl_trace_block(this, lock->frame, fd, NULL, cmd,
- &lock->user_flock, NULL);
continue;
} else {
gf_log(this->name, GF_LOG_DEBUG, "returning EAGAIN");
--
1.8.3.1

1253
SOURCES/0455-locks-prevent-deletion-of-locked-entries.patch Normal file
View File

File diff suppressed because it is too large Load Diff

74
SOURCES/0456-add-clean-local-after-grant-lock.patch Normal file
View File

@ -0,0 +1,74 @@
From c38b38249fdf951565f6501ce8e9a4d01142d43e Mon Sep 17 00:00:00 2001
From: l17zhou <cynthia.zhou@nokia-sbell.com>
Date: Tue, 3 Dec 2019 07:43:35 +0200
Subject: [PATCH 456/456] add clean local after grant lock
found by flock test, without correct ref number of fd,
lock will not be correctly released.
Upstream patch:
> Upstream patch link: https://review.gluster.org/c/glusterfs/+/23794
> Fixes: bz#1779089
> Change-Id: I3e466b17c852eb219c8778e43af8ad670a8449cc
> Signed-off-by: l17zhou <cynthia.zhou@nokia-sbell.com>
BUG: 1854165
Change-Id: I3e466b17c852eb219c8778e43af8ad670a8449cc
Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/206673
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/features/locks/src/common.c | 15 ++++++++-------
1 file changed, 8 insertions(+), 7 deletions(-)
diff --git a/xlators/features/locks/src/common.c b/xlators/features/locks/src/common.c
index 0c52853..cddbfa6 100644
--- a/xlators/features/locks/src/common.c
+++ b/xlators/features/locks/src/common.c
@@ -961,7 +961,7 @@ grant_blocked_locks(xlator_t *this, pl_inode_t *pl_inode)
struct list_head granted_list;
posix_lock_t *tmp = NULL;
posix_lock_t *lock = NULL;
-
+ pl_local_t *local = NULL;
INIT_LIST_HEAD(&granted_list);
pthread_mutex_lock(&pl_inode->mutex);
@@ -976,9 +976,9 @@ grant_blocked_locks(xlator_t *this, pl_inode_t *pl_inode)
pl_trace_out(this, lock->frame, NULL, NULL, F_SETLKW, &lock->user_flock,
0, 0, NULL);
-
- STACK_UNWIND_STRICT(lk, lock->frame, 0, 0, &lock->user_flock, NULL);
-
+ local = lock->frame->local;
+ PL_STACK_UNWIND_AND_FREE(local, lk, lock->frame, 0, 0,
+ &lock->user_flock, NULL);
__destroy_lock(lock);
}
@@ -997,6 +997,7 @@ pl_send_prelock_unlock(xlator_t *this, pl_inode_t *pl_inode,
struct list_head granted_list;
posix_lock_t *tmp = NULL;
posix_lock_t *lock = NULL;
+ pl_local_t *local = NULL;
int ret = -1;
@@ -1024,9 +1025,9 @@ pl_send_prelock_unlock(xlator_t *this, pl_inode_t *pl_inode,
pl_trace_out(this, lock->frame, NULL, NULL, F_SETLKW, &lock->user_flock,
0, 0, NULL);
-
- STACK_UNWIND_STRICT(lk, lock->frame, 0, 0, &lock->user_flock, NULL);
-
+ local = lock->frame->local;
+ PL_STACK_UNWIND_AND_FREE(local, lk, lock->frame, 0, 0,
+ &lock->user_flock, NULL);
__destroy_lock(lock);
}
--
1.8.3.1

409
SOURCES/0457-cluster-ec-Improve-detection-of-new-heals.patch Normal file
View File

@ -0,0 +1,409 @@
From 3e8b3a2c2c6f83635486035fc8040c87d89813d2 Mon Sep 17 00:00:00 2001
From: Xavi Hernandez <xhernandez@redhat.com>
Date: Thu, 2 Jul 2020 18:08:52 +0200
Subject: [PATCH 457/465] cluster/ec: Improve detection of new heals
When EC successfully healed a directory it assumed that maybe other
entries inside that directory could have been created, which could
require additional heal cycles. For this reason, when the heal happened
as part of one index heal iteration, it triggered a new iteration.
The problem happened when the directory was healthy, so no new entries
were added, but its index entry was not removed for some reason. In
this case self-heal started and endless loop healing the same directory
continuously, cause high CPU utilization.
This patch improves detection of new files added to the heal index so
that a new index heal iteration is only triggered if there is new work
to do.
>Upstream patch: https://review.gluster.org/#/c/glusterfs/+/24665/
>Fixes: #1354
Change-Id: I2355742b85fbfa6de758bccc5d2e1a283c82b53f
BUG: 1852736
Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/208041
Tested-by: Ashish Pandey <aspandey@redhat.com>
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Ashish Pandey <aspandey@redhat.com>
---
xlators/cluster/ec/src/ec-common.c | 2 +-
xlators/cluster/ec/src/ec-heal.c | 58 +++++++++++++++++++++++-----------
xlators/cluster/ec/src/ec-heald.c | 24 ++++++++++----
xlators/cluster/ec/src/ec-inode-read.c | 27 ++++++++++++++--
xlators/cluster/ec/src/ec-types.h | 4 +--
xlators/cluster/ec/src/ec.h | 1 +
6 files changed, 86 insertions(+), 30 deletions(-)
diff --git a/xlators/cluster/ec/src/ec-common.c b/xlators/cluster/ec/src/ec-common.c
index e580bfb..e3f8769 100644
--- a/xlators/cluster/ec/src/ec-common.c
+++ b/xlators/cluster/ec/src/ec-common.c
@@ -230,7 +230,7 @@ ec_child_next(ec_t *ec, ec_fop_data_t *fop, uint32_t idx)
int32_t
ec_heal_report(call_frame_t *frame, void *cookie, xlator_t *this,
int32_t op_ret, int32_t op_errno, uintptr_t mask, uintptr_t good,
- uintptr_t bad, dict_t *xdata)
+ uintptr_t bad, uint32_t pending, dict_t *xdata)
{
if (op_ret < 0) {
gf_msg(this->name, GF_LOG_DEBUG, op_errno, EC_MSG_HEAL_FAIL,
diff --git a/xlators/cluster/ec/src/ec-heal.c b/xlators/cluster/ec/src/ec-heal.c
index 06a7016..e2de879 100644
--- a/xlators/cluster/ec/src/ec-heal.c
+++ b/xlators/cluster/ec/src/ec-heal.c
@@ -72,6 +72,7 @@ struct ec_name_data {
char *name;
inode_t *parent;
default_args_cbk_t *replies;
+ uint32_t heal_pending;
};
static char *ec_ignore_xattrs[] = {GF_SELINUX_XATTR_KEY, QUOTA_SIZE_KEY, NULL};
@@ -996,6 +997,7 @@ ec_set_new_entry_dirty(ec_t *ec, loc_t *loc, struct iatt *ia,
ret = -ENOTCONN;
goto out;
}
+
out:
if (xattr)
dict_unref(xattr);
@@ -1164,6 +1166,7 @@ ec_create_name(call_frame_t *frame, ec_t *ec, inode_t *parent, char *name,
dict_t *xdata = NULL;
char *linkname = NULL;
ec_config_t config;
+
/* There should be just one gfid key */
EC_REPLIES_ALLOC(replies, ec->nodes);
if (gfid_db->count != 1) {
@@ -1408,6 +1411,11 @@ __ec_heal_name(call_frame_t *frame, ec_t *ec, inode_t *parent, char *name,
ret = ec_create_name(frame, ec, parent, name, replies, gfid_db, enoent,
participants);
+ if (ret >= 0) {
+ /* If ec_create_name() succeeded we return 1 to indicate that a new
+ * file has been created and it will need to be healed. */
+ ret = 1;
+ }
out:
cluster_replies_wipe(replies, ec->nodes);
loc_wipe(&loc);
@@ -1485,18 +1493,22 @@ ec_name_heal_handler(xlator_t *subvol, gf_dirent_t *entry, loc_t *parent,
ret = ec_heal_name(name_data->frame, ec, parent->inode, entry->d_name,
name_on);
- if (ret < 0)
+ if (ret < 0) {
memset(name_on, 0, ec->nodes);
+ } else {
+ name_data->heal_pending += ret;
+ }
for (i = 0; i < ec->nodes; i++)
if (name_data->participants[i] && !name_on[i])
name_data->failed_on[i] = 1;
+
return 0;
}
int
ec_heal_names(call_frame_t *frame, ec_t *ec, inode_t *inode,
- unsigned char *participants)
+ unsigned char *participants, uint32_t *pending)
{
int i = 0;
int j = 0;
@@ -1509,7 +1521,7 @@ ec_heal_names(call_frame_t *frame, ec_t *ec, inode_t *inode,
name_data.frame = frame;
name_data.participants = participants;
name_data.failed_on = alloca0(ec->nodes);
- ;
+ name_data.heal_pending = 0;
for (i = 0; i < ec->nodes; i++) {
if (!participants[i])
@@ -1528,6 +1540,8 @@ ec_heal_names(call_frame_t *frame, ec_t *ec, inode_t *inode,
break;
}
}
+ *pending += name_data.heal_pending;
+
loc_wipe(&loc);
return ret;
}
@@ -1535,7 +1549,7 @@ ec_heal_names(call_frame_t *frame, ec_t *ec, inode_t *inode,
int
__ec_heal_entry(call_frame_t *frame, ec_t *ec, inode_t *inode,
unsigned char *heal_on, unsigned char *sources,
- unsigned char *healed_sinks)
+ unsigned char *healed_sinks, uint32_t *pending)
{
unsigned char *locked_on = NULL;
unsigned char *output = NULL;
@@ -1580,7 +1594,7 @@ unlock:
if (sources[i] || healed_sinks[i])
participants[i] = 1;
}
- ret = ec_heal_names(frame, ec, inode, participants);
+ ret = ec_heal_names(frame, ec, inode, participants, pending);
if (EC_COUNT(participants, ec->nodes) <= ec->fragments)
goto out;
@@ -1601,7 +1615,8 @@ out:
int
ec_heal_entry(call_frame_t *frame, ec_t *ec, inode_t *inode,
- unsigned char *sources, unsigned char *healed_sinks)
+ unsigned char *sources, unsigned char *healed_sinks,
+ uint32_t *pending)
{
unsigned char *locked_on = NULL;
unsigned char *up_subvols = NULL;
@@ -1632,7 +1647,7 @@ ec_heal_entry(call_frame_t *frame, ec_t *ec, inode_t *inode,
goto unlock;
}
ret = __ec_heal_entry(frame, ec, inode, locked_on, sources,
- healed_sinks);
+ healed_sinks, pending);
}
unlock:
cluster_uninodelk(ec->xl_list, locked_on, ec->nodes, replies, output, frame,
@@ -1953,14 +1968,14 @@ ec_manager_heal_block(ec_fop_data_t *fop, int32_t state)
if (fop->cbks.heal) {
fop->cbks.heal(fop->req_frame, fop, fop->xl, 0, 0,
(heal->good | heal->bad), heal->good, heal->bad,
- NULL);
+ 0, NULL);
}
return EC_STATE_END;
case -EC_STATE_REPORT:
if (fop->cbks.heal) {
- fop->cbks.heal(fop->req_frame, fop, fop->xl, -1, fop->error, 0,
- 0, 0, NULL);
+ fop->cbks.heal(fop->req_frame, fop->data, fop->xl, -1,
+ fop->error, 0, 0, 0, 0, NULL);
}
return EC_STATE_END;
@@ -1997,14 +2012,15 @@ out:
if (fop != NULL) {
ec_manager(fop, error);
} else {
- func(frame, NULL, this, -1, error, 0, 0, 0, NULL);
+ func(frame, heal, this, -1, error, 0, 0, 0, 0, NULL);
}
}
int32_t
ec_heal_block_done(call_frame_t *frame, void *cookie, xlator_t *this,
int32_t op_ret, int32_t op_errno, uintptr_t mask,
- uintptr_t good, uintptr_t bad, dict_t *xdata)
+ uintptr_t good, uintptr_t bad, uint32_t pending,
+ dict_t *xdata)
{
ec_fop_data_t *fop = cookie;
ec_heal_t *heal = fop->data;
@@ -2489,6 +2505,7 @@ ec_heal_do(xlator_t *this, void *data, loc_t *loc, int32_t partial)
intptr_t mbad = 0;
intptr_t good = 0;
intptr_t bad = 0;
+ uint32_t pending = 0;
ec_fop_data_t *fop = data;
gf_boolean_t blocking = _gf_false;
ec_heal_need_t need_heal = EC_HEAL_NONEED;
@@ -2524,7 +2541,7 @@ ec_heal_do(xlator_t *this, void *data, loc_t *loc, int32_t partial)
if (loc->name && strlen(loc->name)) {
ret = ec_heal_name(frame, ec, loc->parent, (char *)loc->name,
participants);
- if (ret == 0) {
+ if (ret >= 0) {
gf_msg_debug(this->name, 0,
"%s: name heal "
"successful on %" PRIXPTR,
@@ -2542,7 +2559,7 @@ ec_heal_do(xlator_t *this, void *data, loc_t *loc, int32_t partial)
/* Mount triggers heal only when it detects that it must need heal, shd
* triggers heals periodically which need not be thorough*/
- if (ec->shd.iamshd) {
+ if (ec->shd.iamshd && (ret <= 0)) {
ec_heal_inspect(frame, ec, loc->inode, up_subvols, _gf_false, _gf_false,
&need_heal);
@@ -2552,13 +2569,15 @@ ec_heal_do(xlator_t *this, void *data, loc_t *loc, int32_t partial)
goto out;
}
}
+
sources = alloca0(ec->nodes);
healed_sinks = alloca0(ec->nodes);
if (IA_ISREG(loc->inode->ia_type)) {
ret = ec_heal_data(frame, ec, blocking, loc->inode, sources,
healed_sinks);
} else if (IA_ISDIR(loc->inode->ia_type) && !partial) {
- ret = ec_heal_entry(frame, ec, loc->inode, sources, healed_sinks);
+ ret = ec_heal_entry(frame, ec, loc->inode, sources, healed_sinks,
+ &pending);
} else {
ret = 0;
memcpy(sources, participants, ec->nodes);
@@ -2588,10 +2607,11 @@ out:
if (fop->cbks.heal) {
fop->cbks.heal(fop->req_frame, fop, fop->xl, op_ret, op_errno,
ec_char_array_to_mask(participants, ec->nodes),
- mgood & good, mbad & bad, NULL);
+ mgood & good, mbad & bad, pending, NULL);
}
if (frame)
STACK_DESTROY(frame->root);
+
return;
}
@@ -2638,8 +2658,8 @@ void
ec_heal_fail(ec_t *ec, ec_fop_data_t *fop)
{
if (fop->cbks.heal) {
- fop->cbks.heal(fop->req_frame, NULL, ec->xl, -1, fop->error, 0, 0, 0,
- NULL);
+ fop->cbks.heal(fop->req_frame, fop->data, ec->xl, -1, fop->error, 0, 0,
+ 0, 0, NULL);
}
ec_fop_data_release(fop);
}
@@ -2826,7 +2846,7 @@ fail:
if (fop)
ec_fop_data_release(fop);
if (func)
- func(frame, NULL, this, -1, err, 0, 0, 0, NULL);
+ func(frame, data, this, -1, err, 0, 0, 0, 0, NULL);
}
int
diff --git a/xlators/cluster/ec/src/ec-heald.c b/xlators/cluster/ec/src/ec-heald.c
index cba111a..4f4b6aa 100644
--- a/xlators/cluster/ec/src/ec-heald.c
+++ b/xlators/cluster/ec/src/ec-heald.c
@@ -156,15 +156,27 @@ int
ec_shd_selfheal(struct subvol_healer *healer, int child, loc_t *loc,
gf_boolean_t full)
{
+ dict_t *xdata = NULL;
+ uint32_t count;
int32_t ret;
- ret = syncop_getxattr(healer->this, loc, NULL, EC_XATTR_HEAL, NULL, NULL);
- if (!full && (ret >= 0) && (loc->inode->ia_type == IA_IFDIR)) {
+ ret = syncop_getxattr(healer->this, loc, NULL, EC_XATTR_HEAL, NULL, &xdata);
+ if (!full && (loc->inode->ia_type == IA_IFDIR)) {
/* If we have just healed a directory, it's possible that
- * other index entries have appeared to be healed. We put a
- * mark so that we can check it later and restart a scan
- * without delay. */
- healer->rerun = _gf_true;
+ * other index entries have appeared to be healed. */
+ if ((xdata != NULL) &&
+ (dict_get_uint32(xdata, EC_XATTR_HEAL_NEW, &count) == 0) &&
+ (count > 0)) {
+ /* Force a rerun of the index healer. */
+ gf_msg_debug(healer->this->name, 0, "%d more entries to heal",
+ count);
+
+ healer->rerun = _gf_true;
+ }
+ }
+
+ if (xdata != NULL) {
+ dict_unref(xdata);
}
return ret;
diff --git a/xlators/cluster/ec/src/ec-inode-read.c b/xlators/cluster/ec/src/ec-inode-read.c
index f87a94a..e82e8f6 100644
--- a/xlators/cluster/ec/src/ec-inode-read.c
+++ b/xlators/cluster/ec/src/ec-inode-read.c
@@ -393,7 +393,8 @@ ec_manager_getxattr(ec_fop_data_t *fop, int32_t state)
int32_t
ec_getxattr_heal_cbk(call_frame_t *frame, void *cookie, xlator_t *xl,
int32_t op_ret, int32_t op_errno, uintptr_t mask,
- uintptr_t good, uintptr_t bad, dict_t *xdata)
+ uintptr_t good, uintptr_t bad, uint32_t pending,
+ dict_t *xdata)
{
ec_fop_data_t *fop = cookie;
fop_getxattr_cbk_t func = fop->data;
@@ -402,6 +403,25 @@ ec_getxattr_heal_cbk(call_frame_t *frame, void *cookie, xlator_t *xl,
char *str;
char bin1[65], bin2[65];
+ /* We try to return the 'pending' information in xdata, but if this cannot
+ * be set, we will ignore it silently. We prefer to report the success or
+ * failure of the heal itself. */
+ if (xdata == NULL) {
+ xdata = dict_new();
+ } else {
+ dict_ref(xdata);
+ }
+ if (xdata != NULL) {
+ if (dict_set_uint32(xdata, EC_XATTR_HEAL_NEW, pending) != 0) {
+ /* dict_set_uint32() is marked as 'warn_unused_result' and gcc
+ * enforces to check the result in this case. However we don't
+ * really care if it succeeded or not. We'll just do the same.
+ *
+ * This empty 'if' avoids the warning, and it will be removed by
+ * the optimizer. */
+ }
+ }
+
if (op_ret >= 0) {
dict = dict_new();
if (dict == NULL) {
@@ -435,11 +455,14 @@ ec_getxattr_heal_cbk(call_frame_t *frame, void *cookie, xlator_t *xl,
}
out:
- func(frame, NULL, xl, op_ret, op_errno, dict, NULL);
+ func(frame, NULL, xl, op_ret, op_errno, dict, xdata);
if (dict != NULL) {
dict_unref(dict);
}
+ if (xdata != NULL) {
+ dict_unref(xdata);
+ }
return 0;
}
diff --git a/xlators/cluster/ec/src/ec-types.h b/xlators/cluster/ec/src/ec-types.h
index 34a9768..f15429d 100644
--- a/xlators/cluster/ec/src/ec-types.h
+++ b/xlators/cluster/ec/src/ec-types.h
@@ -186,10 +186,10 @@ struct _ec_inode {
typedef int32_t (*fop_heal_cbk_t)(call_frame_t *, void *, xlator_t *, int32_t,
int32_t, uintptr_t, uintptr_t, uintptr_t,
- dict_t *);
+ uint32_t, dict_t *);
typedef int32_t (*fop_fheal_cbk_t)(call_frame_t *, void *, xlator_t *, int32_t,
int32_t, uintptr_t, uintptr_t, uintptr_t,
- dict_t *);
+ uint32_t, dict_t *);
union _ec_cbk {
fop_access_cbk_t access;
diff --git a/xlators/cluster/ec/src/ec.h b/xlators/cluster/ec/src/ec.h
index 1b210d9..6f6de6d 100644
--- a/xlators/cluster/ec/src/ec.h
+++ b/xlators/cluster/ec/src/ec.h
@@ -18,6 +18,7 @@
#define EC_XATTR_SIZE EC_XATTR_PREFIX "size"
#define EC_XATTR_VERSION EC_XATTR_PREFIX "version"
#define EC_XATTR_HEAL EC_XATTR_PREFIX "heal"
+#define EC_XATTR_HEAL_NEW EC_XATTR_PREFIX "heal-new"
#define EC_XATTR_DIRTY EC_XATTR_PREFIX "dirty"
#define EC_STRIPE_CACHE_MAX_SIZE 10
#define EC_VERSION_SIZE 2
--
1.8.3.1

182
SOURCES/0458-features-bit-rot-stub-clean-the-mutex-after-cancelli.patch Normal file
View File

@ -0,0 +1,182 @@
From ed73f2046dd3fbb22341bf9fc004087d90dfbe6d Mon Sep 17 00:00:00 2001
From: Raghavendra Bhat <raghavendra@redhat.com>
Date: Mon, 15 Apr 2019 14:09:34 -0400
Subject: [PATCH 458/465] features/bit-rot-stub: clean the mutex after
cancelling the signer thread
When bit-rot feature is disabled, the signer thread from the bit-rot-stub
xlator (the thread which performs the setxattr of the signature on to the
disk) is cancelled. But, if the cancelled signer thread had already held
the mutex (&priv->lock) which it uses to monitor the queue of files to
be signed, then the mutex is never released. This creates problems in
future when the feature is enabled again. Both the new instance of the
signer thread and the regular thread which enqueues the files to be
signed will be blocked on this mutex.
So, as part of cancelling the signer thread, unlock the mutex associated
with it as well using pthread_cleanup_push and pthread_cleanup_pop.
Upstream patch:
> patch: https://review.gluster.org/22572
> fixes: #bz1700078
> Change-Id: Ib761910caed90b268e69794ddeb108165487af40
Change-Id: Ib761910caed90b268e69794ddeb108165487af40
BUG: 1851424
Signed-off-by: Raghavendra M <raghavendra@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/208304
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
.../bit-rot/src/stub/bit-rot-stub-messages.h | 4 +-
xlators/features/bit-rot/src/stub/bit-rot-stub.c | 62 +++++++++++++++++++---
2 files changed, 59 insertions(+), 7 deletions(-)
diff --git a/xlators/features/bit-rot/src/stub/bit-rot-stub-messages.h b/xlators/features/bit-rot/src/stub/bit-rot-stub-messages.h
index 7f07f29..155802b 100644
--- a/xlators/features/bit-rot/src/stub/bit-rot-stub-messages.h
+++ b/xlators/features/bit-rot/src/stub/bit-rot-stub-messages.h
@@ -39,6 +39,8 @@ GLFS_MSGID(BITROT_STUB, BRS_MSG_NO_MEMORY, BRS_MSG_SET_EVENT_FAILED,
BRS_MSG_BAD_HANDLE_DIR_NULL, BRS_MSG_BAD_OBJ_THREAD_FAIL,
BRS_MSG_BAD_OBJ_DIR_CLOSE_FAIL, BRS_MSG_LINK_FAIL,
BRS_MSG_BAD_OBJ_UNLINK_FAIL, BRS_MSG_DICT_SET_FAILED,
- BRS_MSG_PATH_GET_FAILED, BRS_MSG_NULL_LOCAL);
+ BRS_MSG_PATH_GET_FAILED, BRS_MSG_NULL_LOCAL,
+ BRS_MSG_SPAWN_SIGN_THRD_FAILED, BRS_MSG_KILL_SIGN_THREAD,
+ BRS_MSG_NON_BITD_PID, BRS_MSG_SIGN_PREPARE_FAIL);
#endif /* !_BITROT_STUB_MESSAGES_H_ */
diff --git a/xlators/features/bit-rot/src/stub/bit-rot-stub.c b/xlators/features/bit-rot/src/stub/bit-rot-stub.c
index 3f48a4b..c3f81bc 100644
--- a/xlators/features/bit-rot/src/stub/bit-rot-stub.c
+++ b/xlators/features/bit-rot/src/stub/bit-rot-stub.c
@@ -26,6 +26,15 @@
#define BR_STUB_REQUEST_COOKIE 0x1
+void
+br_stub_lock_cleaner(void *arg)
+{
+ pthread_mutex_t *clean_mutex = arg;
+
+ pthread_mutex_unlock(clean_mutex);
+ return;
+}
+
void *
br_stub_signth(void *);
@@ -166,8 +175,11 @@ init(xlator_t *this)
ret = gf_thread_create(&priv->signth, NULL, br_stub_signth, this,
"brssign");
- if (ret != 0)
+ if (ret != 0) {
+ gf_msg(this->name, GF_LOG_WARNING, 0, BRS_MSG_SPAWN_SIGN_THRD_FAILED,
+ "failed to create the new thread for signer");
goto cleanup_lock;
+ }
ret = br_stub_bad_object_container_init(this, priv);
if (ret) {
@@ -214,11 +226,15 @@ reconfigure(xlator_t *this, dict_t *options)
priv = this->private;
GF_OPTION_RECONF("bitrot", priv->do_versioning, options, bool, err);
- if (priv->do_versioning) {
+ if (priv->do_versioning && !priv->signth) {
ret = gf_thread_create(&priv->signth, NULL, br_stub_signth, this,
"brssign");
- if (ret != 0)
+ if (ret != 0) {
+ gf_msg(this->name, GF_LOG_WARNING, 0,
+ BRS_MSG_SPAWN_SIGN_THRD_FAILED,
+ "failed to create the new thread for signer");
goto err;
+ }
ret = br_stub_bad_object_container_init(this, priv);
if (ret) {
@@ -232,8 +248,11 @@ reconfigure(xlator_t *this, dict_t *options)
gf_msg(this->name, GF_LOG_ERROR, 0,
BRS_MSG_CANCEL_SIGN_THREAD_FAILED,
"Could not cancel sign serializer thread");
+ } else {
+ gf_msg(this->name, GF_LOG_INFO, 0, BRS_MSG_KILL_SIGN_THREAD,
+ "killed the signer thread");
+ priv->signth = 0;
}
- priv->signth = 0;
}
if (priv->container.thread) {
@@ -902,6 +921,24 @@ br_stub_signth(void *arg)
THIS = this;
while (1) {
+ /*
+ * Disabling bit-rot feature leads to this particular thread
+ * getting cleaned up by reconfigure via a call to the function
+ * gf_thread_cleanup_xint (which in turn calls pthread_cancel
+ * and pthread_join). But, if this thread had held the mutex
+ * &priv->lock at the time of cancellation, then it leads to
+ * deadlock in future when bit-rot feature is enabled (which
+ * again spawns this thread which cant hold the lock as the
+ * mutex is still held by the previous instance of the thread
+ * which got killed). Also, the br_stub_handle_object_signature
+ * function which is called whenever file has to be signed
+ * also gets blocked as it too attempts to acquire &priv->lock.
+ *
+ * So, arrange for the lock to be unlocked as part of the
+ * cleanup of this thread using pthread_cleanup_push and
+ * pthread_cleanup_pop.
+ */
+ pthread_cleanup_push(br_stub_lock_cleaner, &priv->lock);
pthread_mutex_lock(&priv->lock);
{
while (list_empty(&priv->squeue))
@@ -912,6 +949,7 @@ br_stub_signth(void *arg)
list_del_init(&sigstub->list);
}
pthread_mutex_unlock(&priv->lock);
+ pthread_cleanup_pop(0);
call_resume(sigstub->stub);
@@ -1042,12 +1080,22 @@ br_stub_handle_object_signature(call_frame_t *frame, xlator_t *this, fd_t *fd,
priv = this->private;
- if (frame->root->pid != GF_CLIENT_PID_BITD)
+ if (frame->root->pid != GF_CLIENT_PID_BITD) {
+ gf_msg(this->name, GF_LOG_WARNING, op_errno, BRS_MSG_NON_BITD_PID,
+ "PID %d from where signature request"
+ "came, does not belong to bit-rot daemon."
+ "Unwinding the fop",
+ frame->root->pid);
goto dofop;
+ }
ret = br_stub_prepare_signature(this, dict, fd->inode, sign, &fakesuccess);
- if (ret)
+ if (ret) {
+ gf_msg(this->name, GF_LOG_WARNING, 0, BRS_MSG_SIGN_PREPARE_FAIL,
+ "failed to prepare the signature for %s. Unwinding the fop",
+ uuid_utoa(fd->inode->gfid));
goto dofop;
+ }
if (fakesuccess) {
op_ret = op_errno = 0;
goto dofop;
@@ -1387,6 +1435,8 @@ br_stub_fsetxattr(call_frame_t *frame, xlator_t *this, fd_t *fd, dict_t *dict,
/* object signature request */
ret = dict_get_bin(dict, GLUSTERFS_SET_OBJECT_SIGNATURE, (void **)&sign);
if (!ret) {
+ gf_msg_debug(this->name, 0, "got SIGNATURE request on %s",
+ uuid_utoa(fd->inode->gfid));
br_stub_handle_object_signature(frame, this, fd, dict, sign, xdata);
goto done;
}
--
1.8.3.1

181
SOURCES/0459-features-bit-rot-Unconditionally-sign-the-files-duri.patch Normal file
View File

@ -0,0 +1,181 @@
From 80eef2f52bb92ed740ac00eeb11ee7a3e7fffff2 Mon Sep 17 00:00:00 2001
From: Raghavendra Bhat <raghavendra@redhat.com>
Date: Mon, 11 Mar 2019 12:16:50 -0400
Subject: [PATCH 459/465] features/bit-rot: Unconditionally sign the files
during oneshot crawl
Currently bit-rot feature has an issue with disabling and reenabling it
on the same volume. Consider enabling bit-rot detection which goes on to
crawl and sign all the files present in the volume. Then some files are
modified and the bit-rot daemon goes on to sign the modified files with
the correct signature. Now, disable bit-rot feature. While, signing and
scrubbing are not happening, previous checksums of the files continue to
exist as extended attributes. Now, if some files with checksum xattrs get
modified, they are not signed with new signature as the feature is off.
At this point, if the feature is enabled again, the bit rot daemon will
go and sign those files which does not have any bit-rot specific xattrs
(i.e. those files which were created after bit-rot was disabled). Whereas
the files with bit-rot xattrs wont get signed with proper new checksum.
At this point if scrubber runs, it finds the on disk checksum and the actual
checksum of the file to be different (because the file got modified) and
marks the file as corrupted.
FIX:
The fix is to unconditionally sign the files when the bit-rot daemon
comes up (instead of skipping the files with bit-rot xattrs).
upstream fix:
> patch: https://review.gluster.org/#/c/glusterfs/+/22360/
> fixes: #bz1700078
> Change-ID: Iadfb47dd39f7e2e77f22d549a4a07a385284f4f5
Change-Id: Iadfb47dd39f7e2e77f22d549a4a07a385284f4f5
BUG: 1851424
Signed-off-by: Raghavendra M <raghavendra@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/208305
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/bitrot/bug-1700078.t | 87 +++++++++++++++++++++++++++++
xlators/features/bit-rot/src/bitd/bit-rot.c | 15 ++++-
2 files changed, 101 insertions(+), 1 deletion(-)
create mode 100644 tests/bitrot/bug-1700078.t
diff --git a/tests/bitrot/bug-1700078.t b/tests/bitrot/bug-1700078.t
new file mode 100644
index 0000000..f273742
--- /dev/null
+++ b/tests/bitrot/bug-1700078.t
@@ -0,0 +1,87 @@
+#!/bin/bash
+
+. $(dirname $0)/../include.rc
+. $(dirname $0)/../volume.rc
+
+cleanup;
+
+## Start glusterd
+TEST glusterd;
+TEST pidof glusterd;
+
+## Lets create and start the volume
+TEST $CLI volume create $V0 $H0:$B0/${V0}1
+TEST $CLI volume start $V0
+
+## Enable bitrot for volume $V0
+TEST $CLI volume bitrot $V0 enable
+
+## Turn off quick-read so that it wont cache the contents
+# of the file in lookup. For corrupted files, it might
+# end up in reads being served from the cache instead of
+# an error.
+TEST $CLI volume set $V0 performance.quick-read off
+
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" get_bitd_count
+
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT 'Active' scrub_status $V0 'State of scrub'
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT '/var/log/glusterfs/bitd.log' scrub_status $V0 'Bitrot error log location'
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT '/var/log/glusterfs/scrub.log' scrub_status $V0 'Scrubber error log location'
+
+## Set expiry-timeout to 1 sec
+TEST $CLI volume set $V0 features.expiry-time 1
+
+##Mount $V0
+TEST glusterfs --volfile-id=$V0 --volfile-server=$H0 $M0 --attribute-timeout=0 --entry-timeout=0
+
+## Turn off quick-read xlator so that, the contents are not served from the
+# quick-read cache.
+TEST $CLI volume set $V0 performance.quick-read off
+
+#Create sample file
+TEST `echo "1234" > $M0/FILE1`
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT 'trusted.bit-rot.signature' check_for_xattr 'trusted.bit-rot.signature' "/$B0/${V0}1/FILE1"
+
+##disable bitrot
+TEST $CLI volume bitrot $V0 disable
+
+## modify the file
+TEST `echo "write" >> $M0/FILE1`
+
+# unmount and remount when the file has to be accessed.
+# This is to ensure that, when the remount happens,
+# and the file is read, its contents are served from the
+# brick instead of cache.
+EXPECT_WITHIN $UMOUNT_TIMEOUT "Y" force_umount $M0
+
+##enable bitrot
+TEST $CLI volume bitrot $V0 enable
+
+# expiry time is set to 1 second. Hence sleep for 2 seconds for the
+# oneshot crawler to finish its crawling and sign the file properly.
+sleep 2
+
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" get_bitd_count
+
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT 'Active' scrub_status $V0 'State of scrub'
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT '/var/log/glusterfs/bitd.log' scrub_status $V0 'Bitrot error log location'
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT '/var/log/glusterfs/scrub.log' scrub_status $V0 'Scrubber error log location'
+
+## Ondemand scrub
+TEST $CLI volume bitrot $V0 scrub ondemand
+
+# the scrub ondemand CLI command, just ensures that
+# the scrubber has received the ondemand scrub directive
+# and started. sleep for 2 seconds for scrubber to finish
+# crawling and marking file(s) as bad (if if finds that
+# corruption has happened) which are filesystem operations.
+sleep 2
+
+TEST ! getfattr -n 'trusted.bit-rot.bad-file' $B0/${V0}1/FILE1
+
+##Mount $V0
+TEST glusterfs --volfile-id=$V0 --volfile-server=$H0 $M0 --attribute-timeout=0 --entry-timeout=0
+
+TEST cat $M0/FILE1
+
+cleanup;
diff --git a/xlators/features/bit-rot/src/bitd/bit-rot.c b/xlators/features/bit-rot/src/bitd/bit-rot.c
index b8feef7..424c0d5 100644
--- a/xlators/features/bit-rot/src/bitd/bit-rot.c
+++ b/xlators/features/bit-rot/src/bitd/bit-rot.c
@@ -973,6 +973,7 @@ bitd_oneshot_crawl(xlator_t *subvol, gf_dirent_t *entry, loc_t *parent,
int32_t ret = -1;
inode_t *linked_inode = NULL;
gf_boolean_t need_signing = _gf_false;
+ gf_boolean_t need_reopen = _gf_true;
GF_VALIDATE_OR_GOTO("bit-rot", subvol, out);
GF_VALIDATE_OR_GOTO("bit-rot", data, out);
@@ -1046,6 +1047,18 @@ bitd_oneshot_crawl(xlator_t *subvol, gf_dirent_t *entry, loc_t *parent,
uuid_utoa(linked_inode->gfid));
} else {
need_signing = br_check_object_need_sign(this, xattr, child);
+
+ /*
+ * If we are here means, bitrot daemon has started. Is it just
+ * a simple restart of the daemon or is it started because the
+ * feature is enabled is something hard to determine. Hence,
+ * if need_signing is false (because bit-rot version and signature
+ * are present), then still go ahead and sign it.
+ */
+ if (!need_signing) {
+ need_signing = _gf_true;
+ need_reopen = _gf_true;
+ }
}
if (!need_signing)
@@ -1054,7 +1067,7 @@ bitd_oneshot_crawl(xlator_t *subvol, gf_dirent_t *entry, loc_t *parent,
gf_msg(this->name, GF_LOG_INFO, 0, BRB_MSG_TRIGGER_SIGN,
"Triggering signing for %s [GFID: %s | Brick: %s]", loc.path,
uuid_utoa(linked_inode->gfid), child->brick_path);
- br_trigger_sign(this, child, linked_inode, &loc, _gf_true);
+ br_trigger_sign(this, child, linked_inode, &loc, need_reopen);
ret = 0;
--
1.8.3.1

152
SOURCES/0460-cluster-ec-Remove-stale-entries-from-indices-xattrop.patch Normal file
View File

@ -0,0 +1,152 @@
From b166826b283d9071532174ebbec857dea600064b Mon Sep 17 00:00:00 2001
From: Ashish Pandey <aspandey@redhat.com>
Date: Thu, 23 Jul 2020 11:07:32 +0530
Subject: [PATCH 460/465] cluster/ec: Remove stale entries from indices/xattrop
folder
Problem:
If a gfid is present in indices/xattrop folder while
the file/dir is actaully healthy and all the xattrs are healthy,
it causes lot of lookups by shd on an entry which does not need
to be healed.
This whole process eats up lot of CPU usage without doing meaningful
work.
Solution:
Set trusted.ec.dirty xattr of the entry so that actual heal process
happens and at the end of it, during unset of dirty, gfid enrty from
indices/xattrop will be removed.
>Upstream patch : https://review.gluster.org/#/c/glusterfs/+/24765/
>Fixes: #1385
Change-Id: Ib1b9377d8dda384bba49523e9ff6ba9f0699cc1b
BUG: 1785714
Signed-off-by: Ashish Pandey <aspandey@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/208591
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/cluster/ec/src/ec-heal.c | 73 ++++++++++++++++++++++++++++++++++++++-
xlators/cluster/ec/src/ec-types.h | 7 +++-
2 files changed, 78 insertions(+), 2 deletions(-)
diff --git a/xlators/cluster/ec/src/ec-heal.c b/xlators/cluster/ec/src/ec-heal.c
index e2de879..7d25853 100644
--- a/xlators/cluster/ec/src/ec-heal.c
+++ b/xlators/cluster/ec/src/ec-heal.c
@@ -2488,6 +2488,59 @@ out:
return ret;
}
+int
+ec_heal_set_dirty_without_lock(call_frame_t *frame, ec_t *ec, inode_t *inode)
+{
+ int i = 0;
+ int ret = 0;
+ dict_t **xattr = NULL;
+ loc_t loc = {0};
+ uint64_t dirty_xattr[EC_VERSION_SIZE] = {0};
+ unsigned char *on = NULL;
+ default_args_cbk_t *replies = NULL;
+ dict_t *dict = NULL;
+
+ /* Allocate the required memory */
+ loc.inode = inode_ref(inode);
+ gf_uuid_copy(loc.gfid, inode->gfid);
+ on = alloca0(ec->nodes);
+ EC_REPLIES_ALLOC(replies, ec->nodes);
+ xattr = GF_CALLOC(ec->nodes, sizeof(*xattr), gf_common_mt_pointer);
+ if (!xattr) {
+ ret = -ENOMEM;
+ goto out;
+ }
+ dict = dict_new();
+ if (!dict) {
+ ret = -ENOMEM;
+ goto out;
+ }
+ for (i = 0; i < ec->nodes; i++) {
+ xattr[i] = dict;
+ on[i] = 1;
+ }
+ dirty_xattr[EC_METADATA_TXN] = hton64(1);
+ ret = dict_set_static_bin(dict, EC_XATTR_DIRTY, dirty_xattr,
+ (sizeof(*dirty_xattr) * EC_VERSION_SIZE));
+ if (ret < 0) {
+ ret = -ENOMEM;
+ goto out;
+ }
+ PARALLEL_FOP_ONLIST(ec->xl_list, on, ec->nodes, replies, frame,
+ ec_wind_xattrop_parallel, &loc, GF_XATTROP_ADD_ARRAY64,
+ xattr, NULL);
+out:
+ if (dict) {
+ dict_unref(dict);
+ }
+ if (xattr) {
+ GF_FREE(xattr);
+ }
+ cluster_replies_wipe(replies, ec->nodes);
+ loc_wipe(&loc);
+ return ret;
+}
+
void
ec_heal_do(xlator_t *this, void *data, loc_t *loc, int32_t partial)
{
@@ -2563,7 +2616,18 @@ ec_heal_do(xlator_t *this, void *data, loc_t *loc, int32_t partial)
ec_heal_inspect(frame, ec, loc->inode, up_subvols, _gf_false, _gf_false,
&need_heal);
- if (need_heal == EC_HEAL_NONEED) {
+ if (need_heal == EC_HEAL_PURGE_INDEX) {
+ gf_msg(ec->xl->name, GF_LOG_INFO, 0, EC_MSG_HEAL_FAIL,
+ "Index entry needs to be purged for: %s ",
+ uuid_utoa(loc->gfid));
+ /* We need to send xattrop to set dirty flag so that it can be
+ * healed and index entry could be removed. We need not to take lock
+ * on this entry to do so as we are just setting dirty flag which
+ * actually increases the trusted.ec.dirty count and does not set
+ * the new value.
+ * This will make sure that it is not interfering in other fops.*/
+ ec_heal_set_dirty_without_lock(frame, ec, loc->inode);
+ } else if (need_heal == EC_HEAL_NONEED) {
gf_msg(ec->xl->name, GF_LOG_DEBUG, 0, EC_MSG_HEAL_FAIL,
"Heal is not required for : %s ", uuid_utoa(loc->gfid));
goto out;
@@ -2958,6 +3022,13 @@ _need_heal_calculate(ec_t *ec, uint64_t *dirty, unsigned char *sources,
goto out;
}
}
+ /* If lock count is 0, all dirty flags are 0 and all the
+ * versions are macthing then why are we here. It looks
+ * like something went wrong while removing the index entries
+ * after completing a successful heal or fop. In this case
+ * we need to remove this index entry to avoid triggering heal
+ * in a loop and causing lookups again and again*/
+ *need_heal = EC_HEAL_PURGE_INDEX;
} else {
for (i = 0; i < ec->nodes; i++) {
/* Since each lock can only increment the dirty
diff --git a/xlators/cluster/ec/src/ec-types.h b/xlators/cluster/ec/src/ec-types.h
index f15429d..700dc39 100644
--- a/xlators/cluster/ec/src/ec-types.h
+++ b/xlators/cluster/ec/src/ec-types.h
@@ -130,7 +130,12 @@ typedef void (*ec_resume_f)(ec_fop_data_t *, int32_t);
enum _ec_read_policy { EC_ROUND_ROBIN, EC_GFID_HASH, EC_READ_POLICY_MAX };
-enum _ec_heal_need { EC_HEAL_NONEED, EC_HEAL_MAYBE, EC_HEAL_MUST };
+enum _ec_heal_need {
+ EC_HEAL_NONEED,
+ EC_HEAL_MAYBE,
+ EC_HEAL_MUST,
+ EC_HEAL_PURGE_INDEX
+};
enum _ec_stripe_part { EC_STRIPE_HEAD, EC_STRIPE_TAIL };
--
1.8.3.1

127
SOURCES/0461-geo-replication-Fix-IPv6-parsing.patch Normal file
View File

@ -0,0 +1,127 @@
From d425ed54261d5bc19aa853854cc3b64647e3c897 Mon Sep 17 00:00:00 2001
From: Aravinda Vishwanathapura <aravinda@kadalu.io>
Date: Sun, 12 Jul 2020 12:42:36 +0530
Subject: [PATCH 461/465] geo-replication: Fix IPv6 parsing
Brick paths in Volinfo used `:` as delimiter, Geo-rep uses split
based on `:` char. This will go wrong with IPv6.
This patch handles the IPv6 case and handles the split properly.
Backport of:
>Upstream Patch: https://review.gluster.org/#/c/glusterfs/+/24706
>Fixes: #1366
>Change-Id: I25e88d693744381c0ccf3c1dbf1541b84be2499d
>Signed-off-by: Aravinda Vishwanathapura <aravinda@kadalu.io>
BUG: 1855966
Change-Id: I25e88d693744381c0ccf3c1dbf1541b84be2499d
Signed-off-by: Sunny Kumar <sunkumar@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/208610
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
geo-replication/syncdaemon/master.py | 5 ++--
geo-replication/syncdaemon/syncdutils.py | 43 +++++++++++++++++++++++++++++---
2 files changed, 43 insertions(+), 5 deletions(-)
diff --git a/geo-replication/syncdaemon/master.py b/geo-replication/syncdaemon/master.py
index 3f98337..08e98f8 100644
--- a/geo-replication/syncdaemon/master.py
+++ b/geo-replication/syncdaemon/master.py
@@ -26,7 +26,8 @@ from rconf import rconf
from syncdutils import Thread, GsyncdError, escape_space_newline
from syncdutils import unescape_space_newline, gauxpfx, escape
from syncdutils import lstat, errno_wrap, FreeObject, lf, matching_disk_gfid
-from syncdutils import NoStimeAvailable, PartialHistoryAvailable
+from syncdutils import NoStimeAvailable, PartialHistoryAvailable, host_brick_split
+
URXTIME = (-1, 0)
@@ -1466,7 +1467,7 @@ class GMasterChangelogMixin(GMasterCommon):
node = rconf.args.resource_remote
node_data = node.split("@")
node = node_data[-1]
- remote_node_ip = node.split(":")[0]
+ remote_node_ip, _ = host_brick_split(node)
self.status.set_slave_node(remote_node_ip)
def changelogs_batch_process(self, changes):
diff --git a/geo-replication/syncdaemon/syncdutils.py b/geo-replication/syncdaemon/syncdutils.py
index 7560fa1..f43e13b 100644
--- a/geo-replication/syncdaemon/syncdutils.py
+++ b/geo-replication/syncdaemon/syncdutils.py
@@ -883,6 +883,19 @@ class Popen(subprocess.Popen):
self.errfail()
+def host_brick_split(value):
+ """
+ IPv6 compatible way to split and get the host
+ and brick information. Example inputs:
+ node1.example.com:/exports/bricks/brick1/brick
+ fe80::af0f:df82:844f:ef66%utun0:/exports/bricks/brick1/brick
+ """
+ parts = value.split(":")
+ brick = parts[-1]
+ hostparts = parts[0:-1]
+ return (":".join(hostparts), brick)
+
+
class Volinfo(object):
def __init__(self, vol, host='localhost', prelude=[], master=True):
@@ -925,7 +938,7 @@ class Volinfo(object):
@memoize
def bricks(self):
def bparse(b):
- host, dirp = b.find("name").text.split(':', 2)
+ host, dirp = host_brick_split(b.find("name").text)
return {'host': host, 'dir': dirp, 'uuid': b.find("hostUuid").text}
return [bparse(b) for b in self.get('brick')]
@@ -1001,6 +1014,16 @@ class VolinfoFromGconf(object):
def is_hot(self, brickpath):
return False
+ def is_uuid(self, value):
+ try:
+ uuid.UUID(value)
+ return True
+ except ValueError:
+ return False
+
+ def possible_path(self, value):
+ return "/" in value
+
@property
@memoize
def bricks(self):
@@ -1014,8 +1037,22 @@ class VolinfoFromGconf(object):
out = []
for b in bricks_data:
parts = b.split(":")
- bpath = parts[2] if len(parts) == 3 else ""
- out.append({"host": parts[1], "dir": bpath, "uuid": parts[0]})
+ b_uuid = None
+ if self.is_uuid(parts[0]):
+ b_uuid = parts[0]
+ # Set all parts except first
+ parts = parts[1:]
+
+ if self.possible_path(parts[-1]):
+ bpath = parts[-1]
+ # Set all parts except last
+ parts = parts[0:-1]
+
+ out.append({
+ "host": ":".join(parts), # if remaining parts are IPv6 name
+ "dir": bpath,
+ "uuid": b_uuid
+ })
return out
--
1.8.3.1

43
SOURCES/0462-Issue-with-gf_fill_iatt_for_dirent.patch Normal file
View File

@ -0,0 +1,43 @@
From f027734165374979bd0bff8ea059dfaadca85e07 Mon Sep 17 00:00:00 2001
From: Soumya Koduri <skoduri@redhat.com>
Date: Thu, 2 Jul 2020 02:07:56 +0530
Subject: [PATCH 462/465] Issue with gf_fill_iatt_for_dirent
In "gf_fill_iatt_for_dirent()", while calculating inode_path for loc,
the inode should be of parent's. Instead it is loc.inode which results in error
and eventually lookup/readdirp fails.
This patch fixes the same.
This is backport of below mainstream fix :
> Change-Id: Ied086234a4634e8cb13520521ac547c87b3c76b5
> Fixes: #1351
> Upstream patch: https://review.gluster.org/#/c/glusterfs/+/24661/
Change-Id: Ied086234a4634e8cb13520521ac547c87b3c76b5
BUG: 1853189
Signed-off-by: Soumya Koduri <skoduri@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/208691
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
libglusterfs/src/gf-dirent.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/libglusterfs/src/gf-dirent.c b/libglusterfs/src/gf-dirent.c
index f289723..3fa67f2 100644
--- a/libglusterfs/src/gf-dirent.c
+++ b/libglusterfs/src/gf-dirent.c
@@ -277,7 +277,7 @@ gf_fill_iatt_for_dirent(gf_dirent_t *entry, inode_t *parent, xlator_t *subvol)
gf_uuid_copy(loc.pargfid, parent->gfid);
loc.name = entry->d_name;
loc.parent = inode_ref(parent);
- ret = inode_path(loc.inode, entry->d_name, &path);
+ ret = inode_path(loc.parent, entry->d_name, &path);
loc.path = path;
if (ret < 0)
goto out;
--
1.8.3.1

87
SOURCES/0463-cluster-ec-Change-handling-of-heal-failure-to-avoid-.patch Normal file
View File

@ -0,0 +1,87 @@
From 7d87933f648092ae55d57a96fd06e3df975d764c Mon Sep 17 00:00:00 2001
From: Ashish Pandey <aspandey@redhat.com>
Date: Tue, 18 Aug 2020 10:33:48 +0530
Subject: [PATCH 463/465] cluster/ec: Change handling of heal failure to avoid
crash
Problem:
ec_getxattr_heal_cbk was called with NULL as second argument
in case heal was failing.
This function was dereferencing "cookie" argument which caused crash.
Solution:
Cookie is changed to carry the value that was supposed to be
stored in fop->data, so even in the case when fop is NULL in error
case, there won't be any NULL dereference.
Thanks to Xavi for the suggestion about the fix.
>Upstream patch: https://review.gluster.org/#/c/glusterfs/+/23050/
>fixes: bz#1729085
Change-Id: I0798000d5cadb17c3c2fbfa1baf77033ffc2bb8c
BUG: 1852736
Reviewed-on: https://code.engineering.redhat.com/gerrit/209012
Tested-by: Ashish Pandey <aspandey@redhat.com>
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Xavi Hernandez Juan <xhernandez@redhat.com>
---
xlators/cluster/ec/src/ec-heal.c | 11 ++++++-----
xlators/cluster/ec/src/ec-inode-read.c | 4 ++--
2 files changed, 8 insertions(+), 7 deletions(-)
diff --git a/xlators/cluster/ec/src/ec-heal.c b/xlators/cluster/ec/src/ec-heal.c
index 7d25853..6e6948b 100644
--- a/xlators/cluster/ec/src/ec-heal.c
+++ b/xlators/cluster/ec/src/ec-heal.c
@@ -1966,7 +1966,7 @@ ec_manager_heal_block(ec_fop_data_t *fop, int32_t state)
case EC_STATE_REPORT:
if (fop->cbks.heal) {
- fop->cbks.heal(fop->req_frame, fop, fop->xl, 0, 0,
+ fop->cbks.heal(fop->req_frame, fop->data, fop->xl, 0, 0,
(heal->good | heal->bad), heal->good, heal->bad,
0, NULL);
}
@@ -2022,10 +2022,11 @@ ec_heal_block_done(call_frame_t *frame, void *cookie, xlator_t *this,
uintptr_t good, uintptr_t bad, uint32_t pending,
dict_t *xdata)
{
- ec_fop_data_t *fop = cookie;
- ec_heal_t *heal = fop->data;
+ ec_heal_t *heal = cookie;
- fop->heal = NULL;
+ if (heal->fop) {
+ heal->fop->heal = NULL;
+ }
heal->fop = NULL;
heal->error = op_ret < 0 ? op_errno : 0;
syncbarrier_wake(heal->data);
@@ -2669,7 +2670,7 @@ ec_heal_do(xlator_t *this, void *data, loc_t *loc, int32_t partial)
out:
ec_reset_entry_healing(fop);
if (fop->cbks.heal) {
- fop->cbks.heal(fop->req_frame, fop, fop->xl, op_ret, op_errno,
+ fop->cbks.heal(fop->req_frame, fop->data, fop->xl, op_ret, op_errno,
ec_char_array_to_mask(participants, ec->nodes),
mgood & good, mbad & bad, pending, NULL);
}
diff --git a/xlators/cluster/ec/src/ec-inode-read.c b/xlators/cluster/ec/src/ec-inode-read.c
index e82e8f6..c50d0ad 100644
--- a/xlators/cluster/ec/src/ec-inode-read.c
+++ b/xlators/cluster/ec/src/ec-inode-read.c
@@ -396,8 +396,8 @@ ec_getxattr_heal_cbk(call_frame_t *frame, void *cookie, xlator_t *xl,
uintptr_t good, uintptr_t bad, uint32_t pending,
dict_t *xdata)
{
- ec_fop_data_t *fop = cookie;
- fop_getxattr_cbk_t func = fop->data;
+ fop_getxattr_cbk_t func = cookie;
+
ec_t *ec = xl->private;
dict_t *dict = NULL;
char *str;
--
1.8.3.1

102
SOURCES/0464-storage-posix-Remove-nr_files-usage.patch Normal file
View File

@ -0,0 +1,102 @@
From 7c51addf7912a94320e6b148bd66f2dbf274c533 Mon Sep 17 00:00:00 2001
From: Pranith Kumar K <pkarampu@redhat.com>
Date: Mon, 11 Mar 2019 14:04:39 +0530
Subject: [PATCH 464/465] storage/posix: Remove nr_files usage
nr_files is supposed to represent the number of files opened in posix.
Present logic doesn't seem to handle anon-fds because of which the
counts would always be wrong.
I don't remember anyone using this value in debugging any problem probably
because we always have 'ls -l /proc/<pid>/fd' which not only prints the
fds that are active but also prints their paths. It also handles directories
and anon-fds which actually opened the file. So removing this code
instead of fixing the buggy logic to have the nr_files.
> fixes bz#1688106
> Change-Id: Ibf8713fdfdc1ef094e08e6818152637206a54040
> Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
> (Cherry pick from commit f5987d38f216a3142dfe45f03bf66ff4827d9b55)
> (Reviewed on upstream link https://review.gluster.org/#/c/glusterfs/+/22333/)
Change-Id: Ibf8713fdfdc1ef094e08e6818152637206a54040
BUG: 1851989
Signed-off-by: Mohit Agrawal<moagrawa@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/209468
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/storage/posix/src/posix-common.c | 2 --
xlators/storage/posix/src/posix-entry-ops.c | 2 --
xlators/storage/posix/src/posix-inode-fd-ops.c | 2 --
xlators/storage/posix/src/posix.h | 1 -
4 files changed, 7 deletions(-)
diff --git a/xlators/storage/posix/src/posix-common.c b/xlators/storage/posix/src/posix-common.c
index ac53796..b317627 100644
--- a/xlators/storage/posix/src/posix-common.c
+++ b/xlators/storage/posix/src/posix-common.c
@@ -128,7 +128,6 @@ posix_priv(xlator_t *this)
gf_proc_dump_write("max_read", "%" PRId64, GF_ATOMIC_GET(priv->read_value));
gf_proc_dump_write("max_write", "%" PRId64,
GF_ATOMIC_GET(priv->write_value));
- gf_proc_dump_write("nr_files", "%" PRId64, GF_ATOMIC_GET(priv->nr_files));
return 0;
}
@@ -815,7 +814,6 @@ posix_init(xlator_t *this)
}
LOCK_INIT(&_private->lock);
- GF_ATOMIC_INIT(_private->nr_files, 0);
GF_ATOMIC_INIT(_private->read_value, 0);
GF_ATOMIC_INIT(_private->write_value, 0);
diff --git a/xlators/storage/posix/src/posix-entry-ops.c b/xlators/storage/posix/src/posix-entry-ops.c
index 65650b3..b3a5381 100644
--- a/xlators/storage/posix/src/posix-entry-ops.c
+++ b/xlators/storage/posix/src/posix-entry-ops.c
@@ -2243,8 +2243,6 @@ fill_stat:
gf_msg(this->name, GF_LOG_WARNING, 0, P_MSG_FD_PATH_SETTING_FAILED,
"failed to set the fd context path=%s fd=%p", real_path, fd);
- GF_ATOMIC_INC(priv->nr_files);
-
op_ret = 0;
out:
diff --git a/xlators/storage/posix/src/posix-inode-fd-ops.c b/xlators/storage/posix/src/posix-inode-fd-ops.c
index d135d8b..81f4a6b 100644
--- a/xlators/storage/posix/src/posix-inode-fd-ops.c
+++ b/xlators/storage/posix/src/posix-inode-fd-ops.c
@@ -1605,7 +1605,6 @@ posix_open(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
gf_msg(this->name, GF_LOG_WARNING, 0, P_MSG_FD_PATH_SETTING_FAILED,
"failed to set the fd context path=%s fd=%p", real_path, fd);
- GF_ATOMIC_INC(priv->nr_files);
op_ret = 0;
out:
@@ -2526,7 +2525,6 @@ posix_release(xlator_t *this, fd_t *fd)
if (!priv)
goto out;
- GF_ATOMIC_DEC(priv->nr_files);
out:
return 0;
}
diff --git a/xlators/storage/posix/src/posix.h b/xlators/storage/posix/src/posix.h
index 61495a7..124dbb4 100644
--- a/xlators/storage/posix/src/posix.h
+++ b/xlators/storage/posix/src/posix.h
@@ -154,7 +154,6 @@ struct posix_private {
gf_atomic_t read_value; /* Total read, from init */
gf_atomic_t write_value; /* Total write, from init */
- gf_atomic_t nr_files;
/*
In some cases, two exported volumes may reside on the same
partition on the server. Sending statvfs info for both
--
1.8.3.1

384
SOURCES/0465-posix-Implement-a-janitor-thread-to-close-fd.patch Normal file
View File

@ -0,0 +1,384 @@
From 143b93b230b429cc712353243ed794b68494c040 Mon Sep 17 00:00:00 2001
From: Mohit Agrawal <moagrawa@redhat.com>
Date: Mon, 27 Jul 2020 18:08:00 +0530
Subject: [PATCH 465/465] posix: Implement a janitor thread to close fd
Problem: In the commit fb20713b380e1df8d7f9e9df96563be2f9144fd6 we use
syntask to close fd but we have found the patch is reducing the
performance
Solution: Use janitor thread to close fd's and save the pfd ctx into
ctx janitor list and also save the posix_xlator into pfd object to
avoid the race condition during cleanup in brick_mux environment
> Change-Id: Ifb3d18a854b267333a3a9e39845bfefb83fbc092
> Fixes: #1396
> Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
> (Reviewed on upstream link https://review.gluster.org/#/c/glusterfs/+/24755/)
> (Cherry pick from commit 41b9616435cbdf671805856e487e373060c9455b
Change-Id: Ifb3d18a854b267333a3a9e39845bfefb83fbc092
BUG: 1851989
Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/209448
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
glusterfsd/src/glusterfsd.c | 4 ++
libglusterfs/src/glusterfs/glusterfs.h | 7 ++
rpc/rpc-lib/src/rpcsvc.c | 6 --
xlators/storage/posix/src/posix-common.c | 34 +++++++++-
xlators/storage/posix/src/posix-helpers.c | 93 ++++++++++++++++++++++++++
xlators/storage/posix/src/posix-inode-fd-ops.c | 33 ++++-----
xlators/storage/posix/src/posix.h | 7 ++
7 files changed, 161 insertions(+), 23 deletions(-)
diff --git a/glusterfsd/src/glusterfsd.c b/glusterfsd/src/glusterfsd.c
index 9821180..955bf1d 100644
--- a/glusterfsd/src/glusterfsd.c
+++ b/glusterfsd/src/glusterfsd.c
@@ -1839,6 +1839,10 @@ glusterfs_ctx_defaults_init(glusterfs_ctx_t *ctx)
INIT_LIST_HEAD(&cmd_args->xlator_options);
INIT_LIST_HEAD(&cmd_args->volfile_servers);
+ ctx->pxl_count = 0;
+ pthread_mutex_init(&ctx->fd_lock, NULL);
+ pthread_cond_init(&ctx->fd_cond, NULL);
+ INIT_LIST_HEAD(&ctx->janitor_fds);
lim.rlim_cur = RLIM_INFINITY;
lim.rlim_max = RLIM_INFINITY;
diff --git a/libglusterfs/src/glusterfs/glusterfs.h b/libglusterfs/src/glusterfs/glusterfs.h
index 495a4d7..bf6a987 100644
--- a/libglusterfs/src/glusterfs/glusterfs.h
+++ b/libglusterfs/src/glusterfs/glusterfs.h
@@ -733,6 +733,13 @@ struct _glusterfs_ctx {
} stats;
struct list_head volfile_list;
+ /* Add members to manage janitor threads for cleanup fd */
+ struct list_head janitor_fds;
+ pthread_cond_t fd_cond;
+ pthread_mutex_t fd_lock;
+ pthread_t janitor;
+ /* The variable is use to save total posix xlator count */
+ uint32_t pxl_count;
char volume_id[GF_UUID_BUF_SIZE]; /* Used only in protocol/client */
};
diff --git a/rpc/rpc-lib/src/rpcsvc.c b/rpc/rpc-lib/src/rpcsvc.c
index 23ca1fd..3f184bf 100644
--- a/rpc/rpc-lib/src/rpcsvc.c
+++ b/rpc/rpc-lib/src/rpcsvc.c
@@ -375,12 +375,6 @@ rpcsvc_program_actor(rpcsvc_request_t *req)
req->ownthread = program->ownthread;
req->synctask = program->synctask;
- if (((req->procnum == GFS3_OP_RELEASE) ||
- (req->procnum == GFS3_OP_RELEASEDIR)) &&
- (program->prognum == GLUSTER_FOP_PROGRAM)) {
- req->ownthread = _gf_false;
- req->synctask = _gf_true;
- }
err = SUCCESS;
gf_log(GF_RPCSVC, GF_LOG_TRACE, "Actor found: %s - %s for %s",
diff --git a/xlators/storage/posix/src/posix-common.c b/xlators/storage/posix/src/posix-common.c
index b317627..c5a43a1 100644
--- a/xlators/storage/posix/src/posix-common.c
+++ b/xlators/storage/posix/src/posix-common.c
@@ -150,6 +150,7 @@ posix_notify(xlator_t *this, int32_t event, void *data, ...)
struct timespec sleep_till = {
0,
};
+ glusterfs_ctx_t *ctx = this->ctx;
switch (event) {
case GF_EVENT_PARENT_UP: {
@@ -160,8 +161,6 @@ posix_notify(xlator_t *this, int32_t event, void *data, ...)
case GF_EVENT_PARENT_DOWN: {
if (!victim->cleanup_starting)
break;
- gf_log(this->name, GF_LOG_INFO, "Sending CHILD_DOWN for brick %s",
- victim->name);
if (priv->janitor) {
pthread_mutex_lock(&priv->janitor_mutex);
@@ -187,6 +186,16 @@ posix_notify(xlator_t *this, int32_t event, void *data, ...)
GF_FREE(priv->janitor);
}
priv->janitor = NULL;
+ pthread_mutex_lock(&ctx->fd_lock);
+ {
+ while (priv->rel_fdcount > 0) {
+ pthread_cond_wait(&priv->fd_cond, &ctx->fd_lock);
+ }
+ }
+ pthread_mutex_unlock(&ctx->fd_lock);
+
+ gf_log(this->name, GF_LOG_INFO, "Sending CHILD_DOWN for brick %s",
+ victim->name);
default_notify(this->parents->xlator, GF_EVENT_CHILD_DOWN, data);
} break;
default:
@@ -1038,7 +1047,13 @@ posix_init(xlator_t *this)
pthread_cond_init(&_private->fsync_cond, NULL);
pthread_mutex_init(&_private->janitor_mutex, NULL);
pthread_cond_init(&_private->janitor_cond, NULL);
+ pthread_cond_init(&_private->fd_cond, NULL);
INIT_LIST_HEAD(&_private->fsyncs);
+ _private->rel_fdcount = 0;
+ ret = posix_spawn_ctx_janitor_thread(this);
+ if (ret)
+ goto out;
+
ret = gf_thread_create(&_private->fsyncer, NULL, posix_fsyncer, this,
"posixfsy");
if (ret) {
@@ -1133,6 +1148,8 @@ posix_fini(xlator_t *this)
{
struct posix_private *priv = this->private;
gf_boolean_t health_check = _gf_false;
+ glusterfs_ctx_t *ctx = this->ctx;
+ uint32_t count;
int ret = 0;
if (!priv)
@@ -1166,6 +1183,19 @@ posix_fini(xlator_t *this)
priv->janitor = NULL;
}
+ pthread_mutex_lock(&ctx->fd_lock);
+ {
+ count = --ctx->pxl_count;
+ if (count == 0) {
+ pthread_cond_signal(&ctx->fd_cond);
+ }
+ }
+ pthread_mutex_unlock(&ctx->fd_lock);
+
+ if (count == 0) {
+ pthread_join(ctx->janitor, NULL);
+ }
+
if (priv->fsyncer) {
(void)gf_thread_cleanup_xint(priv->fsyncer);
priv->fsyncer = 0;
diff --git a/xlators/storage/posix/src/posix-helpers.c b/xlators/storage/posix/src/posix-helpers.c
index 39dbcce..73a44be 100644
--- a/xlators/storage/posix/src/posix-helpers.c
+++ b/xlators/storage/posix/src/posix-helpers.c
@@ -1582,6 +1582,99 @@ unlock:
return;
}
+static struct posix_fd *
+janitor_get_next_fd(glusterfs_ctx_t *ctx)
+{
+ struct posix_fd *pfd = NULL;
+
+ while (list_empty(&ctx->janitor_fds)) {
+ if (ctx->pxl_count == 0) {
+ return NULL;
+ }
+
+ pthread_cond_wait(&ctx->fd_cond, &ctx->fd_lock);
+ }
+
+ pfd = list_first_entry(&ctx->janitor_fds, struct posix_fd, list);
+ list_del_init(&pfd->list);
+
+ return pfd;
+}
+
+static void
+posix_close_pfd(xlator_t *xl, struct posix_fd *pfd)
+{
+ THIS = xl;
+
+ if (pfd->dir == NULL) {
+ gf_msg_trace(xl->name, 0, "janitor: closing file fd=%d", pfd->fd);
+ sys_close(pfd->fd);
+ } else {
+ gf_msg_debug(xl->name, 0, "janitor: closing dir fd=%p", pfd->dir);
+ sys_closedir(pfd->dir);
+ }
+
+ GF_FREE(pfd);
+}
+
+static void *
+posix_ctx_janitor_thread_proc(void *data)
+{
+ xlator_t *xl;
+ struct posix_fd *pfd;
+ glusterfs_ctx_t *ctx = NULL;
+ struct posix_private *priv_fd;
+
+ ctx = data;
+
+ pthread_mutex_lock(&ctx->fd_lock);
+
+ while ((pfd = janitor_get_next_fd(ctx)) != NULL) {
+ pthread_mutex_unlock(&ctx->fd_lock);
+
+ xl = pfd->xl;
+ posix_close_pfd(xl, pfd);
+
+ pthread_mutex_lock(&ctx->fd_lock);
+
+ priv_fd = xl->private;
+ priv_fd->rel_fdcount--;
+ if (!priv_fd->rel_fdcount)
+ pthread_cond_signal(&priv_fd->fd_cond);
+ }
+
+ pthread_mutex_unlock(&ctx->fd_lock);
+
+ return NULL;
+}
+
+int
+posix_spawn_ctx_janitor_thread(xlator_t *this)
+{
+ int ret = 0;
+ glusterfs_ctx_t *ctx = NULL;
+
+ ctx = this->ctx;
+
+ pthread_mutex_lock(&ctx->fd_lock);
+ {
+ if (ctx->pxl_count++ == 0) {
+ ret = gf_thread_create(&ctx->janitor, NULL,
+ posix_ctx_janitor_thread_proc, ctx,
+ "posixctxjan");
+
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, errno, P_MSG_THREAD_FAILED,
+ "spawning janitor thread failed");
+ ctx->pxl_count--;
+ }
+ }
+ }
+ pthread_mutex_unlock(&ctx->fd_lock);
+
+ return ret;
+}
+
static int
is_fresh_file(int64_t ctime_sec)
{
diff --git a/xlators/storage/posix/src/posix-inode-fd-ops.c b/xlators/storage/posix/src/posix-inode-fd-ops.c
index 81f4a6b..21119ea 100644
--- a/xlators/storage/posix/src/posix-inode-fd-ops.c
+++ b/xlators/storage/posix/src/posix-inode-fd-ops.c
@@ -1352,6 +1352,22 @@ out:
return 0;
}
+static void
+posix_add_fd_to_cleanup(xlator_t *this, struct posix_fd *pfd)
+{
+ glusterfs_ctx_t *ctx = this->ctx;
+ struct posix_private *priv = this->private;
+
+ pfd->xl = this;
+ pthread_mutex_lock(&ctx->fd_lock);
+ {
+ list_add_tail(&pfd->list, &ctx->janitor_fds);
+ priv->rel_fdcount++;
+ pthread_cond_signal(&ctx->fd_cond);
+ }
+ pthread_mutex_unlock(&ctx->fd_lock);
+}
+
int32_t
posix_releasedir(xlator_t *this, fd_t *fd)
{
@@ -1374,11 +1390,7 @@ posix_releasedir(xlator_t *this, fd_t *fd)
"pfd->dir is NULL for fd=%p", fd);
goto out;
}
-
- gf_msg_debug(this->name, 0, "janitor: closing dir fd=%p", pfd->dir);
-
- sys_closedir(pfd->dir);
- GF_FREE(pfd);
+ posix_add_fd_to_cleanup(this, pfd);
out:
return 0;
@@ -2494,7 +2506,6 @@ out:
int32_t
posix_release(xlator_t *this, fd_t *fd)
{
- struct posix_private *priv = NULL;
struct posix_fd *pfd = NULL;
int ret = -1;
uint64_t tmp_pfd = 0;
@@ -2502,8 +2513,6 @@ posix_release(xlator_t *this, fd_t *fd)
VALIDATE_OR_GOTO(this, out);
VALIDATE_OR_GOTO(fd, out);
- priv = this->private;
-
ret = fd_ctx_del(fd, this, &tmp_pfd);
if (ret < 0) {
gf_msg(this->name, GF_LOG_WARNING, 0, P_MSG_PFD_NULL,
@@ -2517,13 +2526,7 @@ posix_release(xlator_t *this, fd_t *fd)
"pfd->dir is %p (not NULL) for file fd=%p", pfd->dir, fd);
}
- gf_msg_debug(this->name, 0, "janitor: closing dir fd=%p", pfd->dir);
-
- sys_close(pfd->fd);
- GF_FREE(pfd);
-
- if (!priv)
- goto out;
+ posix_add_fd_to_cleanup(this, pfd);
out:
return 0;
diff --git a/xlators/storage/posix/src/posix.h b/xlators/storage/posix/src/posix.h
index 124dbb4..07f367b 100644
--- a/xlators/storage/posix/src/posix.h
+++ b/xlators/storage/posix/src/posix.h
@@ -134,6 +134,8 @@ struct posix_fd {
off_t dir_eof; /* offset at dir EOF */
int odirect;
struct list_head list; /* to add to the janitor list */
+ xlator_t *xl;
+ char _pad[4]; /* manual padding */
};
struct posix_private {
@@ -204,6 +206,7 @@ struct posix_private {
pthread_cond_t fsync_cond;
pthread_mutex_t janitor_mutex;
pthread_cond_t janitor_cond;
+ pthread_cond_t fd_cond;
int fsync_queue_count;
enum {
@@ -259,6 +262,7 @@ struct posix_private {
gf_boolean_t fips_mode_rchecksum;
gf_boolean_t ctime;
gf_boolean_t janitor_task_stop;
+ uint32_t rel_fdcount;
};
typedef struct {
@@ -665,6 +669,9 @@ posix_cs_maintenance(xlator_t *this, fd_t *fd, loc_t *loc, int *pfd,
int
posix_check_dev_file(xlator_t *this, inode_t *inode, char *fop, int *op_errno);
+int
+posix_spawn_ctx_janitor_thread(xlator_t *this);
+
void
posix_update_iatt_buf(struct iatt *buf, int fd, char *loc, dict_t *xdata);
--
1.8.3.1

68
SOURCES/0466-cluster-ec-Change-stale-index-handling.patch Normal file
View File

@ -0,0 +1,68 @@
From b603170ae5f583037b8177a9d19e56c7821edf0b Mon Sep 17 00:00:00 2001
From: Pranith Kumar K <pkarampu@redhat.com>
Date: Tue, 25 Aug 2020 04:19:54 +0530
Subject: [PATCH 466/466] cluster/ec: Change stale index handling
Problem:
Earlier approach is setting dirty bit which requires extra heal
Fix:
Send zero-xattrop which deletes stale index without any need
for extra heal.
> Fixes: #1385
> Upstream-patch: https://review.gluster.org/c/glusterfs/+/24911/
BUG: 1785714
Change-Id: I7e97a1d8b5516f7be47cae55d0e56b14332b6cae
Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/209904
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Tested-by: Ashish Pandey <aspandey@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/cluster/ec/src/ec-heal.c | 14 +++++---------
1 file changed, 5 insertions(+), 9 deletions(-)
diff --git a/xlators/cluster/ec/src/ec-heal.c b/xlators/cluster/ec/src/ec-heal.c
index 6e6948b..06bafa5 100644
--- a/xlators/cluster/ec/src/ec-heal.c
+++ b/xlators/cluster/ec/src/ec-heal.c
@@ -2490,7 +2490,7 @@ out:
}
int
-ec_heal_set_dirty_without_lock(call_frame_t *frame, ec_t *ec, inode_t *inode)
+ec_heal_purge_stale_index(call_frame_t *frame, ec_t *ec, inode_t *inode)
{
int i = 0;
int ret = 0;
@@ -2520,7 +2520,6 @@ ec_heal_set_dirty_without_lock(call_frame_t *frame, ec_t *ec, inode_t *inode)
xattr[i] = dict;
on[i] = 1;
}
- dirty_xattr[EC_METADATA_TXN] = hton64(1);
ret = dict_set_static_bin(dict, EC_XATTR_DIRTY, dirty_xattr,
(sizeof(*dirty_xattr) * EC_VERSION_SIZE));
if (ret < 0) {
@@ -2621,13 +2620,10 @@ ec_heal_do(xlator_t *this, void *data, loc_t *loc, int32_t partial)
gf_msg(ec->xl->name, GF_LOG_INFO, 0, EC_MSG_HEAL_FAIL,
"Index entry needs to be purged for: %s ",
uuid_utoa(loc->gfid));
- /* We need to send xattrop to set dirty flag so that it can be
- * healed and index entry could be removed. We need not to take lock
- * on this entry to do so as we are just setting dirty flag which
- * actually increases the trusted.ec.dirty count and does not set
- * the new value.
- * This will make sure that it is not interfering in other fops.*/
- ec_heal_set_dirty_without_lock(frame, ec, loc->inode);
+ /* We need to send zero-xattrop so that stale index entry could be
+ * removed. We need not take lock on this entry to do so as
+ * xattrop on a brick is atomic. */
+ ec_heal_purge_stale_index(frame, ec, loc->inode);
} else if (need_heal == EC_HEAL_NONEED) {
gf_msg(ec->xl->name, GF_LOG_DEBUG, 0, EC_MSG_HEAL_FAIL,
"Heal is not required for : %s ", uuid_utoa(loc->gfid));
--
1.8.3.1

38
SOURCES/0467-build-Added-dependency-for-glusterfs-selinux.patch Normal file
View File

@ -0,0 +1,38 @@
From 9176ee8f10c3c33f31d00261995ed27e8680934a Mon Sep 17 00:00:00 2001
From: Rinku Kothiya <rkothiya@redhat.com>
Date: Thu, 3 Sep 2020 11:46:38 +0000
Subject: [PATCH 467/467] build: Added dependency for glusterfs-selinux
> Fixes: #1442
> Upstream-patch: https://review.gluster.org/#/c/glusterfs/+/24876/
> Change-Id: I7d79bceff329db4d525bc8a77ba7ffe41bf53c97
> Signed-off-by: Rinku Kothiya <rkothiya@redhat.com>
BUG: 1460657
Change-Id: I7d79bceff329db4d525bc8a77ba7ffe41bf53c97
Signed-off-by: Rinku Kothiya <rkothiya@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/210637
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
glusterfs.spec.in | 3 +++
1 file changed, 3 insertions(+)
diff --git a/glusterfs.spec.in b/glusterfs.spec.in
index 9def416..ed6bdf3 100644
--- a/glusterfs.spec.in
+++ b/glusterfs.spec.in
@@ -650,6 +650,9 @@ Summary: Clustered file-system server
Requires: %{name}%{?_isa} = %{version}-%{release}
Requires: %{name}-cli%{?_isa} = %{version}-%{release}
Requires: %{name}-libs%{?_isa} = %{version}-%{release}
+%if ( 0%{?fedora} && 0%{?fedora} >= 30 || ( 0%{?rhel} && 0%{?rhel} >= 8 ) )
+Requires: glusterfs-selinux >= 0.1.0-2
+%endif
# some daemons (like quota) use a fuse-mount, glusterfsd is part of -fuse
Requires: %{name}-fuse%{?_isa} = %{version}-%{release}
# self-heal daemon, rebalance, nfs-server etc. are actually clients
--
1.8.3.1

36
SOURCES/0468-build-Update-the-glusterfs-selinux-version.patch Normal file
View File

@ -0,0 +1,36 @@
From 4b72f5e7704d480bac869f7a32ac891898bb994f Mon Sep 17 00:00:00 2001
From: Rinku Kothiya <rkothiya@redhat.com>
Date: Thu, 3 Sep 2020 14:56:27 +0000
Subject: [PATCH 468/468] build: Update the glusterfs-selinux version
Updated the glusterfs-selinux version according to
the downstream official version.
Label: DOWNSTREAM ONLY
BUG: 1460657
Change-Id: I7b8bbf53f71f6f56103042950d8910f0cb63a685
Signed-off-by: Rinku Kothiya <rkothiya@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/210685
Tested-by: RHGS Build Bot <nigelb@redhat.com>
---
glusterfs.spec.in | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/glusterfs.spec.in b/glusterfs.spec.in
index ed6bdf3..30d7162 100644
--- a/glusterfs.spec.in
+++ b/glusterfs.spec.in
@@ -651,7 +651,7 @@ Requires: %{name}%{?_isa} = %{version}-%{release}
Requires: %{name}-cli%{?_isa} = %{version}-%{release}
Requires: %{name}-libs%{?_isa} = %{version}-%{release}
%if ( 0%{?fedora} && 0%{?fedora} >= 30 || ( 0%{?rhel} && 0%{?rhel} >= 8 ) )
-Requires: glusterfs-selinux >= 0.1.0-2
+Requires: glusterfs-selinux >= 1.0-1
%endif
# some daemons (like quota) use a fuse-mount, glusterfsd is part of -fuse
Requires: %{name}-fuse%{?_isa} = %{version}-%{release}
--
1.8.3.1

33
SOURCES/0469-cluster-ec-Don-t-trigger-heal-for-stale-index.patch Normal file
View File

@ -0,0 +1,33 @@
From 6fed6cfcb26e6ed3c9640c5f889629315bbd83c2 Mon Sep 17 00:00:00 2001
From: Pranith Kumar K <pkarampu@redhat.com>
Date: Mon, 31 Aug 2020 12:22:05 +0530
Subject: [PATCH 469/469] cluster/ec: Don't trigger heal for stale index
> Fixes: #1385
> Upstream-patch: https://review.gluster.org/c/glusterfs/+/24930
BUG: 1785714
Change-Id: I3609dd2e1f63c4bd6a19d528b935bf5b05443824
Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/210731
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/cluster/ec/src/ec-heal.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/xlators/cluster/ec/src/ec-heal.c b/xlators/cluster/ec/src/ec-heal.c
index 06bafa5..f6376cd 100644
--- a/xlators/cluster/ec/src/ec-heal.c
+++ b/xlators/cluster/ec/src/ec-heal.c
@@ -2624,6 +2624,7 @@ ec_heal_do(xlator_t *this, void *data, loc_t *loc, int32_t partial)
* removed. We need not take lock on this entry to do so as
* xattrop on a brick is atomic. */
ec_heal_purge_stale_index(frame, ec, loc->inode);
+ goto out;
} else if (need_heal == EC_HEAL_NONEED) {
gf_msg(ec->xl->name, GF_LOG_DEBUG, 0, EC_MSG_HEAL_FAIL,
"Heal is not required for : %s ", uuid_utoa(loc->gfid));
--
1.8.3.1

63
SOURCES/0470-extras-snap_scheduler-changes-in-gluster-shared-stor.patch Normal file
View File

@ -0,0 +1,63 @@
From 8e427716f4e2855093b1a1a0e3a9ec79ebac7faf Mon Sep 17 00:00:00 2001
From: Shwetha K Acharya <sacharya@redhat.com>
Date: Thu, 10 Sep 2020 13:49:09 +0530
Subject: [PATCH 470/473] extras/snap_scheduler: changes in
gluster-shared-storage mount path
The patch https://review.gluster.org/#/c/glusterfs/+/24934/, changes mount point
of gluster_shared_storage from /var/run to /run to address the issue of symlink
at mount path in fstab.
NOTE: mount point /var/run is symlink to /run
The required changes with respect to gluster_shared_storage mount path are
introduced with this patch in snap_scheduler.
>Fixes: #1476
>Change-Id: I9ce88c2f624c6aa5982de04edfee2d0a9f160d62
>Signed-off-by: Shwetha K Acharya <sacharya@redhat.com>
backport of https://review.gluster.org/#/c/glusterfs/+/24971/
BUG: 1873469
Change-Id: I9ce88c2f624c6aa5982de04edfee2d0a9f160d62
Signed-off-by: Shwetha K Acharya <sacharya@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/211391
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
extras/snap_scheduler/gcron.py | 4 ++--
extras/snap_scheduler/snap_scheduler.py | 2 +-
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/extras/snap_scheduler/gcron.py b/extras/snap_scheduler/gcron.py
index cc16310..0e4df77 100755
--- a/extras/snap_scheduler/gcron.py
+++ b/extras/snap_scheduler/gcron.py
@@ -19,10 +19,10 @@ import logging.handlers
import fcntl
-GCRON_TASKS = "/var/run/gluster/shared_storage/snaps/glusterfs_snap_cron_tasks"
+GCRON_TASKS = "/run/gluster/shared_storage/snaps/glusterfs_snap_cron_tasks"
GCRON_CROND_TASK = "/etc/cron.d/glusterfs_snap_cron_tasks"
GCRON_RELOAD_FLAG = "/var/run/gluster/crond_task_reload_flag"
-LOCK_FILE_DIR = "/var/run/gluster/shared_storage/snaps/lock_files/"
+LOCK_FILE_DIR = "/run/gluster/shared_storage/snaps/lock_files/"
log = logging.getLogger("gcron-logger")
start_time = 0.0
diff --git a/extras/snap_scheduler/snap_scheduler.py b/extras/snap_scheduler/snap_scheduler.py
index 5a29d41..e8fcc44 100755
--- a/extras/snap_scheduler/snap_scheduler.py
+++ b/extras/snap_scheduler/snap_scheduler.py
@@ -67,7 +67,7 @@ except ImportError:
SCRIPT_NAME = "snap_scheduler"
scheduler_enabled = False
log = logging.getLogger(SCRIPT_NAME)
-SHARED_STORAGE_DIR="/var/run/gluster/shared_storage"
+SHARED_STORAGE_DIR="/run/gluster/shared_storage"
GCRON_DISABLED = SHARED_STORAGE_DIR+"/snaps/gcron_disabled"
GCRON_ENABLED = SHARED_STORAGE_DIR+"/snaps/gcron_enabled"
GCRON_TASKS = SHARED_STORAGE_DIR+"/snaps/glusterfs_snap_cron_tasks"
--
1.8.3.1

73
SOURCES/0471-nfs-ganesha-gluster_shared_storage-fails-to-automoun.patch Normal file
View File

@ -0,0 +1,73 @@
From d23ad767281af85cf07f5c3f63de482d40ee1953 Mon Sep 17 00:00:00 2001
From: Shwetha K Acharya <sacharya@redhat.com>
Date: Thu, 10 Sep 2020 13:16:12 +0530
Subject: [PATCH 471/473] nfs-ganesha: gluster_shared_storage fails to
automount on node reboot on rhel 8
The patch https://review.gluster.org/#/c/glusterfs/+/24934/, changes mount point
of gluster_shared_storage from /var/run to /run to address the issue of symlink
at mount path in fstab.
NOTE: mount point /var/run is symlink to /run
The required changes with respect to gluster_shared_storage mount path are
introduced with this patch in nfs-ganesha.
>Fixes: #1475
>Change-Id: I9c7677a053e1291f71476d47ba6fa2e729f59625
>Signed-off-by: Shwetha K Acharya <sacharya@redhat.com>
backport of https://review.gluster.org/#/c/glusterfs/+/24970/
BUG: 1873469
Change-Id: I9c7677a053e1291f71476d47ba6fa2e729f59625
Signed-off-by: Shwetha K Acharya <sacharya@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/211392
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
extras/ganesha/ocf/ganesha_nfsd | 2 +-
extras/ganesha/scripts/ganesha-ha.sh | 2 +-
extras/hook-scripts/start/post/S31ganesha-start.sh | 2 +-
3 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/extras/ganesha/ocf/ganesha_nfsd b/extras/ganesha/ocf/ganesha_nfsd
index 93fc8be..f91e8b6 100644
--- a/extras/ganesha/ocf/ganesha_nfsd
+++ b/extras/ganesha/ocf/ganesha_nfsd
@@ -36,7 +36,7 @@ else
. ${OCF_FUNCTIONS_DIR}/ocf-shellfuncs
fi
-OCF_RESKEY_ha_vol_mnt_default="/var/run/gluster/shared_storage"
+OCF_RESKEY_ha_vol_mnt_default="/run/gluster/shared_storage"
: ${OCF_RESKEY_ha_vol_mnt=${OCF_RESKEY_ha_vol_mnt_default}}
ganesha_meta_data() {
diff --git a/extras/ganesha/scripts/ganesha-ha.sh b/extras/ganesha/scripts/ganesha-ha.sh
index a6814b1..9790a71 100644
--- a/extras/ganesha/scripts/ganesha-ha.sh
+++ b/extras/ganesha/scripts/ganesha-ha.sh
@@ -24,7 +24,7 @@ GANESHA_HA_SH=$(realpath $0)
HA_NUM_SERVERS=0
HA_SERVERS=""
HA_VOL_NAME="gluster_shared_storage"
-HA_VOL_MNT="/var/run/gluster/shared_storage"
+HA_VOL_MNT="/run/gluster/shared_storage"
HA_CONFDIR=$HA_VOL_MNT"/nfs-ganesha"
SERVICE_MAN="DISTRO_NOT_FOUND"
diff --git a/extras/hook-scripts/start/post/S31ganesha-start.sh b/extras/hook-scripts/start/post/S31ganesha-start.sh
index 90ba6bc..7ad6f23 100755
--- a/extras/hook-scripts/start/post/S31ganesha-start.sh
+++ b/extras/hook-scripts/start/post/S31ganesha-start.sh
@@ -4,7 +4,7 @@ OPTSPEC="volname:,gd-workdir:"
VOL=
declare -i EXPORT_ID
ganesha_key="ganesha.enable"
-GANESHA_DIR="/var/run/gluster/shared_storage/nfs-ganesha"
+GANESHA_DIR="/run/gluster/shared_storage/nfs-ganesha"
CONF1="$GANESHA_DIR/ganesha.conf"
GLUSTERD_WORKDIR=
--
1.8.3.1

98
SOURCES/0472-geo-rep-gluster_shared_storage-fails-to-automount-on.patch Normal file
View File

@ -0,0 +1,98 @@
From ccd45222c46b91b4d0cd57db9ea8b1515c97ada0 Mon Sep 17 00:00:00 2001
From: Shwetha K Acharya <sacharya@redhat.com>
Date: Mon, 31 Aug 2020 20:08:39 +0530
Subject: [PATCH 472/473] geo-rep: gluster_shared_storage fails to automount on
node reboot on rhel 8.
Issue: On reboot, all the mounts get wiped out.
Only the mounts mentioned in /etc/fstab automatically gets mounted
during boot/reboot.
But /etc/fstab complains on not getting a canonical path
(it gets path containing a symlink)
This is because the gluster_shared_storage, is mounted to
/var/run which is symlink to /run. This is a general practice
followed by most operating systems.
[root@ ~]# ls -lsah /var/run
0 lrwxrwxrwx. 1 root root 6 Jul 22 19:39 /var/run -> ../run
Fix: Mount gluster_shared_storage on /run.
(Also It is seen that /var/run is mostly
used by old or legacy systems, thus it is a good practice to
update /var/run to /run)
>fixes: #1459
>Change-Id: I8c16653be8cd746c84f01abf0eea19284fb97c77
>Signed-off-by: Shwetha K Acharya <sacharya@redhat.com>
backport of https://review.gluster.org/#/c/glusterfs/+/24934/
BUG: 1873469
Change-Id: I8c16653be8cd746c84f01abf0eea19284fb97c77
Signed-off-by: Shwetha K Acharya <sacharya@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/211387
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
.../set/post/S32gluster_enable_shared_storage.sh | 18 +++++++++---------
geo-replication/gsyncd.conf.in | 2 +-
2 files changed, 10 insertions(+), 10 deletions(-)
diff --git a/extras/hook-scripts/set/post/S32gluster_enable_shared_storage.sh b/extras/hook-scripts/set/post/S32gluster_enable_shared_storage.sh
index 885ed03..3bae37c 100755
--- a/extras/hook-scripts/set/post/S32gluster_enable_shared_storage.sh
+++ b/extras/hook-scripts/set/post/S32gluster_enable_shared_storage.sh
@@ -79,9 +79,9 @@ done
if [ "$option" == "disable" ]; then
# Unmount the volume on all the nodes
- umount /var/run/gluster/shared_storage
- cat /etc/fstab | grep -v "gluster_shared_storage /var/run/gluster/shared_storage/" > /var/run/gluster/fstab.tmp
- mv /var/run/gluster/fstab.tmp /etc/fstab
+ umount /run/gluster/shared_storage
+ cat /etc/fstab | grep -v "gluster_shared_storage /run/gluster/shared_storage/" > /run/gluster/fstab.tmp
+ mv /run/gluster/fstab.tmp /etc/fstab
fi
if [ "$is_originator" == 1 ]; then
@@ -105,7 +105,7 @@ function check_volume_status()
}
mount_cmd="mount -t glusterfs $local_node_hostname:/gluster_shared_storage \
- /var/run/gluster/shared_storage"
+ /run/gluster/shared_storage"
if [ "$option" == "enable" ]; then
retry=0;
@@ -120,10 +120,10 @@ if [ "$option" == "enable" ]; then
status=$(check_volume_status)
done
# Mount the volume on all the nodes
- umount /var/run/gluster/shared_storage
- mkdir -p /var/run/gluster/shared_storage
+ umount /run/gluster/shared_storage
+ mkdir -p /run/gluster/shared_storage
$mount_cmd
- cp /etc/fstab /var/run/gluster/fstab.tmp
- echo "$local_node_hostname:/gluster_shared_storage /var/run/gluster/shared_storage/ glusterfs defaults 0 0" >> /var/run/gluster/fstab.tmp
- mv /var/run/gluster/fstab.tmp /etc/fstab
+ cp /etc/fstab /run/gluster/fstab.tmp
+ echo "$local_node_hostname:/gluster_shared_storage /run/gluster/shared_storage/ glusterfs defaults 0 0" >> /run/gluster/fstab.tmp
+ mv /run/gluster/fstab.tmp /etc/fstab
fi
diff --git a/geo-replication/gsyncd.conf.in b/geo-replication/gsyncd.conf.in
index 11e57fd..9688c79 100644
--- a/geo-replication/gsyncd.conf.in
+++ b/geo-replication/gsyncd.conf.in
@@ -123,7 +123,7 @@ type=bool
help=Use this to set Active Passive mode to meta-volume.
[meta-volume-mnt]
-value=/var/run/gluster/shared_storage
+value=/run/gluster/shared_storage
help=Meta Volume or Shared Volume mount path
[allow-network]
--
1.8.3.1

75
SOURCES/0473-glusterd-Fix-Add-brick-with-increasing-replica-count.patch Normal file
View File

@ -0,0 +1,75 @@
From 80f1b3aedcde02ae25b341519857ba9a5b2fa722 Mon Sep 17 00:00:00 2001
From: Sheetal Pamecha <spamecha@redhat.com>
Date: Thu, 24 Sep 2020 19:43:29 +0530
Subject: [PATCH 473/473] glusterd: Fix Add-brick with increasing replica count
failure
Problem: add-brick operation fails with multiple bricks on same
server error when replica count is increased.
This was happening because of extra runs in a loop to compare
hostnames and if bricks supplied were less than "replica" count,
the bricks will get compared to itself resulting in above error.
>Upstream-patch: https://review.gluster.org/#/c/glusterfs/+/25029
>Fixes: #1508
BUG: 1881823
Change-Id: I8668e964340b7bf59728bb838525d2db062197ed
Signed-off-by: Sheetal Pamecha <spamecha@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/213064
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/bugs/glusterd/brick-order-check-add-brick.t | 21 +++++++++++++++++++++
xlators/mgmt/glusterd/src/glusterd-utils.c | 4 ++++
2 files changed, 25 insertions(+)
diff --git a/tests/bugs/glusterd/brick-order-check-add-brick.t b/tests/bugs/glusterd/brick-order-check-add-brick.t
index 29f0ed1..0be31da 100644
--- a/tests/bugs/glusterd/brick-order-check-add-brick.t
+++ b/tests/bugs/glusterd/brick-order-check-add-brick.t
@@ -37,4 +37,25 @@ EXPECT '3 x 3 = 9' volinfo_field $V0 'Number of Bricks'
TEST $CLI_1 volume add-brick $V0 $H1:$L1/${V0}_3 $H1:$L1/${V0}_4 $H1:$L1/${V0}_5 force
EXPECT '4 x 3 = 12' volinfo_field $V0 'Number of Bricks'
+TEST $CLI_1 volume stop $V0
+TEST $CLI_1 volume delete $V0
+
+TEST $CLI_1 volume create $V0 replica 2 $H1:$L1/${V0}1 $H2:$L2/${V0}1
+EXPECT '1 x 2 = 2' volinfo_field $V0 'Number of Bricks'
+EXPECT 'Created' volinfo_field $V0 'Status'
+
+TEST $CLI_1 volume start $V0
+EXPECT 'Started' volinfo_field $V0 'Status'
+
+#Add-brick with Increasing replica count
+TEST $CLI_1 volume add-brick $V0 replica 3 $H3:$L3/${V0}1
+EXPECT '1 x 3 = 3' volinfo_field $V0 'Number of Bricks'
+
+#Add-brick with Increasing replica count from same host should fail
+TEST ! $CLI_1 volume add-brick $V0 replica 5 $H1:$L1/${V0}2 $H1:$L1/${V0}3
+
+#adding multiple bricks from same host should fail the brick order check
+TEST ! $CLI_1 volume add-brick $V0 replica 3 $H1:$L1/${V0}{4..6} $H2:$L2/${V0}{7..9}
+EXPECT '1 x 3 = 3' volinfo_field $V0 'Number of Bricks'
+
cleanup
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index 545e688..d25fc8a 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -14908,6 +14908,10 @@ glusterd_check_brick_order(dict_t *dict, char *err_str, int32_t type,
i = 0;
ai_list_tmp1 = cds_list_entry(ai_list->list.next, addrinfo_list_t, list);
+ if (brick_count < sub_count) {
+ sub_count = brick_count;
+ }
+
/* Check for bad brick order */
while (i < brick_count) {
++i;
--
1.8.3.1

49
SOURCES/0474-features-locks-posixlk-clear-lock-should-set-error-a.patch Normal file
View File

@ -0,0 +1,49 @@
From 3612b3a46c33d19bb7d4aee6eb6625d8d903d459 Mon Sep 17 00:00:00 2001
From: Pranith Kumar K <pkarampu@redhat.com>
Date: Wed, 17 Jun 2020 10:44:37 +0530
Subject: [PATCH 474/478] features/locks: posixlk-clear-lock should set error
as EINTR
Problem:
fuse on receiving interrupt for setlk sends clear-lock "fop"
using virtual-getxattr. At the moment blocked locks which are
cleared return EAGAIN errno as opposed to EINTR errno
Fix:
Return EINTR errno.
Upstream:
> Reviewed-on: https://review.gluster.org/24587
> Updates: #1310
> Change-Id: I47de0fcaec370b267f2f5f89deeb37e1b9c0ee9b
> Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
BUG: 1821743
Change-Id: Id8301ce6e21c009949e88db5904d8b6ecc278f66
Signed-off-by: Csaba Henk <csaba@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/216157
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/features/locks/src/clear.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/xlators/features/locks/src/clear.c b/xlators/features/locks/src/clear.c
index 116aed6..ab1eac6 100644
--- a/xlators/features/locks/src/clear.c
+++ b/xlators/features/locks/src/clear.c
@@ -181,9 +181,9 @@ clrlk_clear_posixlk(xlator_t *this, pl_inode_t *pl_inode, clrlk_args *args,
if (plock->blocked) {
bcount++;
pl_trace_out(this, plock->frame, NULL, NULL, F_SETLKW,
- &plock->user_flock, -1, EAGAIN, NULL);
+ &plock->user_flock, -1, EINTR, NULL);
- STACK_UNWIND_STRICT(lk, plock->frame, -1, EAGAIN,
+ STACK_UNWIND_STRICT(lk, plock->frame, -1, EINTR,
&plock->user_flock, NULL);
} else {
--
1.8.3.1

46
SOURCES/0475-fuse-lock-interrupt-fix-flock_interrupt.t.patch Normal file
View File

@ -0,0 +1,46 @@
From 47d8c316f622850d060af90d1d939528ace5607a Mon Sep 17 00:00:00 2001
From: Csaba Henk <csaba@redhat.com>
Date: Thu, 14 Feb 2019 02:01:38 +0100
Subject: [PATCH 475/478] fuse lock interrupt: fix flock_interrupt.t
Upstream:
> Reviewed-on: https://review.gluster.org/22213
> updates: bz#1193929
> Change-Id: I347de62755100cd69e3cf341434767ae23fd1ba4
> Signed-off-by: Csaba Henk <csaba@redhat.com>
BUG: 1821743
Change-Id: I0088f804bca215152e7ca2c490402c11f7b5333a
Signed-off-by: Csaba Henk <csaba@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/216158
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/features/flock_interrupt.t | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/tests/features/flock_interrupt.t b/tests/features/flock_interrupt.t
index 8603b65..964a4bc 100644
--- a/tests/features/flock_interrupt.t
+++ b/tests/features/flock_interrupt.t
@@ -22,12 +22,12 @@ EXPECT 'Started' volinfo_field $V0 'Status';
TEST $GFS --volfile-id=$V0 --volfile-server=$H0 $M0;
TEST touch $M0/testfile;
-function flock_interrupt {
- flock $MO/testfile sleep 3 & flock -w 1 $M0/testfile true;
- echo ok;
-}
+echo > got_lock
+flock $M0/testfile sleep 6 & { sleep 0.3; flock -w 2 $M0/testfile true; echo ok > got_lock; } &
-EXPECT_WITHIN 2 ok flock_interrupt;
+EXPECT_WITHIN 4 ok cat got_lock;
## Finish up
+sleep 7;
+rm -f got_lock;
cleanup;
--
1.8.3.1

114
SOURCES/0476-mount-fuse-use-cookies-to-get-fuse-interrupt-record-.patch Normal file
View File

@ -0,0 +1,114 @@
From 40519185067d891f06818c574301ea1af4b36479 Mon Sep 17 00:00:00 2001
From: Pranith Kumar K <pkarampu@redhat.com>
Date: Wed, 17 Jun 2020 10:45:19 +0530
Subject: [PATCH 476/478] mount/fuse: use cookies to get fuse-interrupt-record
instead of xdata
Problem:
On executing tests/features/flock_interrupt.t the following error log
appears
[2020-06-16 11:51:54.631072 +0000] E
[fuse-bridge.c:4791:fuse_setlk_interrupt_handler_cbk] 0-glusterfs-fuse:
interrupt record not found
This happens because fuse-interrupt-record is never sent on the wire by
getxattr fop and there is no guarantee that in the cbk it will be
available in case of failures.
Fix:
wind getxattr fop with fuse-interrupt-record as cookie and recover it
in the cbk
Upstream:
> Reviewed-on: https://review.gluster.org/24588
> Fixes: #1310
> Change-Id: I4cfff154321a449114fc26e9440db0f08e5c7daa
> Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
BUG: 1821743
Change-Id: If9576801654d4d743bd66ae90ca259c4d34746a7
Signed-off-by: Csaba Henk <csaba@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/216159
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/features/flock_interrupt.t | 1 -
xlators/mount/fuse/src/fuse-bridge.c | 28 +++++++---------------------
2 files changed, 7 insertions(+), 22 deletions(-)
diff --git a/tests/features/flock_interrupt.t b/tests/features/flock_interrupt.t
index 964a4bc..b8717e3 100644
--- a/tests/features/flock_interrupt.t
+++ b/tests/features/flock_interrupt.t
@@ -28,6 +28,5 @@ flock $M0/testfile sleep 6 & { sleep 0.3; flock -w 2 $M0/testfile true; echo ok
EXPECT_WITHIN 4 ok cat got_lock;
## Finish up
-sleep 7;
rm -f got_lock;
cleanup;
diff --git a/xlators/mount/fuse/src/fuse-bridge.c b/xlators/mount/fuse/src/fuse-bridge.c
index f61fa39..1bddac2 100644
--- a/xlators/mount/fuse/src/fuse-bridge.c
+++ b/xlators/mount/fuse/src/fuse-bridge.c
@@ -4768,16 +4768,8 @@ fuse_setlk_interrupt_handler_cbk(call_frame_t *frame, void *cookie,
int32_t op_errno, dict_t *dict, dict_t *xdata)
{
fuse_interrupt_state_t intstat = INTERRUPT_NONE;
- fuse_interrupt_record_t *fir;
+ fuse_interrupt_record_t *fir = cookie;
fuse_state_t *state = NULL;
- int ret = 0;
-
- ret = dict_get_bin(xdata, "fuse-interrupt-record", (void **)&fir);
- if (ret < 0) {
- gf_log("glusterfs-fuse", GF_LOG_ERROR, "interrupt record not found");
-
- goto out;
- }
intstat = op_ret >= 0 ? INTERRUPT_HANDLED : INTERRUPT_SQUELCHED;
@@ -4789,7 +4781,6 @@ fuse_setlk_interrupt_handler_cbk(call_frame_t *frame, void *cookie,
GF_FREE(state);
}
-out:
STACK_DESTROY(frame->root);
return 0;
@@ -4827,9 +4818,10 @@ fuse_setlk_interrupt_handler(xlator_t *this, fuse_interrupt_record_t *fir)
frame->op = GF_FOP_GETXATTR;
state->name = xattr_name;
- STACK_WIND(frame, fuse_setlk_interrupt_handler_cbk, state->active_subvol,
- state->active_subvol->fops->fgetxattr, state->fd, xattr_name,
- state->xdata);
+ STACK_WIND_COOKIE(frame, fuse_setlk_interrupt_handler_cbk, fir,
+ state->active_subvol,
+ state->active_subvol->fops->fgetxattr, state->fd,
+ xattr_name, state->xdata);
return;
@@ -4852,15 +4844,9 @@ fuse_setlk_resume(fuse_state_t *state)
fir = fuse_interrupt_record_new(state->finh, fuse_setlk_interrupt_handler);
state_clone = gf_memdup(state, sizeof(*state));
if (state_clone) {
- /*
- * Calling this allocator with fir casted to (char *) seems like
- * an abuse of this API, but in fact the API is stupid to assume
- * a (char *) argument (in the funcion it's casted to (void *)
- * anyway).
- */
- state_clone->xdata = dict_for_key_value(
- "fuse-interrupt-record", (char *)fir, sizeof(*fir), _gf_true);
+ state_clone->xdata = dict_new();
}
+
if (!fir || !state_clone || !state_clone->xdata) {
if (fir) {
GF_FREE(fir);
--
1.8.3.1

51
SOURCES/0477-glusterd-snapshot-Snapshot-prevalidation-failure-not.patch Normal file
View File

@ -0,0 +1,51 @@
From 3d50207b346cb5d95af94aa010ebd1ec3e795554 Mon Sep 17 00:00:00 2001
From: srijan-sivakumar <ssivakum@redhat.com>
Date: Wed, 4 Nov 2020 11:44:51 +0530
Subject: [PATCH 477/478] glusterd/snapshot: Snapshot prevalidation failure not
failing.
The value of `ret` is to be set to `-1` to indicate failure
or else the prevalidation which is supposed to be a failure
as the snapshot isn't even activated for cloning will move
to next stage.
Label: DOWNSTREAM ONLY
BUG: 1837926
Change-Id: I95122c3a261332630efa00033a1892a8f95fc00b
Signed-off-by: srijan-sivakumar <ssivakum@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/216920
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Shwetha Acharya <sacharya@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/mgmt/glusterd/src/glusterd-snapshot.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/xlators/mgmt/glusterd/src/glusterd-snapshot.c b/xlators/mgmt/glusterd/src/glusterd-snapshot.c
index 5b8ae97..ee3cea0 100644
--- a/xlators/mgmt/glusterd/src/glusterd-snapshot.c
+++ b/xlators/mgmt/glusterd/src/glusterd-snapshot.c
@@ -2298,8 +2298,8 @@ glusterd_snapshot_clone_prevalidate(dict_t *dict, char **op_errstr,
goto out;
}
-
if (!glusterd_is_volume_started(snap_vol)) {
+ ret = -1;
snprintf(err_str, sizeof(err_str),
"Snapshot %s is "
"not activated",
@@ -9361,7 +9361,8 @@ glusterd_handle_snapshot_fn(rpcsvc_request_t *req)
"for a snapshot");
op_errno = EG_OPNOTSUP;
gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_UNSUPPORTED_VERSION,
- "%s (%d < %d)", err_str, conf->op_version, GD_OP_VERSION_RHS_3_0);
+ "%s (%d < %d)", err_str, conf->op_version,
+ GD_OP_VERSION_RHS_3_0);
ret = -1;
goto out;
}
--
1.8.3.1

119
SOURCES/0478-DHT-Fixing-rebalance-failure-on-issuing-stop-command.patch Normal file
View File

@ -0,0 +1,119 @@
From e772bef5631017145cd0270d72a9ada1378e022a Mon Sep 17 00:00:00 2001
From: Barak Sason Rofman <bsasonro@redhat.com>
Date: Fri, 30 Oct 2020 08:27:47 +0200
Subject: [PATCH 478/478] DHT - Fixing rebalance failure on issuing stop
command
Issuing a stop command for an ongoing rebalance process results in an error.
This issue was brought up in https://bugzilla.redhat.com/1286171 and a patch
(https://review.gluster.org/24103/) was submitted to resolve the issue.
However the submitted patch resolved only part of the
problem by reducing the number of log messages that were printed (since
rebalnace is currently a recursive process, an error message was printed
for every directory) but didn't fully resolve the root cause for the
failure.
This patch fixes the issue by modifying the code-path which handles the
termination of the rebalance process by issuing a stop command.
Upstream:
> Reviewed-on: https://github.com/gluster/glusterfs/pull/1628
> fixes: #1627
> Change-Id: I604f2b0f8b1ccb1026b8425a14200bbd1dc5bd03
> Signed-off-by: Barak Sason Rofman bsasonro@redhat.com
BUG: 1286171
Change-Id: I604f2b0f8b1ccb1026b8425a14200bbd1dc5bd03
Signed-off-by: Barak Sason Rofman <bsasonro@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/216896
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/cluster/dht/src/dht-rebalance.c | 22 ++++++++++++----------
1 file changed, 12 insertions(+), 10 deletions(-)
diff --git a/xlators/cluster/dht/src/dht-rebalance.c b/xlators/cluster/dht/src/dht-rebalance.c
index abc10fc..d49a719 100644
--- a/xlators/cluster/dht/src/dht-rebalance.c
+++ b/xlators/cluster/dht/src/dht-rebalance.c
@@ -3113,12 +3113,10 @@ int static gf_defrag_get_entry(xlator_t *this, int i,
struct dht_container *tmp_container = NULL;
if (defrag->defrag_status != GF_DEFRAG_STATUS_STARTED) {
- ret = -1;
goto out;
}
if (dir_dfmeta->offset_var[i].readdir_done == 1) {
- ret = 0;
goto out;
}
@@ -3135,7 +3133,6 @@ int static gf_defrag_get_entry(xlator_t *this, int i,
&(dir_dfmeta->equeue[i]), xattr_req, NULL);
if (ret == 0) {
dir_dfmeta->offset_var[i].readdir_done = 1;
- ret = 0;
goto out;
}
@@ -3161,7 +3158,6 @@ int static gf_defrag_get_entry(xlator_t *this, int i,
while (1) {
if (defrag->defrag_status != GF_DEFRAG_STATUS_STARTED) {
- ret = -1;
goto out;
}
@@ -3273,12 +3269,14 @@ int static gf_defrag_get_entry(xlator_t *this, int i,
}
out:
- if (ret == 0) {
- *container = tmp_container;
- } else {
- if (tmp_container) {
+ if (defrag->defrag_status == GF_DEFRAG_STATUS_STARTED) {
+ if (ret == 0) {
+ *container = tmp_container;
+ } else {
gf_defrag_free_container(tmp_container);
}
+ } else {
+ gf_defrag_free_container(tmp_container);
}
return ret;
@@ -3487,7 +3485,7 @@ gf_defrag_process_dir(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
migrate_data, dir_dfmeta, xattr_req,
&should_commit_hash, perrno);
- if (defrag->defrag_status == GF_DEFRAG_STATUS_STOPPED) {
+ if (defrag->defrag_status != GF_DEFRAG_STATUS_STARTED) {
goto out;
}
@@ -3947,7 +3945,7 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
ret = gf_defrag_fix_layout(this, defrag, &entry_loc, fix_layout,
migrate_data);
- if (defrag->defrag_status == GF_DEFRAG_STATUS_STOPPED) {
+ if (defrag->defrag_status != GF_DEFRAG_STATUS_STARTED) {
goto out;
}
@@ -4015,6 +4013,10 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
(defrag->cmd != GF_DEFRAG_CMD_START_LAYOUT_FIX)) {
ret = gf_defrag_process_dir(this, defrag, loc, migrate_data, &perrno);
+ if (defrag->defrag_status != GF_DEFRAG_STATUS_STARTED) {
+ goto out;
+ }
+
if (ret && (ret != 2)) {
if (perrno == ENOENT || perrno == ESTALE) {
ret = 0;
--
1.8.3.1

53
SOURCES/0479-ganesha-ha-revised-regex-exprs-for-status.patch Normal file
View File

@ -0,0 +1,53 @@
From 9036c9f0fd081c83c5c4fcd1ecba858421442777 Mon Sep 17 00:00:00 2001
From: "Kaleb S. KEITHLEY" <kkeithle@redhat.com>
Date: Tue, 10 Nov 2020 07:39:14 -0500
Subject: [PATCH 479/479] ganesha-ha: revised regex exprs for --status
better whitespace in regex
This has worked for years, but somehow no longer works on rhel8
> Updates: #1000
> Change-Id: I2c1a3537573d125608334772ba1a263c55407dd4
> Signed-off-by: Kaleb S. KEITHLEY <kkeithle@redhat.com>
> https://github.com/gluster/glusterfs/commit/4026fe9a956238d8e4785cf39c3b7290eae90f03
BUG: 1895301
Change-Id: I2c1a3537573d125608334772ba1a263c55407dd4
Signed-off-by: Kaleb S. KEITHLEY <kkeithle@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/217480
Tested-by: RHGS Build Bot <nigelb@redhat.com>
---
extras/ganesha/scripts/ganesha-ha.sh | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/extras/ganesha/scripts/ganesha-ha.sh b/extras/ganesha/scripts/ganesha-ha.sh
index 9790a71..491c61d 100644
--- a/extras/ganesha/scripts/ganesha-ha.sh
+++ b/extras/ganesha/scripts/ganesha-ha.sh
@@ -948,18 +948,18 @@ status()
# check if the VIP and port block/unblock RAs are on the expected nodes
for n in ${nodes[*]}; do
- grep -E -x "${n}-nfs_block \(ocf::heartbeat:portblock\): Started ${n}" > /dev/null 2>&1 ${scratch}
+ grep -E -x "${n}-nfs_block +\(ocf::heartbeat:portblock\): +Started ${n}" > /dev/null 2>&1 ${scratch}
result=$?
((healthy+=${result}))
- grep -E -x "${n}-cluster_ip-1 \(ocf::heartbeat:IPaddr\): Started ${n}" > /dev/null 2>&1 ${scratch}
+ grep -E -x "${n}-cluster_ip-1 +\(ocf::heartbeat:IPaddr\): +Started ${n}" > /dev/null 2>&1 ${scratch}
result=$?
((healthy+=${result}))
- grep -E -x "${n}-nfs_unblock \(ocf::heartbeat:portblock\): Started ${n}" > /dev/null 2>&1 ${scratch}
+ grep -E -x "${n}-nfs_unblock +\(ocf::heartbeat:portblock\): +Started ${n}" > /dev/null 2>&1 ${scratch}
result=$?
((healthy+=${result}))
done
- grep -E "\):\ Stopped|FAILED" > /dev/null 2>&1 ${scratch}
+ grep -E "\): +Stopped|FAILED" > /dev/null 2>&1 ${scratch}
result=$?
if [ ${result} -eq 0 ]; then
--
1.8.3.1

255
SOURCES/0480-DHT-Rebalance-Ensure-Rebalance-reports-status-only-o.patch Normal file
View File

@ -0,0 +1,255 @@
From 759c12fc016a6399bb179aa0f930602c87d1e0f8 Mon Sep 17 00:00:00 2001
From: Barak Sason Rofman <bsasonro@redhat.com>
Date: Tue, 24 Nov 2020 12:56:10 +0200
Subject: [PATCH 480/480] DHT/Rebalance - Ensure Rebalance reports status only
once upon stopping
Upon issuing rebalance stop command, the status of rebalance is being
logged twice to the log file, which can sometime result in an
inconsistent reports (one report states status stopped, while the other
may report something else).
This fix ensures rebalance reports it's status only once and that the
correct status is being reported.
Upstream:
> Reviewed-on: https://github.com/gluster/glusterfs/pull/1783
> fixes: #1782
> Change-Id: Id3206edfad33b3db60e9df8e95a519928dc7cb37
> Signed-off-by: Barak Sason Rofman bsasonro@redhat.com
BUG: 1286171
Change-Id: Id3206edfad33b3db60e9df8e95a519928dc7cb37
Signed-off-by: Barak Sason Rofman <bsasonro@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/218953
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Csaba Henk <chenk@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/bugs/distribute/bug-1286171.t | 75 +++++++++++++++++++++++++++++++++
xlators/cluster/dht/src/dht-common.c | 2 +-
xlators/cluster/dht/src/dht-common.h | 2 +-
xlators/cluster/dht/src/dht-rebalance.c | 63 ++++++++++++++-------------
4 files changed, 108 insertions(+), 34 deletions(-)
create mode 100644 tests/bugs/distribute/bug-1286171.t
diff --git a/tests/bugs/distribute/bug-1286171.t b/tests/bugs/distribute/bug-1286171.t
new file mode 100644
index 0000000..a2ca36f
--- /dev/null
+++ b/tests/bugs/distribute/bug-1286171.t
@@ -0,0 +1,75 @@
+#!/bin/bash
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../cluster.rc
+. $(dirname $0)/../../volume.rc
+
+# Initialize
+#------------------------------------------------------------
+cleanup;
+
+volname=bug-1286171
+
+# Start glusterd
+TEST glusterd;
+TEST pidof glusterd;
+TEST $CLI volume info;
+
+# Create a volume
+TEST $CLI volume create $volname $H0:$B0/${volname}{1,2}
+
+# Verify volume creation
+EXPECT "$volname" volinfo_field $volname 'Volume Name';
+EXPECT 'Created' volinfo_field $volname 'Status';
+
+# Start volume and verify successful start
+TEST $CLI volume start $volname;
+EXPECT 'Started' volinfo_field $volname 'Status';
+TEST glusterfs --volfile-id=$volname --volfile-server=$H0 --entry-timeout=0 $M0;
+#------------------------------------------------------------
+
+# Create a nested dir structure and some file under MP
+cd $M0;
+for i in {1..5}
+do
+ mkdir dir$i
+ cd dir$i
+ for j in {1..5}
+ do
+ mkdir dir$i$j
+ cd dir$i$j
+ for k in {1..5}
+ do
+ mkdir dir$i$j$k
+ cd dir$i$j$k
+ touch {1..300}
+ cd ..
+ done
+ touch {1..300}
+ cd ..
+ done
+ touch {1..300}
+ cd ..
+done
+touch {1..300}
+
+# Add-brick and start rebalance
+TEST $CLI volume add-brick $volname $H0:$B0/${volname}4;
+TEST $CLI volume rebalance $volname start;
+
+# Let rebalance run for a while
+sleep 5
+
+# Stop rebalance
+TEST $CLI volume rebalance $volname stop;
+
+# Allow rebalance to stop
+sleep 5
+
+# Examine the logfile for errors
+cd /var/log/glusterfs;
+failures=`grep "failures:" ${volname}-rebalance.log | tail -1 | sed 's/.*failures: //; s/,.*//'`;
+
+TEST [ $failures == 0 ];
+
+cleanup;
diff --git a/xlators/cluster/dht/src/dht-common.c b/xlators/cluster/dht/src/dht-common.c
index 23cc80c..4db89df 100644
--- a/xlators/cluster/dht/src/dht-common.c
+++ b/xlators/cluster/dht/src/dht-common.c
@@ -10969,7 +10969,7 @@ dht_notify(xlator_t *this, int event, void *data, ...)
if ((cmd == GF_DEFRAG_CMD_STATUS) ||
(cmd == GF_DEFRAG_CMD_STATUS_TIER) ||
(cmd == GF_DEFRAG_CMD_DETACH_STATUS))
- gf_defrag_status_get(conf, output);
+ gf_defrag_status_get(conf, output, _gf_false);
else if (cmd == GF_DEFRAG_CMD_START_DETACH_TIER)
gf_defrag_start_detach_tier(defrag);
else if (cmd == GF_DEFRAG_CMD_DETACH_START)
diff --git a/xlators/cluster/dht/src/dht-common.h b/xlators/cluster/dht/src/dht-common.h
index 9ec5b51..92f1b89 100644
--- a/xlators/cluster/dht/src/dht-common.h
+++ b/xlators/cluster/dht/src/dht-common.h
@@ -1252,7 +1252,7 @@ dht_fxattrop_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
int32_t op_ret, int32_t op_errno, dict_t *dict, dict_t *xdata);
int
-gf_defrag_status_get(dht_conf_t *conf, dict_t *dict);
+gf_defrag_status_get(dht_conf_t *conf, dict_t *dict, gf_boolean_t log_status);
void
gf_defrag_set_pause_state(gf_tier_conf_t *tier_conf, tier_pause_state_t state);
diff --git a/xlators/cluster/dht/src/dht-rebalance.c b/xlators/cluster/dht/src/dht-rebalance.c
index d49a719..16ac16c 100644
--- a/xlators/cluster/dht/src/dht-rebalance.c
+++ b/xlators/cluster/dht/src/dht-rebalance.c
@@ -2720,7 +2720,6 @@ gf_defrag_migrate_single_file(void *opaque)
iatt_ptr = &entry->d_stat;
if (defrag->defrag_status != GF_DEFRAG_STATUS_STARTED) {
- ret = -1;
goto out;
}
@@ -3833,7 +3832,6 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
list_for_each_entry_safe(entry, tmp, &entries.list, list)
{
if (defrag->defrag_status != GF_DEFRAG_STATUS_STARTED) {
- ret = 1;
goto out;
}
@@ -4863,7 +4861,7 @@ out:
LOCK(&defrag->lock);
{
status = dict_new();
- gf_defrag_status_get(conf, status);
+ gf_defrag_status_get(conf, status, _gf_true);
if (ctx && ctx->notify)
ctx->notify(GF_EN_DEFRAG_STATUS, status);
if (status)
@@ -4998,7 +4996,7 @@ out:
}
int
-gf_defrag_status_get(dht_conf_t *conf, dict_t *dict)
+gf_defrag_status_get(dht_conf_t *conf, dict_t *dict, gf_boolean_t log_status)
{
int ret = 0;
uint64_t files = 0;
@@ -5095,34 +5093,35 @@ gf_defrag_status_get(dht_conf_t *conf, dict_t *dict)
gf_log(THIS->name, GF_LOG_WARNING, "failed to set time-left");
log:
- switch (defrag->defrag_status) {
- case GF_DEFRAG_STATUS_NOT_STARTED:
- status = "not started";
- break;
- case GF_DEFRAG_STATUS_STARTED:
- status = "in progress";
- break;
- case GF_DEFRAG_STATUS_STOPPED:
- status = "stopped";
- break;
- case GF_DEFRAG_STATUS_COMPLETE:
- status = "completed";
- break;
- case GF_DEFRAG_STATUS_FAILED:
- status = "failed";
- break;
- default:
- break;
- }
+ if (log_status) {
+ switch (defrag->defrag_status) {
+ case GF_DEFRAG_STATUS_NOT_STARTED:
+ status = "not started";
+ break;
+ case GF_DEFRAG_STATUS_STARTED:
+ status = "in progress";
+ break;
+ case GF_DEFRAG_STATUS_STOPPED:
+ status = "stopped";
+ break;
+ case GF_DEFRAG_STATUS_COMPLETE:
+ status = "completed";
+ break;
+ case GF_DEFRAG_STATUS_FAILED:
+ status = "failed";
+ break;
+ default:
+ break;
+ }
- gf_msg(THIS->name, GF_LOG_INFO, 0, DHT_MSG_REBALANCE_STATUS,
- "Rebalance is %s. Time taken is %.2f secs", status, elapsed);
- gf_msg(THIS->name, GF_LOG_INFO, 0, DHT_MSG_REBALANCE_STATUS,
- "Files migrated: %" PRIu64 ", size: %" PRIu64 ", lookups: %" PRIu64
- ", failures: %" PRIu64
- ", skipped: "
- "%" PRIu64,
- files, size, lookup, failures, skipped);
+ gf_msg("DHT", GF_LOG_INFO, 0, DHT_MSG_REBALANCE_STATUS,
+ "Rebalance is %s. Time taken is %.2f secs "
+ "Files migrated: %" PRIu64 ", size: %" PRIu64
+ ", lookups: %" PRIu64 ", failures: %" PRIu64
+ ", skipped: "
+ "%" PRIu64,
+ status, elapsed, files, size, lookup, failures, skipped);
+ }
out:
return 0;
}
@@ -5299,7 +5298,7 @@ gf_defrag_stop(dht_conf_t *conf, gf_defrag_status_t status, dict_t *output)
defrag->defrag_status = status;
if (output)
- gf_defrag_status_get(conf, output);
+ gf_defrag_status_get(conf, output, _gf_false);
ret = 0;
out:
gf_msg_debug("", 0, "Returning %d", ret);
--
1.8.3.1

26
SOURCES/0481-Update-rfc.sh-to-rhgs-3.5.4.patch Normal file
View File

@ -0,0 +1,26 @@
From 828be8e789db3c77587c708f930d7fe8c9456e3b Mon Sep 17 00:00:00 2001
From: Rinku Kothiya <rkothiya@redhat.com>
Date: Fri, 4 Dec 2020 05:18:45 +0530
Subject: [PATCH 481/511] Update rfc.sh to rhgs-3.5.4
Signed-off-by: Rinku Kothiya <rkothiya@redhat.com>
---
rfc.sh | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/rfc.sh b/rfc.sh
index 1dca29f..c0559b9 100755
--- a/rfc.sh
+++ b/rfc.sh
@@ -18,7 +18,7 @@ done
shift $((OPTIND-1))
-branch="rhgs-3.5.3";
+branch="rhgs-3.5.4";
set_hooks_commit_msg()
{
--
1.8.3.1

49
SOURCES/0482-logger-Always-print-errors-in-english.patch Normal file
View File

@ -0,0 +1,49 @@
From e43af5b15d14e43c3201fd0fb7bf02663e3e0127 Mon Sep 17 00:00:00 2001
From: Rinku Kothiya <rkothiya@redhat.com>
Date: Sat, 7 Nov 2020 12:09:36 +0530
Subject: [PATCH 482/511] logger: Always print errors in english
Upstream:
> Reviewed-on: https://github.com/gluster/glusterfs/pull/1657
> fixes: #1302
> Change-Id: If0e21f016155276a953c64a8dd13ff3eb281d09d
> Signed-off-by: Rinku Kothiya <rkothiya@redhat.com>
BUG: 1896425
Change-Id: If0e21f016155276a953c64a8dd13ff3eb281d09d
Signed-off-by: Rinku Kothiya <rkothiya@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/219999
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
libglusterfs/src/logging.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/libglusterfs/src/logging.c b/libglusterfs/src/logging.c
index 7f0eff6..5874c34 100644
--- a/libglusterfs/src/logging.c
+++ b/libglusterfs/src/logging.c
@@ -513,6 +513,7 @@ gf_openlog(const char *ident, int option, int facility)
{
int _option = option;
int _facility = facility;
+ char *language = NULL;
if (-1 == _option) {
_option = LOG_PID | LOG_NDELAY;
@@ -522,7 +523,10 @@ gf_openlog(const char *ident, int option, int facility)
}
/* TODO: Should check for errors here and return appropriately */
- setlocale(LC_ALL, "");
+ language = setlocale(LC_ALL, "en_US.UTF-8");
+ if (!language)
+ setlocale(LC_ALL, "");
+
setlocale(LC_NUMERIC, "C"); /* C-locale for strtod, ... */
/* close the previous syslog if open as we are changing settings */
closelog();
--
1.8.3.1

Some files were not shown because too many files have changed in this diff Show More