rpms/glusterfs
5
0
Fork 0
Code Issues Pull Requests Releases Activity

autobuild v3.12.2-33

Browse Source 
Resolves: bz#1350745 bz#1362129 bz#1541568 bz#1597252 bz#1599220
Resolves: bz#1633177 bz#1637564 bz#1639476 bz#1639568 bz#1643370
Resolves: bz#1645480 bz#1648296 bz#1648893 bz#1651040 bz#1651460
Resolves: bz#1652466 bz#1652537 bz#1653224 bz#1653613 bz#1654103
Resolves: bz#1654161 bz#1655385 bz#1655578 bz#1656357 bz#1659439
Signed-off-by: Milind Changire <mchangir@redhat.com>
This commit is contained in:
Milind Changire 2018-12-18 11:23:13 -05:00
parent 9f6b7acb4a
commit 1f2f23ddef
39 changed files with 8558 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

27
0456-Update-rfc.sh-to-rhgs-3.4.3.patch Normal file
View File

@ -0,0 +1,27 @@
From cd5b8a65ea1df291d936bda1f2514cc5805d4209 Mon Sep 17 00:00:00 2001
From: Milind Changire <mchangir@redhat.com>
Date: Tue, 11 Dec 2018 12:57:44 +0530
Subject: [PATCH 456/493] Update rfc.sh to rhgs-3.4.3
Change-Id: Ida4e354010cf83aa121baa02a2d071ccb8867fc7
Signed-off-by: Milind Changire <mchangir@redhat.com>
---
rfc.sh | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/rfc.sh b/rfc.sh
index 5b09aad..bd52851 100755
--- a/rfc.sh
+++ b/rfc.sh
@@ -17,7 +17,7 @@ done
shift $((OPTIND-1))
-branch="rhgs-3.4.2";
+branch="rhgs-3.4.3";
set_hooks_commit_msg()
{
--
1.8.3.1

122
0457-cluster-dht-sync-brick-root-perms-on-add-brick.patch Normal file
View File

@ -0,0 +1,122 @@
From aad0d32376e6ca56770e5c2d4dc5a1462b1e7167 Mon Sep 17 00:00:00 2001
From: N Balachandran <nbalacha@redhat.com>
Date: Thu, 13 Dec 2018 12:06:10 +0530
Subject: [PATCH 457/493] cluster/dht: sync brick root perms on add brick
If a single brick is added to the volume and the
newly added brick is the first to respond to a
dht_revalidate call, its stbuf will not be merged
into local->stbuf as the brick does not yet have
a layout. The is_permission_different check therefore
fails to detect that an attr heal is required as it
only considers the stbuf values from existing bricks.
To fix this, merge all stbuf values into local->stbuf
and use local->prebuf to store the correct directory
attributes.
upstream patch: https://review.gluster.org/#/c/glusterfs/+/21611/
> Change-Id: Ic9e8b04a1ab9ed1248b6b056e3450bbafe32e1bc
> fixes: bz#1648298
> Signed-off-by: N Balachandran <nbalacha@redhat.com>
Change-Id: I329ce48555d15f741d7247290e749f0800d12df8
BUG: 1648296
Signed-off-by: N Balachandran <nbalacha@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158493
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Susant Palai <spalai@redhat.com>
Reviewed-by: Raghavendra Gowdappa <rgowdapp@redhat.com>
---
tests/bugs/distribute/bug-1368012.t | 11 +++++------
xlators/cluster/dht/src/dht-common.c | 26 ++++++++------------------
2 files changed, 13 insertions(+), 24 deletions(-)
diff --git a/tests/bugs/distribute/bug-1368012.t b/tests/bugs/distribute/bug-1368012.t
index b861554..0b62635 100644
--- a/tests/bugs/distribute/bug-1368012.t
+++ b/tests/bugs/distribute/bug-1368012.t
@@ -15,7 +15,7 @@ TEST pidof glusterd;
TEST $CLI volume info;
## Lets create volume
-TEST $CLI volume create $V0 $H0:/${V0}{1,2};
+TEST $CLI volume create $V0 $H0:$B0/${V0}{1,2};
## Verify volume is created
EXPECT "$V0" volinfo_field $V0 'Volume Name';
@@ -36,17 +36,16 @@ TEST chmod 444 $M0
TEST permission_root=`stat -c "%A" $M0`
TEST echo $permission_root
#Add-brick
-TEST $CLI volume add-brick $V0 $H0:/${V0}3
+TEST $CLI volume add-brick $V0 $H0:$B0/${V0}3
EXPECT_WITHIN $PROCESS_UP_TIMEOUT "3" online_brick_count
#Allow one lookup to happen
-TEST pushd $M0
-TEST ls
+TEST ls $M0
#Generate another lookup
echo 3 > /proc/sys/vm/drop_caches
-TEST ls
+TEST ls $M0
#check root permission
EXPECT_WITHIN "5" $permission_root get_permission $M0
#check permission on the new-brick
-EXPECT $permission_root get_permission /${V0}3
+EXPECT $permission_root get_permission $B0/${V0}3
cleanup
diff --git a/xlators/cluster/dht/src/dht-common.c b/xlators/cluster/dht/src/dht-common.c
index d3a0c8b..2e19036 100644
--- a/xlators/cluster/dht/src/dht-common.c
+++ b/xlators/cluster/dht/src/dht-common.c
@@ -1717,14 +1717,17 @@ dht_revalidate_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
if (is_dir) {
ret = dht_dir_has_layout (xattr, conf->xattr_name);
if (ret >= 0) {
- if (is_greater_time(local->stbuf.ia_ctime,
- local->stbuf.ia_ctime_nsec,
+ if (is_greater_time(local->prebuf.ia_ctime,
+ local->prebuf.ia_ctime_nsec,
stbuf->ia_ctime,
stbuf->ia_ctime_nsec)) {
/* Choose source */
local->prebuf.ia_gid = stbuf->ia_gid;
local->prebuf.ia_uid = stbuf->ia_uid;
+ local->prebuf.ia_ctime = stbuf->ia_ctime;
+ local->prebuf.ia_ctime_nsec = stbuf->ia_ctime_nsec;
+
if (__is_root_gfid (stbuf->ia_gfid))
local->prebuf.ia_prot = stbuf->ia_prot;
}
@@ -1792,22 +1795,9 @@ dht_revalidate_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
}
}
-
- /* Update stbuf from the servers where layout is present. This
- * is an indication that the server is not a newly added brick.
- * Merging stbuf from newly added brick may result in the added
- * brick being the source of heal for uid/gid */
- if (!is_dir || (is_dir &&
- dht_dir_has_layout (xattr, conf->xattr_name) >= 0)
- || conf->subvolume_cnt == 1) {
-
- dht_iatt_merge (this, &local->stbuf, stbuf, prev);
- dht_iatt_merge (this, &local->postparent, postparent,
- prev);
- } else {
- /* copy the gfid anyway */
- gf_uuid_copy (local->stbuf.ia_gfid, stbuf->ia_gfid);
- }
+ gf_uuid_copy (local->stbuf.ia_gfid, stbuf->ia_gfid);
+ dht_iatt_merge (this, &local->stbuf, stbuf, prev);
+ dht_iatt_merge (this, &local->postparent, postparent, prev);
local->op_ret = 0;
--
1.8.3.1

42
0458-glusterd-fix-crash.patch Normal file
View File

@ -0,0 +1,42 @@
From 3a1484c401d4293531c80532fa96c2f7cfc8aa2d Mon Sep 17 00:00:00 2001
From: Sanju Rakonde <srakonde@redhat.com>
Date: Wed, 19 Sep 2018 19:49:51 +0530
Subject: [PATCH 458/493] glusterd: fix crash
When huge number of volumes are created, glusterd crash is seen.
With the core dump, got to know that mgmt_lock_timer became NULL.
Adding a null check for the same, need to explore about the root
cause.
>updates: bz#1630922
>Change-Id: I0770063fcbbbf4b24bef29e94b857b20bdfb5b85
>Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
upstream patch: https://review.gluster.org/#/c/glusterfs/+/21228/
Change-Id: I0770063fcbbbf4b24bef29e94b857b20bdfb5b85
BUG: 1599220
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158542
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/mgmt/glusterd/src/glusterd-locks.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/xlators/mgmt/glusterd/src/glusterd-locks.c b/xlators/mgmt/glusterd/src/glusterd-locks.c
index d75452d..d62d9dd 100644
--- a/xlators/mgmt/glusterd/src/glusterd-locks.c
+++ b/xlators/mgmt/glusterd/src/glusterd-locks.c
@@ -890,7 +890,7 @@ glusterd_mgmt_v3_unlock (const char *name, uuid_t uuid, char *type)
type, name);
/* Release owner reference which was held during lock */
- if (mgmt_lock_timer->timer) {
+ if (mgmt_lock_timer && mgmt_lock_timer->timer) {
ret = -1;
mgmt_lock_timer_xl = mgmt_lock_timer->xl;
GF_VALIDATE_OR_GOTO (this->name, mgmt_lock_timer_xl, out);
--
1.8.3.1

223
0459-glfsheal-add-a-nolog-flag.patch Normal file
View File

@ -0,0 +1,223 @@
From 1637d5018aeea96efc2916afe162c4905ef2c2d9 Mon Sep 17 00:00:00 2001
From: Ravishankar N <ravishankar@redhat.com>
Date: Fri, 14 Dec 2018 12:48:05 +0530
Subject: [PATCH 459/493] glfsheal: add a '--nolog' flag
(Upstream master patch: https://review.gluster.org/#/c/glusterfs/+/21501/)
....and if set, change the log level to GF_LOG_NONE. This is useful for
monitoring applications which invoke the heal info set of commands once
every minute, leading to un-necessary glfsheal* logs in
/var/log/glusterfs/. For example, we can now run
`gluster volume heal <VOLNAME> info --nolog`
`gluster volume heal <VOLNAME> info split-brain --nolog` etc.
The default log level is still retained at GF_LOG_INFO.
The patch also changes glfsheal internally to accept '--xml' instead of 'xml'.
Note: The --nolog flag is *not* displayed in the help anywhere, for the
sake of consistency in how the other flags are not displayed anywhere in
the help.
Change-Id: I932d0f79070880b0f9ca87e164d3c2a3b831c8c4
BUG: 1579293
Signed-off-by: Ravishankar N <ravishankar@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158640
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
cli/src/cli-cmd-volume.c | 8 ++++---
cli/src/cli.c | 5 ++++
cli/src/cli.h | 1 +
heal/src/glfs-heal.c | 59 ++++++++++++++++++++++++++++++++++++------------
4 files changed, 56 insertions(+), 17 deletions(-)
diff --git a/cli/src/cli-cmd-volume.c b/cli/src/cli-cmd-volume.c
index 32efa73..8fca7eb 100644
--- a/cli/src/cli-cmd-volume.c
+++ b/cli/src/cli-cmd-volume.c
@@ -2830,7 +2830,7 @@ cli_launch_glfs_heal (int heal_op, dict_t *options)
switch (heal_op) {
case GF_SHD_OP_INDEX_SUMMARY:
if (global_state->mode & GLUSTER_MODE_XML) {
- runner_add_args (&runner, "xml", NULL);
+ runner_add_args (&runner, "--xml", NULL);
}
break;
case GF_SHD_OP_SBRAIN_HEAL_FROM_BIGGER_FILE:
@@ -2854,7 +2854,7 @@ cli_launch_glfs_heal (int heal_op, dict_t *options)
case GF_SHD_OP_SPLIT_BRAIN_FILES:
runner_add_args (&runner, "split-brain-info", NULL);
if (global_state->mode & GLUSTER_MODE_XML) {
- runner_add_args (&runner, "xml", NULL);
+ runner_add_args (&runner, "--xml", NULL);
}
break;
case GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE:
@@ -2864,12 +2864,14 @@ cli_launch_glfs_heal (int heal_op, dict_t *options)
case GF_SHD_OP_HEAL_SUMMARY:
runner_add_args (&runner, "info-summary", NULL);
if (global_state->mode & GLUSTER_MODE_XML) {
- runner_add_args (&runner, "xml", NULL);
+ runner_add_args (&runner, "--xml", NULL);
}
break;
default:
ret = -1;
}
+ if (global_state->mode & GLUSTER_MODE_GLFSHEAL_NOLOG)
+ runner_add_args(&runner, "--nolog", NULL);
ret = runner_start (&runner);
if (ret == -1)
goto out;
diff --git a/cli/src/cli.c b/cli/src/cli.c
index b64d4ef..3fd7bc5 100644
--- a/cli/src/cli.c
+++ b/cli/src/cli.c
@@ -340,6 +340,11 @@ cli_opt_parse (char *opt, struct cli_state *state)
return 0;
}
+ if (strcmp(opt, "nolog") == 0) {
+ state->mode |= GLUSTER_MODE_GLFSHEAL_NOLOG;
+ return 0;
+ }
+
if (strcmp (opt, "wignore-partition") == 0) {
state->mode |= GLUSTER_MODE_WIGNORE_PARTITION;
return 0;
diff --git a/cli/src/cli.h b/cli/src/cli.h
index 109dcd4..104d601 100644
--- a/cli/src/cli.h
+++ b/cli/src/cli.h
@@ -61,6 +61,7 @@ typedef enum {
#define GLUSTER_MODE_XML (1 << 2)
#define GLUSTER_MODE_WIGNORE (1 << 3)
#define GLUSTER_MODE_WIGNORE_PARTITION (1 << 4)
+#define GLUSTER_MODE_GLFSHEAL_NOLOG (1 << 5)
#define GLUSTERD_GET_QUOTA_LIST_MOUNT_PATH(abspath, volname, path) do { \
diff --git a/heal/src/glfs-heal.c b/heal/src/glfs-heal.c
index 153cd29..12746dc 100644
--- a/heal/src/glfs-heal.c
+++ b/heal/src/glfs-heal.c
@@ -40,6 +40,9 @@ xmlDocPtr glfsh_doc = NULL;
ret = 0; \
} while (0) \
+#define MODE_XML (1 << 0)
+#define MODE_NO_LOG (1 << 1)
+
typedef struct num_entries {
uint64_t num_entries;
uint64_t pending_entries;
@@ -1434,6 +1437,28 @@ out:
return ret;
}
+static void
+parse_flags(int *argc, char **argv, int *flags)
+{
+ int i = 0;
+ char *opt = NULL;
+ int count = 0;
+
+ for (i = 0; i < *argc; i++) {
+ opt = strtail(argv[i], "--");
+ if (!opt)
+ continue;
+ if (strcmp(opt, "nolog") == 0) {
+ *flags |= MODE_NO_LOG;
+ count++;
+ } else if (strcmp(opt, "xml") == 0) {
+ *flags |= MODE_XML;
+ count++;
+ }
+ }
+ *argc = *argc - count;
+}
+
int
glfsh_heal_from_bigger_file_or_mtime (glfs_t *fs, xlator_t *top_subvol,
loc_t *rootloc, char *file,
@@ -1518,6 +1543,8 @@ main (int argc, char **argv)
char *file = NULL;
char *op_errstr = NULL;
gf_xl_afr_op_t heal_op = -1;
+ gf_loglevel_t log_level = GF_LOG_INFO;
+ int flags = 0;
if (argc < 2) {
printf (USAGE_STR, argv[0]);
@@ -1526,6 +1553,13 @@ main (int argc, char **argv)
}
volname = argv[1];
+
+ parse_flags(&argc, argv, &flags);
+ if (flags & MODE_NO_LOG)
+ log_level = GF_LOG_NONE;
+ if (flags & MODE_XML)
+ is_xml = 1;
+
switch (argc) {
case 2:
heal_op = GF_SHD_OP_INDEX_SUMMARY;
@@ -1533,9 +1567,6 @@ main (int argc, char **argv)
case 3:
if (!strcmp (argv[2], "split-brain-info")) {
heal_op = GF_SHD_OP_SPLIT_BRAIN_FILES;
- } else if (!strcmp (argv[2], "xml")) {
- heal_op = GF_SHD_OP_INDEX_SUMMARY;
- is_xml = 1;
} else if (!strcmp (argv[2], "granular-entry-heal-op")) {
heal_op = GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE;
} else if (!strcmp (argv[2], "info-summary")) {
@@ -1547,15 +1578,7 @@ main (int argc, char **argv)
}
break;
case 4:
- if ((!strcmp (argv[2], "split-brain-info"))
- && (!strcmp (argv[3], "xml"))) {
- heal_op = GF_SHD_OP_SPLIT_BRAIN_FILES;
- is_xml = 1;
- } else if ((!strcmp (argv[2], "info-summary"))
- && (!strcmp (argv[3], "xml"))) {
- heal_op = GF_SHD_OP_HEAL_SUMMARY;
- is_xml = 1;
- } else if (!strcmp (argv[2], "bigger-file")) {
+ if (!strcmp (argv[2], "bigger-file")) {
heal_op = GF_SHD_OP_SBRAIN_HEAL_FROM_BIGGER_FILE;
file = argv[3];
} else if (!strcmp (argv[2], "latest-mtime")) {
@@ -1592,7 +1615,15 @@ main (int argc, char **argv)
glfsh_output = &glfsh_human_readable;
if (is_xml) {
#if (HAVE_LIB_XML)
- glfsh_output = &glfsh_xml_output;
+ if ((heal_op == GF_SHD_OP_INDEX_SUMMARY) ||
+ (heal_op == GF_SHD_OP_SPLIT_BRAIN_FILES) ||
+ (heal_op == GF_SHD_OP_HEAL_SUMMARY)) {
+ glfsh_output = &glfsh_xml_output;
+ } else {
+ printf(USAGE_STR, argv[0]);
+ ret = -1;
+ goto out;
+ }
#else
/*No point doing anything, just fail the command*/
exit (EXIT_FAILURE);
@@ -1636,7 +1667,7 @@ main (int argc, char **argv)
}
snprintf (logfilepath, sizeof (logfilepath),
DEFAULT_HEAL_LOG_FILE_DIRECTORY"/glfsheal-%s.log", volname);
- ret = glfs_set_logging(fs, logfilepath, GF_LOG_INFO);
+ ret = glfs_set_logging(fs, logfilepath, log_level);
if (ret < 0) {
ret = -errno;
gf_asprintf (&op_errstr, "Failed to set the log file path, "
--
1.8.3.1

74
0460-cli-add-a-warning-confirmation-message-in-peer-detac.patch Normal file
View File

@ -0,0 +1,74 @@
From 81b19743a97ebecc188d87fbe04dce59260824f8 Mon Sep 17 00:00:00 2001
From: Atin Mukherjee <amukherj@redhat.com>
Date: Tue, 6 Nov 2018 21:35:36 +0530
Subject: [PATCH 460/493] cli: add a warning/confirmation message in peer
detach code path
On a multi node cluster if one of the node is detached which had active
clients mounted through the same server address, this can cause all the
clients to loose any volfile changes. This is due to the lack of infra
in glusterd to let client know the list of IPs and attempt to connect to
other active nodes as failback. Such framework does exist in GD2 but not
in GD1.
This patch ensures to take a preventive measure to have a warning
message thrown to user to ensure all such connected clients are
remounted through a different IP.
> Change-Id: I740b01868abbd75bf0a609cfaf00114d4d78aa96
> Fixes: bz#1647074
> Signed-off-by: Atin Mukherjee <amukherj@redhat.com>
upstream patch: https://review.gluster.org/#/c/glusterfs/+/21572/
Change-Id: I740b01868abbd75bf0a609cfaf00114d4d78aa96
BUG: 1639568
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158630
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
cli/src/cli-cmd-peer.c | 14 +++++++++++++-
1 file changed, 13 insertions(+), 1 deletion(-)
diff --git a/cli/src/cli-cmd-peer.c b/cli/src/cli-cmd-peer.c
index 7df60bc..6f3b744 100644
--- a/cli/src/cli-cmd-peer.c
+++ b/cli/src/cli-cmd-peer.c
@@ -111,13 +111,20 @@ cli_cmd_peer_deprobe_cbk (struct cli_state *state, struct cli_cmd_word *word,
int sent = 0;
int parse_error = 0;
cli_local_t *local = NULL;
+ gf_answer_t answer = GF_ANSWER_NO;
+ const char *question = NULL;
if ((wordcount < 3) || (wordcount > 4)) {
cli_usage_out (word->pattern);
parse_error = 1;
goto out;
}
-
+ question =
+ "All clients mounted through the peer which is getting detached"
+ " need to be remounted using one of the other active peers in "
+ "the trusted storage pool to ensure client gets notification on"
+ " any changes done on the gluster configuration and if the "
+ "same has been done do you want to proceed?";
proc = &cli_rpc_prog->proctable[GLUSTER_CLI_DEPROBE];
frame = create_frame (THIS, THIS->ctx->pool);
@@ -149,6 +156,11 @@ cli_cmd_peer_deprobe_cbk (struct cli_state *state, struct cli_cmd_word *word,
ret = dict_set_int32 (dict, "flags", flags);
if (ret)
goto out;
+ answer = cli_cmd_get_confirmation(state, question);
+ if (GF_ANSWER_NO == answer) {
+ ret = 0;
+ goto out;
+ }
CLI_LOCAL_INIT (local, words, frame, dict);
--
1.8.3.1

835
0461-mount-fuse-Add-support-for-multi-threaded-fuse-reade.patch Normal file
View File

@ -0,0 +1,835 @@
From 668b55b7dd86b23e635cfb2264bc5e50f4cd888d Mon Sep 17 00:00:00 2001
From: Krutika Dhananjay <kdhananj@redhat.com>
Date: Tue, 9 Jan 2018 15:11:00 +0530
Subject: [PATCH 461/493] mount/fuse: Add support for multi-threaded fuse
readers
> Upstream: https://review.gluster.org/19226
> Github issue #412
> Change-Id: I94aa1505e5ae6a133683d473e0e4e0edd139b76b
Usage: Use 'reader-thread-count=<NUM>' as command line option to
set the thread count at the time of mounting the volume.
Next task is to make these threads auto-scale based on the load,
instead of having the user remount the volume everytime to change
the thread count.
Change-Id: I94aa1505e5ae6a133683d473e0e4e0edd139b76b
BUG: 1651040
Signed-off-by: Krutika Dhananjay <kdhananj@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158514
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
glusterfsd/src/glusterfsd.c | 26 ++++
glusterfsd/src/glusterfsd.h | 1 +
libglusterfs/src/glusterfs.h | 1 +
xlators/mount/fuse/src/fuse-bridge.c | 231 ++++++++++++++++++----------
xlators/mount/fuse/src/fuse-bridge.h | 9 +-
xlators/mount/fuse/src/fuse-helpers.c | 3 +
xlators/mount/fuse/src/fuse-mem-types.h | 1 +
xlators/mount/fuse/utils/mount.glusterfs.in | 7 +
8 files changed, 196 insertions(+), 83 deletions(-)
diff --git a/glusterfsd/src/glusterfsd.c b/glusterfsd/src/glusterfsd.c
index 78f3719..03bca24 100644
--- a/glusterfsd/src/glusterfsd.c
+++ b/glusterfsd/src/glusterfsd.c
@@ -238,6 +238,8 @@ static struct argp_option gf_options[] = {
"Enable localtime logging"},
{"event-history", ARGP_FUSE_EVENT_HISTORY_KEY, "BOOL",
OPTION_ARG_OPTIONAL, "disable/enable fuse event-history"},
+ {"reader-thread-count", ARGP_READER_THREAD_COUNT_KEY, "INTEGER",
+ OPTION_ARG_OPTIONAL, "set fuse reader thread count"},
{0, 0, 0, 0, "Miscellaneous Options:"},
{0, }
};
@@ -557,6 +559,17 @@ set_fuse_mount_options (glusterfs_ctx_t *ctx, dict_t *options)
goto err;
}
}
+ if (cmd_args->reader_thread_count) {
+ ret = dict_set_uint32 (options, "reader-thread-count",
+ cmd_args->reader_thread_count);
+ if (ret < 0) {
+ gf_msg ("glusterfsd", GF_LOG_ERROR, 0, glusterfsd_msg_4,
+ "failed to set dict value for key "
+ "reader-thread-count");
+ goto err;
+ }
+ }
+
ret = 0;
err:
return ret;
@@ -1307,6 +1320,19 @@ no_oom_api:
argp_failure (state, -1, 0,
"unknown event-history setting \"%s\"", arg);
break;
+ case ARGP_READER_THREAD_COUNT_KEY:
+ if (gf_string2uint32 (arg, &cmd_args->reader_thread_count)) {
+ argp_failure (state, -1, 0,
+ "unknown reader thread count option %s",
+ arg);
+ } else if ((cmd_args->reader_thread_count < 1) ||
+ (cmd_args->reader_thread_count > 64)) {
+ argp_failure (state, -1, 0,
+ "Invalid reader thread count %s. "
+ "Valid range: [\"1, 64\"]", arg);
+ }
+
+ break;
}
return 0;
diff --git a/glusterfsd/src/glusterfsd.h b/glusterfsd/src/glusterfsd.h
index f66947b..75cb1d8 100644
--- a/glusterfsd/src/glusterfsd.h
+++ b/glusterfsd/src/glusterfsd.h
@@ -99,6 +99,7 @@ enum argp_option_keys {
ARGP_LOCALTIME_LOGGING_KEY = 177,
ARGP_SUBDIR_MOUNT_KEY = 178,
ARGP_FUSE_EVENT_HISTORY_KEY = 179,
+ ARGP_READER_THREAD_COUNT_KEY = 180,
};
struct _gfd_vol_top_priv {
diff --git a/libglusterfs/src/glusterfs.h b/libglusterfs/src/glusterfs.h
index 5e641fd..3e2f426 100644
--- a/libglusterfs/src/glusterfs.h
+++ b/libglusterfs/src/glusterfs.h
@@ -446,6 +446,7 @@ struct _cmd_args {
char *subdir_mount;
char *event_history;
+ uint32_t reader_thread_count;
};
typedef struct _cmd_args cmd_args_t;
diff --git a/xlators/mount/fuse/src/fuse-bridge.c b/xlators/mount/fuse/src/fuse-bridge.c
index fbb4c53..8d1e3a0 100644
--- a/xlators/mount/fuse/src/fuse-bridge.c
+++ b/xlators/mount/fuse/src/fuse-bridge.c
@@ -665,7 +665,8 @@ fuse_lookup_resume (fuse_state_t *state)
}
static void
-fuse_lookup (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_lookup (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
char *name = msg;
fuse_state_t *state = NULL;
@@ -693,7 +694,8 @@ do_forget(xlator_t *this, uint64_t unique, uint64_t nodeid, uint64_t nlookup)
}
static void
-fuse_forget (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_forget (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_forget_in *ffi = msg;
@@ -714,7 +716,8 @@ fuse_forget (xlator_t *this, fuse_in_header_t *finh, void *msg)
#if FUSE_KERNEL_MINOR_VERSION >= 16
static void
-fuse_batch_forget(xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_batch_forget(xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_batch_forget_in *fbfi = msg;
struct fuse_forget_one *ffo = (struct fuse_forget_one *) (fbfi + 1);
@@ -932,7 +935,8 @@ fuse_getattr_resume (fuse_state_t *state)
}
static void
-fuse_getattr (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_getattr (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
#if FUSE_KERNEL_MINOR_VERSION >= 9
struct fuse_getattr_in *fgi = msg;
@@ -1265,7 +1269,8 @@ fuse_setattr_resume (fuse_state_t *state)
}
static void
-fuse_setattr (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_setattr (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_setattr_in *fsi = msg;
@@ -1492,7 +1497,8 @@ fuse_access_resume (fuse_state_t *state)
}
static void
-fuse_access (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_access (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_access_in *fai = msg;
fuse_state_t *state = NULL;
@@ -1566,7 +1572,8 @@ fuse_readlink_resume (fuse_state_t *state)
}
static void
-fuse_readlink (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_readlink (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
fuse_state_t *state = NULL;
@@ -1616,7 +1623,8 @@ fuse_mknod_resume (fuse_state_t *state)
}
static void
-fuse_mknod (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_mknod (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_mknod_in *fmi = msg;
char *name = (char *)(fmi + 1);
@@ -1686,7 +1694,8 @@ fuse_mkdir_resume (fuse_state_t *state)
}
static void
-fuse_mkdir (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_mkdir (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_mkdir_in *fmi = msg;
char *name = (char *)(fmi + 1);
@@ -1738,7 +1747,8 @@ fuse_unlink_resume (fuse_state_t *state)
}
static void
-fuse_unlink (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_unlink (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
char *name = msg;
fuse_state_t *state = NULL;
@@ -1775,7 +1785,8 @@ fuse_rmdir_resume (fuse_state_t *state)
}
static void
-fuse_rmdir (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_rmdir (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
char *name = msg;
fuse_state_t *state = NULL;
@@ -1825,7 +1836,8 @@ fuse_symlink_resume (fuse_state_t *state)
}
static void
-fuse_symlink (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_symlink (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
char *name = msg;
char *linkname = name + strlen (name) + 1;
@@ -1947,7 +1959,8 @@ fuse_rename_resume (fuse_state_t *state)
}
static void
-fuse_rename (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_rename (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_rename_in *fri = msg;
char *oldname = (char *)(fri + 1);
@@ -1997,7 +2010,8 @@ fuse_link_resume (fuse_state_t *state)
}
static void
-fuse_link (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_link (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_link_in *fli = msg;
char *name = (char *)(fli + 1);
@@ -2186,7 +2200,8 @@ fuse_create_resume (fuse_state_t *state)
}
static void
-fuse_create (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_create (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
#if FUSE_KERNEL_MINOR_VERSION >= 12
struct fuse_create_in *fci = msg;
@@ -2280,7 +2295,8 @@ fuse_open_resume (fuse_state_t *state)
}
static void
-fuse_open (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_open (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_open_in *foi = msg;
fuse_state_t *state = NULL;
@@ -2357,7 +2373,8 @@ fuse_readv_resume (fuse_state_t *state)
}
static void
-fuse_readv (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_readv (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_read_in *fri = msg;
@@ -2433,8 +2450,6 @@ void
fuse_write_resume (fuse_state_t *state)
{
struct iobref *iobref = NULL;
- struct iobuf *iobuf = NULL;
-
iobref = iobref_new ();
if (!iobref) {
@@ -2447,8 +2462,7 @@ fuse_write_resume (fuse_state_t *state)
return;
}
- iobuf = ((fuse_private_t *) (state->this->private))->iobuf;
- iobref_add (iobref, iobuf);
+ iobref_add (iobref, state->iobuf);
gf_log ("glusterfs-fuse", GF_LOG_TRACE,
"%"PRIu64": WRITE (%p, size=%"GF_PRI_SIZET", offset=%"PRId64")",
@@ -2462,7 +2476,8 @@ fuse_write_resume (fuse_state_t *state)
}
static void
-fuse_write (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_write (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
/* WRITE is special, metadata is attached to in_header,
* and msg is the payload as-is.
@@ -2505,6 +2520,7 @@ fuse_write (xlator_t *this, fuse_in_header_t *finh, void *msg)
state->vector.iov_base = msg;
state->vector.iov_len = fwi->size;
+ state->iobuf = iobuf;
fuse_resolve_and_resume (state, fuse_write_resume);
@@ -2543,7 +2559,8 @@ fuse_lseek_resume (fuse_state_t *state)
}
static void
-fuse_lseek (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_lseek (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_lseek_in *ffi = msg;
fuse_state_t *state = NULL;
@@ -2579,7 +2596,8 @@ fuse_flush_resume (fuse_state_t *state)
}
static void
-fuse_flush (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_flush (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_flush_in *ffi = msg;
@@ -2615,7 +2633,8 @@ fuse_internal_release (xlator_t *this, fd_t *fd)
}
static void
-fuse_release (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_release (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_release_in *fri = msg;
fd_t *fd = NULL;
@@ -2660,7 +2679,8 @@ fuse_fsync_resume (fuse_state_t *state)
}
static void
-fuse_fsync (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_fsync (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_fsync_in *fsi = msg;
@@ -2735,7 +2755,8 @@ fuse_opendir_resume (fuse_state_t *state)
}
static void
-fuse_opendir (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_opendir (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
/*
struct fuse_open_in *foi = msg;
@@ -2877,7 +2898,8 @@ fuse_readdir_resume (fuse_state_t *state)
}
static void
-fuse_readdir (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_readdir (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_read_in *fri = msg;
@@ -3028,7 +3050,8 @@ fuse_readdirp_resume (fuse_state_t *state)
static void
-fuse_readdirp (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_readdirp (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_read_in *fri = msg;
@@ -3075,7 +3098,8 @@ fuse_fallocate_resume(fuse_state_t *state)
}
static void
-fuse_fallocate(xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_fallocate(xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_fallocate_in *ffi = msg;
fuse_state_t *state = NULL;
@@ -3093,7 +3117,8 @@ fuse_fallocate(xlator_t *this, fuse_in_header_t *finh, void *msg)
#endif /* FUSE minor version >= 19 */
static void
-fuse_releasedir (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_releasedir (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_release_in *fri = msg;
fuse_state_t *state = NULL;
@@ -3134,7 +3159,8 @@ fuse_fsyncdir_resume (fuse_state_t *state)
}
static void
-fuse_fsyncdir (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_fsyncdir (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_fsync_in *fsi = msg;
@@ -3221,7 +3247,8 @@ fuse_statfs_resume (fuse_state_t *state)
static void
-fuse_statfs (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_statfs (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
fuse_state_t *state = NULL;
@@ -3273,7 +3300,8 @@ fuse_setxattr_resume (fuse_state_t *state)
static void
-fuse_setxattr (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_setxattr (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_setxattr_in *fsi = msg;
char *name = (char *)(fsi + 1);
@@ -3604,7 +3632,8 @@ fuse_getxattr_resume (fuse_state_t *state)
static void
-fuse_getxattr (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_getxattr (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_getxattr_in *fgxi = msg;
char *name = (char *)(fgxi + 1);
@@ -3710,7 +3739,8 @@ fuse_listxattr_resume (fuse_state_t *state)
static void
-fuse_listxattr (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_listxattr (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_getxattr_in *fgxi = msg;
fuse_state_t *state = NULL;
@@ -3766,7 +3796,8 @@ fuse_removexattr_resume (fuse_state_t *state)
static void
-fuse_removexattr (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_removexattr (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
char *name = msg;
@@ -3865,7 +3896,8 @@ fuse_getlk_resume (fuse_state_t *state)
static void
-fuse_getlk (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_getlk (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_lk_in *fli = msg;
@@ -3957,7 +3989,8 @@ fuse_setlk_resume (fuse_state_t *state)
static void
-fuse_setlk (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_setlk (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_lk_in *fli = msg;
@@ -4056,7 +4089,8 @@ notify_kernel_loop (void *data)
#endif
static void
-fuse_init (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_init (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
struct fuse_init_in *fini = msg;
struct fuse_init_out fino = {0,};
@@ -4227,7 +4261,8 @@ fuse_init (xlator_t *this, fuse_in_header_t *finh, void *msg)
static void
-fuse_enosys (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_enosys (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
send_fuse_err (this, finh, ENOSYS);
@@ -4236,7 +4271,8 @@ fuse_enosys (xlator_t *this, fuse_in_header_t *finh, void *msg)
static void
-fuse_destroy (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_destroy (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
send_fuse_err (this, finh, 0);
@@ -4826,6 +4862,7 @@ fuse_graph_sync (xlator_t *this)
new_graph_id = priv->next_graph->id;
priv->next_graph = NULL;
need_first_lookup = 1;
+ priv->handle_graph_switch = _gf_true;
while (!priv->event_recvd) {
ret = pthread_cond_wait (&priv->sync_cond,
@@ -4854,6 +4891,8 @@ unlock:
{
old_subvol->switched = 1;
winds_on_old_subvol = old_subvol->winds;
+ priv->handle_graph_switch = _gf_false;
+ pthread_cond_broadcast (&priv->migrate_cond);
}
pthread_mutex_unlock (&priv->sync_mutex);
@@ -4861,6 +4900,13 @@ unlock:
xlator_notify (old_subvol, GF_EVENT_PARENT_DOWN,
old_subvol, NULL);
}
+ } else {
+ pthread_mutex_lock (&priv->sync_mutex);
+ {
+ priv->handle_graph_switch = _gf_false;
+ pthread_cond_broadcast (&priv->migrate_cond);
+ }
+ pthread_mutex_unlock (&priv->sync_mutex);
}
return 0;
@@ -4897,7 +4943,6 @@ fuse_thread_proc (void *data)
const size_t msg0_size = sizeof (*finh) + 128;
fuse_handler_t **fuse_ops = NULL;
struct pollfd pfd[2] = {{0,}};
- gf_boolean_t mount_finished = _gf_false;
this = data;
priv = this->private;
@@ -4914,32 +4959,40 @@ fuse_thread_proc (void *data)
/* THIS has to be reset here */
THIS = this;
- if (!mount_finished) {
- memset(pfd,0,sizeof(pfd));
- pfd[0].fd = priv->status_pipe[0];
- pfd[0].events = POLLIN | POLLHUP | POLLERR;
- pfd[1].fd = priv->fd;
- pfd[1].events = POLLIN | POLLHUP | POLLERR;
- if (poll(pfd,2,-1) < 0) {
- gf_log (this->name, GF_LOG_ERROR,
- "poll error %s", strerror(errno));
- break;
- }
- if (pfd[0].revents & POLLIN) {
- if (fuse_get_mount_status(this) != 0) {
+ pthread_mutex_lock (&priv->sync_mutex);
+ {
+ if (!priv->mount_finished) {
+ memset(pfd, 0, sizeof(pfd));
+ pfd[0].fd = priv->status_pipe[0];
+ pfd[0].events = POLLIN | POLLHUP | POLLERR;
+ pfd[1].fd = priv->fd;
+ pfd[1].events = POLLIN | POLLHUP | POLLERR;
+ if (poll(pfd, 2, -1) < 0) {
+ gf_log (this->name, GF_LOG_ERROR,
+ "poll error %s",
+ strerror(errno));
+ pthread_mutex_unlock (&priv->sync_mutex);
break;
}
- mount_finished = _gf_true;
- }
- else if (pfd[0].revents) {
- gf_log (this->name, GF_LOG_ERROR,
- "mount pipe closed without status");
- break;
- }
- if (!pfd[1].revents) {
- continue;
+ if (pfd[0].revents & POLLIN) {
+ if (fuse_get_mount_status(this) != 0) {
+ pthread_mutex_unlock (&priv->sync_mutex);
+ break;
+ }
+ priv->mount_finished = _gf_true;
+ } else if (pfd[0].revents) {
+ gf_log (this->name, GF_LOG_ERROR,
+ "mount pipe closed without status");
+ pthread_mutex_unlock (&priv->sync_mutex);
+ break;
+ }
+ if (!pfd[1].revents) {
+ pthread_mutex_unlock (&priv->sync_mutex);
+ continue;
+ }
}
}
+ pthread_mutex_unlock (&priv->sync_mutex);
/*
* We don't want to block on readv while we're still waiting
@@ -5034,8 +5087,6 @@ fuse_thread_proc (void *data)
break;
}
- priv->iobuf = iobuf;
-
/*
* This can be moved around a bit, but it's important to do it
* *after* the readv. Otherwise, a graph switch could occur
@@ -5078,9 +5129,9 @@ fuse_thread_proc (void *data)
if (finh->opcode >= FUSE_OP_HIGH)
/* turn down MacFUSE specific messages */
- fuse_enosys (this, finh, msg);
+ fuse_enosys (this, finh, msg, NULL);
else
- fuse_ops[finh->opcode] (this, finh, msg);
+ fuse_ops[finh->opcode] (this, finh, msg, iobuf);
iobuf_unref (iobuf);
continue;
@@ -5152,8 +5203,6 @@ fuse_priv_dump (xlator_t *this)
private->volfile_size);
gf_proc_dump_write("mount_point", "%s",
private->mount_point);
- gf_proc_dump_write("iobuf", "%p",
- private->iobuf);
gf_proc_dump_write("fuse_thread_started", "%d",
(int)private->fuse_thread_started);
gf_proc_dump_write("direct_io_mode", "%d",
@@ -5279,6 +5328,7 @@ unlock:
int
notify (xlator_t *this, int32_t event, void *data, ...)
{
+ int i = 0;
int32_t ret = 0;
fuse_private_t *private = NULL;
gf_boolean_t start_thread = _gf_false;
@@ -5327,14 +5377,21 @@ notify (xlator_t *this, int32_t event, void *data, ...)
pthread_mutex_unlock (&private->sync_mutex);
if (start_thread) {
- ret = gf_thread_create (&private->fuse_thread, NULL,
- fuse_thread_proc, this,
- "fuseproc");
- if (ret != 0) {
- gf_log (this->name, GF_LOG_DEBUG,
- "pthread_create() failed (%s)",
- strerror (errno));
- break;
+ private->fuse_thread = GF_CALLOC (private->reader_thread_count,
+ sizeof (pthread_t),
+ gf_fuse_mt_pthread_t);
+ for (i = 0; i < private->reader_thread_count; i++) {
+
+ ret = gf_thread_create (&private->fuse_thread[i],
+ NULL,
+ fuse_thread_proc, this,
+ "fuseproc");
+ if (ret != 0) {
+ gf_log (this->name, GF_LOG_DEBUG,
+ "pthread_create() failed (%s)",
+ strerror (errno));
+ break;
+ }
}
}
@@ -5441,7 +5498,8 @@ static fuse_handler_t *fuse_dump_ops[FUSE_OP_HIGH];
static void
-fuse_dumper (xlator_t *this, fuse_in_header_t *finh, void *msg)
+fuse_dumper (xlator_t *this, fuse_in_header_t *finh, void *msg,
+ struct iobuf *iobuf)
{
fuse_private_t *priv = NULL;
struct iovec diov[6] = {{0,},};
@@ -5473,7 +5531,7 @@ fuse_dumper (xlator_t *this, fuse_in_header_t *finh, void *msg)
"failed to dump fuse message (R): %s",
strerror (errno));
- priv->fuse_ops0[finh->opcode] (this, finh, msg);
+ priv->fuse_ops0[finh->opcode] (this, finh, msg, NULL);
}
@@ -5578,6 +5636,9 @@ init (xlator_t *this_xl)
GF_OPTION_INIT (ZR_ATTR_TIMEOUT_OPT, priv->attribute_timeout, double,
cleanup_exit);
+ GF_OPTION_INIT ("reader-thread-count", priv->reader_thread_count, uint32,
+ cleanup_exit);
+
GF_OPTION_INIT (ZR_ENTRY_TIMEOUT_OPT, priv->entry_timeout, double,
cleanup_exit);
@@ -5793,6 +5854,7 @@ init (xlator_t *this_xl)
pthread_mutex_init (&priv->fuse_dump_mutex, NULL);
pthread_cond_init (&priv->sync_cond, NULL);
+ pthread_cond_init (&priv->migrate_cond, NULL);
pthread_mutex_init (&priv->sync_mutex, NULL);
priv->event_recvd = 0;
@@ -5992,5 +6054,12 @@ struct volume_options options[] = {
.description = "This option can be used to enable or disable fuse "
"event history.",
},
+ { .key = {"reader-thread-count"},
+ .type = GF_OPTION_TYPE_INT,
+ .default_value = "1",
+ .min = 1,
+ .max = 64,
+ .description = "Sets fuse reader thread count.",
+ },
{ .key = {NULL} },
};
diff --git a/xlators/mount/fuse/src/fuse-bridge.h b/xlators/mount/fuse/src/fuse-bridge.h
index 2dfef64..4ca76e9 100644
--- a/xlators/mount/fuse/src/fuse-bridge.h
+++ b/xlators/mount/fuse/src/fuse-bridge.h
@@ -52,7 +52,7 @@
typedef struct fuse_in_header fuse_in_header_t;
typedef void (fuse_handler_t) (xlator_t *this, fuse_in_header_t *finh,
- void *msg);
+ void *msg, struct iobuf *iobuf);
struct fuse_private {
int fd;
@@ -62,7 +62,8 @@ struct fuse_private {
char *mount_point;
struct iobuf *iobuf;
- pthread_t fuse_thread;
+ pthread_t *fuse_thread;
+ uint32_t reader_thread_count;
char fuse_thread_started;
uint32_t direct_io_mode;
@@ -140,6 +141,9 @@ struct fuse_private {
/* whether to run the unmount daemon */
gf_boolean_t auto_unmount;
+ gf_boolean_t mount_finished;
+ gf_boolean_t handle_graph_switch;
+ pthread_cond_t migrate_cond;
};
typedef struct fuse_private fuse_private_t;
@@ -391,6 +395,7 @@ typedef struct {
int32_t fd_no;
gf_seek_what_t whence;
+ struct iobuf *iobuf;
} fuse_state_t;
typedef struct {
diff --git a/xlators/mount/fuse/src/fuse-helpers.c b/xlators/mount/fuse/src/fuse-helpers.c
index c59ff77..c2d4d0c 100644
--- a/xlators/mount/fuse/src/fuse-helpers.c
+++ b/xlators/mount/fuse/src/fuse-helpers.c
@@ -123,6 +123,9 @@ get_fuse_state (xlator_t *this, fuse_in_header_t *finh)
pthread_mutex_lock (&priv->sync_mutex);
{
+ while (priv->handle_graph_switch)
+ pthread_cond_wait (&priv->migrate_cond,
+ &priv->sync_mutex);
active_subvol = fuse_active_subvol (state->this);
active_subvol->winds++;
}
diff --git a/xlators/mount/fuse/src/fuse-mem-types.h b/xlators/mount/fuse/src/fuse-mem-types.h
index 2b4b473..721b9a3 100644
--- a/xlators/mount/fuse/src/fuse-mem-types.h
+++ b/xlators/mount/fuse/src/fuse-mem-types.h
@@ -23,6 +23,7 @@ enum gf_fuse_mem_types_ {
gf_fuse_mt_graph_switch_args_t,
gf_fuse_mt_gids_t,
gf_fuse_mt_invalidate_node_t,
+ gf_fuse_mt_pthread_t,
gf_fuse_mt_end
};
#endif
diff --git a/xlators/mount/fuse/utils/mount.glusterfs.in b/xlators/mount/fuse/utils/mount.glusterfs.in
index b39bb98..817619e 100755
--- a/xlators/mount/fuse/utils/mount.glusterfs.in
+++ b/xlators/mount/fuse/utils/mount.glusterfs.in
@@ -221,6 +221,10 @@ start_glusterfs ()
cmd_line=$(echo "$cmd_line --event-history=$event_history");
fi
+ if [ -n "$reader_thread_count" ]; then
+ cmd_line=$(echo "$cmd_line --reader-thread-count=$reader_thread_count");
+ fi
+
if [ -n "$volume_name" ]; then
cmd_line=$(echo "$cmd_line --volume-name=$volume_name");
fi
@@ -496,6 +500,9 @@ with_options()
"event-history")
event_history=$value
;;
+ "reader-thread-count")
+ reader_thread_count=$value
+ ;;
"no-root-squash")
if [ $value = "yes" ] ||
[ $value = "on" ] ||
--
1.8.3.1

62
0462-posix-Do-not-log-ENXIO-errors-for-seek-fop.patch Normal file
View File

@ -0,0 +1,62 @@
From a064614b60924c4b0b1dbc4dd18278ce18b46db0 Mon Sep 17 00:00:00 2001
From: Xavi Hernandez <xhernandez@redhat.com>
Date: Fri, 6 Jul 2018 23:26:41 +0200
Subject: [PATCH 462/493] posix: Do not log ENXIO errors for seek fop
When lseek is used with SEEK_DATA and SEEK_HOLE, it's expected that the
last operation fails with ENXIO when offset is beyond the end of file.
In this case it doesn't make sense to report this as an error log message.
This patch reports ENXIO failure messages for seek fops in debug level
instead of error level.
> Change-Id: I62a4f61f99b0e4d7ea6a2cdcd40afe15072794ac
> fixes: bz#1598926
> Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
Upstream patch: https://review.gluster.org/c/glusterfs/+/20475
Change-Id: I62a4f61f99b0e4d7ea6a2cdcd40afe15072794ac
BUG: 1598883
Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158531
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
libglusterfs/src/common-utils.c | 6 ++++++
xlators/storage/posix/src/posix.c | 3 ++-
2 files changed, 8 insertions(+), 1 deletion(-)
diff --git a/libglusterfs/src/common-utils.c b/libglusterfs/src/common-utils.c
index dd6cdb3..25600a9 100644
--- a/libglusterfs/src/common-utils.c
+++ b/libglusterfs/src/common-utils.c
@@ -4369,6 +4369,12 @@ fop_log_level (glusterfs_fop_t fop, int op_errno)
if (op_errno == EEXIST)
return GF_LOG_DEBUG;
+ if (fop == GF_FOP_SEEK) {
+ if (op_errno == ENXIO) {
+ return GF_LOG_DEBUG;
+ }
+ }
+
return GF_LOG_ERROR;
}
diff --git a/xlators/storage/posix/src/posix.c b/xlators/storage/posix/src/posix.c
index e46fe99..13b4aa6 100644
--- a/xlators/storage/posix/src/posix.c
+++ b/xlators/storage/posix/src/posix.c
@@ -1225,7 +1225,8 @@ posix_seek (call_frame_t *frame, xlator_t *this, fd_t *fd, off_t offset,
ret = sys_lseek (pfd->fd, offset, whence);
if (ret == -1) {
err = errno;
- gf_msg (this->name, GF_LOG_ERROR, err, P_MSG_SEEK_FAILED,
+ gf_msg (this->name, fop_log_level(GF_FOP_SEEK, err), err,
+ P_MSG_SEEK_FAILED,
"seek failed on fd %d length %" PRId64 , pfd->fd,
offset);
goto out;
--
1.8.3.1

58
0463-build-glusterfs.spec-.in-firewalld-file-doesn-t-use-.patch Normal file
View File

@ -0,0 +1,58 @@
From f1d10db3bf315bfc9640a532aa39b9248d55e9c6 Mon Sep 17 00:00:00 2001
From: Kaleb S KEITHLEY <kkeithle@redhat.com>
Date: Thu, 13 Dec 2018 11:24:05 +0530
Subject: [PATCH 463/493] build: glusterfs.spec(.in) firewalld file doesn't use
%{_prefix}
.../firewalld/services/glusterfs.xml in %server files section
does not use %{_prefix}
Other firewalld files, e.g., in the firewalld.src.rpm file use
%{_prefix}
N.B. the other 'hardcoded' path, "/sbin/mount.glusterfs" in the
%files fuse section must remain as is, there is no macro for
mount utilities to use as there is for, e.g., the %{_sbindir}
macro for /usr/sbin. This doesn't matter for either RHEL6, where
/sbin and /usr/sbin are distinct directories, or for Fedora and
RHEL7, where /sbin is a symlink to /usr/sbin. E.g. see the nfs-
utils.src.rpm where /sbin/mount.nfs is also 'hardcoded'.
mainline:
> Change-Id: I902f47e3c589526c774d0aceb6fc2815abf86b01
> BUG: 1350793
> Signed-off-by: Kaleb S KEITHLEY <kkeithle@redhat.com>
> Reviewed-on: http://review.gluster.org/14823
> Smoke: Gluster Build System <jenkins@build.gluster.org>
> CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
> NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
> Reviewed-by: Milind Changire <mchangir@redhat.com>
> Reviewed-by: Niels de Vos <ndevos@redhat.com>
Change-Id: I902f47e3c589526c774d0aceb6fc2815abf86b01
BUG: 1350745
Signed-off-by: Kaleb S KEITHLEY <kkeithle@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158502
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Tested-by: Milind Changire <mchangir@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
glusterfs.spec.in | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/glusterfs.spec.in b/glusterfs.spec.in
index b6b7630..2745b50 100644
--- a/glusterfs.spec.in
+++ b/glusterfs.spec.in
@@ -1168,7 +1168,7 @@ exit 0
%exclude %{_tmpfilesdir}/gluster.conf
%endif
%if ( 0%{?_with_firewalld:1} )
-%exclude /usr/lib/firewalld/services/glusterfs.xml
+%exclude %{_prefix}/lib/firewalld/services/glusterfs.xml
%endif
%endif
%doc ChangeLog COPYING-GPLV2 COPYING-LGPLV3 INSTALL README.md THANKS
--
1.8.3.1

37
0464-build-exclude-packaging-crypt.so.patch Normal file
View File

@ -0,0 +1,37 @@
From 93f35b10446fe14a0fd3e5d318056fe399479386 Mon Sep 17 00:00:00 2001
From: Milind Changire <mchangir@redhat.com>
Date: Thu, 13 Dec 2018 11:31:20 +0530
Subject: [PATCH 464/493] build: exclude packaging crypt.so
exclude packaging crypt.so for FIPS compliance
Label: DOWNSTREAM ONLY
BUG: 1653224
Change-Id: Icbf0d3efc90813c5856237213e6cf25af84e4915
Signed-off-by: Milind Changire <mchangir@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158500
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
glusterfs.spec.in | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/glusterfs.spec.in b/glusterfs.spec.in
index 2745b50..baee2fa 100644
--- a/glusterfs.spec.in
+++ b/glusterfs.spec.in
@@ -1198,8 +1198,8 @@ exit 0
%{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/debug/trace.so
%if ( ! ( 0%{?rhel} && 0%{?rhel} < 6 ) )
# RHEL-5 based distributions have a too old openssl
-%dir %{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/encryption
- %{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/encryption/crypt.so
+%exclude %dir %{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/encryption
+%exclude %{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/encryption/crypt.so
%endif
%dir %{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/features
%{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/features/access-control.so
--
1.8.3.1

84
0465-build-add-missing-explicit-package-dependencies.patch Normal file
View File

@ -0,0 +1,84 @@
From ef7af50ae35ac5776057d6355b84ae111d33151e Mon Sep 17 00:00:00 2001
From: Milind Changire <mchangir@redhat.com>
Date: Thu, 13 Dec 2018 12:46:56 +0530
Subject: [PATCH 465/493] build: add missing explicit package dependencies
Add dependencies for glusterfs-libs, and other packages.
This is an Errata Tool whine.
Label: DOWNSTREAM ONLY
BUG: 1656357
Change-Id: Ieaadb6e4ffa84d1811aa740f7891855568ecbcbb
Signed-off-by: Milind Changire <mchangir@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158501
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
glusterfs.spec.in | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/glusterfs.spec.in b/glusterfs.spec.in
index baee2fa..f6a4ab0 100644
--- a/glusterfs.spec.in
+++ b/glusterfs.spec.in
@@ -282,6 +282,7 @@ Summary: GlusterFS api library
Group: System Environment/Daemons
Requires: %{name}%{?_isa} = %{version}-%{release}
Requires: %{name}-client-xlators%{?_isa} = %{version}-%{release}
+Requires: %{name}-libs%{?_isa} = %{version}-%{release}
%description api
GlusterFS is a distributed file-system capable of scaling to several
@@ -300,6 +301,7 @@ Group: Development/Libraries
Requires: %{name}%{?_isa} = %{version}-%{release}
Requires: %{name}-devel%{?_isa} = %{version}-%{release}
Requires: libacl-devel
+Requires: %{name}-api%{?_isa} = %{version}-%{release}
%description api-devel
GlusterFS is a distributed file-system capable of scaling to several
@@ -336,6 +338,8 @@ Requires: %{name}%{?_isa} = %{version}-%{release}
%if ( 0%{!?_without_extra_xlators:1} )
Requires: %{name}-extra-xlators = %{version}-%{release}
%endif
+Requires: %{name}-libs%{?_isa} = %{version}-%{release}
+Requires: %{name}-server%{?_isa} = %{version}-%{release}
%description devel
GlusterFS is a distributed file-system capable of scaling to several
@@ -385,6 +389,7 @@ Requires: %{name}-client-xlators%{?_isa} = %{version}-%{release}
Obsoletes: %{name}-client < %{version}-%{release}
Provides: %{name}-client = %{version}-%{release}
+Requires: %{name}-libs%{?_isa} = %{version}-%{release}
%description fuse
GlusterFS is a distributed file-system capable of scaling to several
@@ -454,6 +459,7 @@ BuildRequires: python-ctypes
Requires: python2-gluster = %{version}-%{release}
Requires: rsync
Requires: util-linux
+Requires: %{name}-libs%{?_isa} = %{version}-%{release}
%description geo-replication
GlusterFS is a distributed file-system capable of scaling to several
@@ -526,6 +532,7 @@ BuildRequires: libibverbs-devel
BuildRequires: librdmacm-devel >= 1.0.15
%endif
Requires: %{name}%{?_isa} = %{version}-%{release}
+Requires: %{name}-libs%{?_isa} = %{version}-%{release}
%description rdma
GlusterFS is a distributed file-system capable of scaling to several
@@ -656,6 +663,7 @@ This package provides the glusterfs server daemon.
%package client-xlators
Summary: GlusterFS client-side translators
Group: Applications/File
+Requires: %{name}-libs%{?_isa} = %{version}-%{release}
%description client-xlators
GlusterFS is a distributed file-system capable of scaling to several
--
1.8.3.1

86
0466-extras-Add-group-distributed-virt-for-single-brick-o.patch Normal file
View File

@ -0,0 +1,86 @@
From 2ef41e27b452f215e56bfc08c8117f8f18c33619 Mon Sep 17 00:00:00 2001
From: Krutika Dhananjay <kdhananj@redhat.com>
Date: Wed, 28 Nov 2018 12:39:31 +0530
Subject: [PATCH 466/493] extras: Add group-distributed-virt for single-brick
ovirt-gluster use-case
> Upstream: https://review.gluster.org/21735
> BUG: 1654138
> Change-Id: I930011327332b7ba30cc76f614efaf5932eb4f3d
Change-Id: I930011327332b7ba30cc76f614efaf5932eb4f3d
BUG: 1653613
Signed-off-by: Krutika Dhananjay <kdhananj@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158487
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Milind Changire <mchangir@redhat.com>
Reviewed-by: Atin Mukherjee <amukherj@redhat.com>
---
extras/Makefile.am | 6 +++++-
extras/group-distributed-virt | 10 ++++++++++
glusterfs.spec.in | 4 ++++
3 files changed, 19 insertions(+), 1 deletion(-)
create mode 100644 extras/group-distributed-virt
diff --git a/extras/Makefile.am b/extras/Makefile.am
index 7b791af..e0e05b5 100644
--- a/extras/Makefile.am
+++ b/extras/Makefile.am
@@ -12,7 +12,9 @@ SUBDIRS = init.d systemd benchmarking hook-scripts $(OCF_SUBDIR) LinuxRPM \
confdir = $(sysconfdir)/glusterfs
conf_DATA = glusterfs-logrotate gluster-rsyslog-7.2.conf gluster-rsyslog-5.8.conf \
- logger.conf.example glusterfs-georep-logrotate group-virt.example group-metadata-cache group-gluster-block group-nl-cache group-db-workload
+ logger.conf.example glusterfs-georep-logrotate group-virt.example \
+ group-metadata-cache group-gluster-block group-nl-cache group-db-workload \
+ group-distributed-virt
voldir = $(sysconfdir)/glusterfs
vol_DATA = glusterd.vol
@@ -49,3 +51,5 @@ install-data-local:
$(DESTDIR)$(GLUSTERD_WORKDIR)/groups/nl-cache
$(INSTALL_DATA) $(top_srcdir)/extras/group-db-workload \
$(DESTDIR)$(GLUSTERD_WORKDIR)/groups/db-workload
+ $(INSTALL_DATA) $(top_srcdir)/extras/group-distributed-virt \
+ $(DESTDIR)$(GLUSTERD_WORKDIR)/groups/distributed-virt
diff --git a/extras/group-distributed-virt b/extras/group-distributed-virt
new file mode 100644
index 0000000..a960b76
--- /dev/null
+++ b/extras/group-distributed-virt
@@ -0,0 +1,10 @@
+performance.quick-read=off
+performance.read-ahead=off
+performance.io-cache=off
+performance.low-prio-threads=32
+network.remote-dio=enable
+features.shard=on
+user.cifs=off
+client.event-threads=4
+server.event-threads=4
+performance.client-io-threads=on
diff --git a/glusterfs.spec.in b/glusterfs.spec.in
index f6a4ab0..a4accd9 100644
--- a/glusterfs.spec.in
+++ b/glusterfs.spec.in
@@ -1522,6 +1522,7 @@ exit 0
%attr(0644,-,-) %{_sharedstatedir}/glusterd/groups/metadata-cache
%attr(0644,-,-) %{_sharedstatedir}/glusterd/groups/gluster-block
%attr(0644,-,-) %{_sharedstatedir}/glusterd/groups/db-workload
+ %attr(0644,-,-) %{_sharedstatedir}/glusterd/groups/distributed-virt
%attr(0644,-,-) %{_sharedstatedir}/glusterd/groups/nl-cache
%dir %attr(0755,-,-) %{_sharedstatedir}/glusterd/glusterfind
%dir %attr(0755,-,-) %{_sharedstatedir}/glusterd/glusterfind/.keys
@@ -2169,6 +2170,9 @@ fi
%endif
%changelog
+* Thu Dec 13 2018 Krutika Dhananjay <kdhananj@redhat.com>
+- Install /var/lib/glusterd/groups/distributed-virt by default (#1653613)
+
* Fri Jul 6 2018 Atin Mukherjee <amukherj@redhat.com>
- Added db group profile (#1597506)
--
1.8.3.1

304
0467-glusterd-glusterd-to-regenerate-volfiles-when-GD_OP_.patch Normal file
View File

@ -0,0 +1,304 @@
From 647b4d4e8edefd256de2a9f3916763b8cfa8429b Mon Sep 17 00:00:00 2001
From: Atin Mukherjee <amukherj@redhat.com>
Date: Tue, 20 Nov 2018 12:32:32 +0530
Subject: [PATCH 467/493] glusterd: glusterd to regenerate volfiles when
GD_OP_VERSION_MAX changes
While glusterd has an infra to allow post install of spec to bring it up
in the interim upgrade mode to allow all the volfiles to be regenerated
with the latest executable, in container world the same methodology is
not followed as container image always point to the specific gluster rpm
and gluster rpm doesn't go through an upgrade process.
This fix does the following:
1. If glusterd.upgrade file doesn't exist, regenerate the volfiles
2. If maximum-operating-version read from glusterd.upgrade doesn't match
with GD_OP_VERSION_MAX, glusterd detects it to be a version where new
options are introduced and regenerate the volfiles.
Tests done:
1. Bring up glusterd, check if glusterd.upgrade file has been created
with GD_OP_VERSION_MAX value.
2. Post 1, restart glusterd and check glusterd hasn't regenerated the
volfiles as there's is no change in the GD_OP_VERSION_MAX vs the
op_version read from the file.
3. Bump up the GD_OP_VERSION_MAX in the code by 1 and post compilation
restart glusterd where the volfiles should be again regenerated.
Note: The old way of having volfiles regenerated during an rpm upgrade
is kept as it is for now but eventually this can be sunset later.
> Change-Id: I75b49a1601c71e99f6a6bc360dd12dd03a96414b
> Fixes: bz#1651463
> Signed-off-by: Atin Mukherjee <amukherj@redhat.com>
upstream patch: https://review.gluster.org/#/c/glusterfs/+/21687/
Change-Id: I75b49a1601c71e99f6a6bc360dd12dd03a96414b
BUG: 1651460
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158645
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Atin Mukherjee <amukherj@redhat.com>
---
xlators/mgmt/glusterd/src/glusterd-store.c | 126 +++++++++++++++++++++++++++--
xlators/mgmt/glusterd/src/glusterd-store.h | 6 ++
xlators/mgmt/glusterd/src/glusterd.c | 27 +++++--
xlators/mgmt/glusterd/src/glusterd.h | 7 ++
4 files changed, 154 insertions(+), 12 deletions(-)
diff --git a/xlators/mgmt/glusterd/src/glusterd-store.c b/xlators/mgmt/glusterd/src/glusterd-store.c
index 37542e7..f276fef 100644
--- a/xlators/mgmt/glusterd/src/glusterd-store.c
+++ b/xlators/mgmt/glusterd/src/glusterd-store.c
@@ -2063,7 +2063,7 @@ glusterd_store_global_info (xlator_t *this)
}
handle->fd = gf_store_mkstemp (handle);
- if (handle->fd <= 0) {
+ if (handle->fd < 0) {
ret = -1;
goto out;
}
@@ -2081,7 +2081,7 @@ glusterd_store_global_info (xlator_t *this)
goto out;
}
- snprintf (op_version_str, 15, "%d", conf->op_version);
+ snprintf (op_version_str, sizeof(op_version_str), "%d", conf->op_version);
ret = gf_store_save_value (handle->fd, GD_OP_VERSION_KEY,
op_version_str);
if (ret) {
@@ -2094,12 +2094,8 @@ glusterd_store_global_info (xlator_t *this)
ret = gf_store_rename_tmppath (handle);
out:
if (handle) {
- if (ret && (handle->fd > 0))
+ if (ret && (handle->fd >= 0))
gf_store_unlink_tmppath (handle);
-
- if (handle->fd > 0) {
- handle->fd = 0;
- }
}
if (uuid_str)
@@ -2114,6 +2110,122 @@ out:
}
int
+glusterd_store_max_op_version(xlator_t *this)
+{
+ int ret = -1;
+ glusterd_conf_t *conf = NULL;
+ char op_version_str[15] = {0,};
+ char path[PATH_MAX] = {0,};
+ gf_store_handle_t *handle = NULL;
+ int32_t len = 0;
+
+ conf = this->private;
+
+ len = snprintf(path, PATH_MAX, "%s/%s", conf->workdir,
+ GLUSTERD_UPGRADE_FILE);
+ if ((len < 0) || (len >= PATH_MAX)) {
+ goto out;
+ }
+ ret = gf_store_handle_new(path, &handle);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0,
+ GD_MSG_STORE_HANDLE_GET_FAIL, "Unable to get store "
+ "handle");
+ goto out;
+ }
+
+ /* These options need to be available for all users */
+ ret = sys_chmod(handle->path, 0644);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, errno, GD_MSG_FILE_OP_FAILED,
+ "chmod error for %s", GLUSTERD_UPGRADE_FILE);
+ goto out;
+ }
+
+ handle->fd = gf_store_mkstemp(handle);
+ if (handle->fd < 0) {
+ ret = -1;
+ goto out;
+ }
+
+ snprintf(op_version_str, sizeof(op_version_str), "%d",
+ GD_OP_VERSION_MAX);
+ ret = gf_store_save_value(handle->fd, GD_MAX_OP_VERSION_KEY,
+ op_version_str);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_OP_VERS_STORE_FAIL,
+ "Storing op-version failed ret = %d", ret);
+ goto out;
+ }
+
+ ret = gf_store_rename_tmppath(handle);
+out:
+ if (handle) {
+ if (ret && (handle->fd >= 0))
+ gf_store_unlink_tmppath(handle);
+ }
+
+ if (ret)
+ gf_msg(this->name, GF_LOG_ERROR, 0,
+ GD_MSG_GLUSTERD_GLOBAL_INFO_STORE_FAIL,
+ "Failed to store max op-version");
+ if (handle)
+ gf_store_handle_destroy(handle);
+ return ret;
+}
+
+int
+glusterd_retrieve_max_op_version(xlator_t *this, int *op_version)
+{
+ char *op_version_str = NULL;
+ glusterd_conf_t *priv = NULL;
+ int ret = -1;
+ int tmp_version = 0;
+ char *tmp = NULL;
+ char path[PATH_MAX] = {0,};
+ gf_store_handle_t *handle = NULL;
+ int32_t len = 0;
+
+ priv = this->private;
+
+ len = snprintf(path, PATH_MAX, "%s/%s", priv->workdir,
+ GLUSTERD_UPGRADE_FILE);
+ if ((len < 0) || (len >= PATH_MAX)) {
+ goto out;
+ }
+ ret = gf_store_handle_retrieve(path, &handle);
+
+ if (ret) {
+ gf_msg_debug(this->name, 0, "Unable to get store handle!");
+ goto out;
+ }
+
+ ret = gf_store_retrieve_value(handle, GD_MAX_OP_VERSION_KEY,
+ &op_version_str);
+ if (ret) {
+ gf_msg_debug(this->name, 0, "No previous op_version present");
+ goto out;
+ }
+
+ tmp_version = strtol(op_version_str, &tmp, 10);
+ if ((tmp_version <= 0) || (tmp && strlen(tmp) > 1)) {
+ gf_msg(this->name, GF_LOG_WARNING, EINVAL,
+ GD_MSG_UNSUPPORTED_VERSION, "invalid version number");
+ goto out;
+ }
+
+ *op_version = tmp_version;
+
+ ret = 0;
+out:
+ if (op_version_str)
+ GF_FREE(op_version_str);
+ if (handle)
+ gf_store_handle_destroy(handle);
+ return ret;
+}
+
+int
glusterd_retrieve_op_version (xlator_t *this, int *op_version)
{
char *op_version_str = NULL;
diff --git a/xlators/mgmt/glusterd/src/glusterd-store.h b/xlators/mgmt/glusterd/src/glusterd-store.h
index 383a475..76c5500 100644
--- a/xlators/mgmt/glusterd/src/glusterd-store.h
+++ b/xlators/mgmt/glusterd/src/glusterd-store.h
@@ -161,6 +161,12 @@ glusterd_retrieve_op_version (xlator_t *this, int *op_version);
int
glusterd_store_global_info (xlator_t *this);
+int
+glusterd_retrieve_max_op_version(xlator_t *this, int *op_version);
+
+int
+glusterd_store_max_op_version(xlator_t *this);
+
int32_t
glusterd_store_retrieve_options (xlator_t *this);
diff --git a/xlators/mgmt/glusterd/src/glusterd.c b/xlators/mgmt/glusterd/src/glusterd.c
index ca17526..29d5de1 100644
--- a/xlators/mgmt/glusterd/src/glusterd.c
+++ b/xlators/mgmt/glusterd/src/glusterd.c
@@ -1428,6 +1428,7 @@ init (xlator_t *this)
gf_boolean_t upgrade = _gf_false;
gf_boolean_t downgrade = _gf_false;
char *localtime_logging = NULL;
+ int op_version = 0;
#ifndef GF_DARWIN_HOST_OS
{
@@ -1976,6 +1977,27 @@ init (xlator_t *this)
}
GF_ATOMIC_INIT(conf->blockers, 0);
+ ret = glusterd_handle_upgrade_downgrade(this->options, conf, upgrade,
+ downgrade);
+ if (ret)
+ goto out;
+
+ ret = glusterd_retrieve_max_op_version(this, &op_version);
+ /* first condition indicates file isn't present which means this code
+ * change is hitting for the first time or someone has deleted it from
+ * the backend.second condition is when max op_version differs, in both
+ * cases volfiles should be regenerated
+ */
+ if (op_version == 0 || op_version != GD_OP_VERSION_MAX) {
+ gf_log(this->name, GF_LOG_INFO,
+ "Regenerating volfiles due to a max op-version mismatch "
+ "or glusterd.upgrade file not being present, op_version "
+ "retrieved: %d, max op_version: %d", op_version,
+ GD_OP_VERSION_MAX);
+ glusterd_recreate_volfiles(conf);
+ ret = glusterd_store_max_op_version(this);
+ }
+
/* If the peer count is less than 2 then this would be the best time to
* spawn process/bricks that may need (re)starting since last time
* (this) glusterd was up. */
@@ -1983,11 +2005,6 @@ init (xlator_t *this)
glusterd_launch_synctask (glusterd_spawn_daemons, NULL);
- ret = glusterd_handle_upgrade_downgrade (this->options, conf, upgrade,
- downgrade);
- if (ret)
- goto out;
-
ret = glusterd_hooks_spawn_worker (this);
if (ret)
goto out;
diff --git a/xlators/mgmt/glusterd/src/glusterd.h b/xlators/mgmt/glusterd/src/glusterd.h
index bfa8310..cbdca52 100644
--- a/xlators/mgmt/glusterd/src/glusterd.h
+++ b/xlators/mgmt/glusterd/src/glusterd.h
@@ -582,6 +582,9 @@ typedef enum {
#define GLUSTERD_DEFAULT_PORT GF_DEFAULT_BASE_PORT
#define GLUSTERD_INFO_FILE "glusterd.info"
+#define GLUSTERD_UPGRADE_FILE \
+ "glusterd.upgrade" /* zero byte file to detect a need for regenerating \
+ volfiles in container mode */
#define GLUSTERD_VOLUME_QUOTA_CONFIG "quota.conf"
#define GLUSTERD_VOLUME_DIR_PREFIX "vols"
#define GLUSTERD_PEER_DIR_PREFIX "peers"
@@ -1333,4 +1336,8 @@ glusterd_tier_prevalidate (dict_t *dict, char **op_errstr,
int
glusterd_options_init (xlator_t *this);
+
+int32_t
+glusterd_recreate_volfiles(glusterd_conf_t *conf);
+
#endif
--
1.8.3.1

40
0468-core-move-invalid-port-logs-to-DEBUG-log-level.patch Normal file
View File

@ -0,0 +1,40 @@
From 8eb95b2ebdedd61e7784cf2f18b9564d2d4ed0b9 Mon Sep 17 00:00:00 2001
From: Milind Changire <mchangir@redhat.com>
Date: Thu, 13 Dec 2018 10:11:58 +0530
Subject: [PATCH 468/493] core: move invalid port logs to DEBUG log level
Stop spamming "invalid port" logs in case sysadmin has reserved a large
number of ports.
manline:
> Change-Id: I244ef7693560cc404b36cadc6b05d92ec0e908d3
> fixes: bz#1656517
> Signed-off-by: Milind Changire <mchangir@redhat.com>
> Reviewed-on: https://review.gluster.org/c/glusterfs/+/21809
Change-Id: I244ef7693560cc404b36cadc6b05d92ec0e908d3
BUG: 1656217
Signed-off-by: Milind Changire <mchangir@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158483
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
libglusterfs/src/common-utils.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/libglusterfs/src/common-utils.c b/libglusterfs/src/common-utils.c
index 25600a9..1243754 100644
--- a/libglusterfs/src/common-utils.c
+++ b/libglusterfs/src/common-utils.c
@@ -3373,7 +3373,7 @@ gf_ports_reserved (char *blocked_port, unsigned char *ports, uint32_t ceiling)
if (blocked_port[strlen(blocked_port) -1] == '\n')
blocked_port[strlen(blocked_port) -1] = '\0';
if (gf_string2int32 (blocked_port, &tmp_port1) == 0) {
- if (tmp_port1 > ceiling
+ if (tmp_port1 > GF_PORT_MAX
|| tmp_port1 < 0) {
gf_msg ("glusterfs-socket", GF_LOG_WARNING, 0,
LG_MSG_INVALID_PORT, "invalid port %d",
--
1.8.3.1

48
0469-nfs-set-ctx-for-every-inode-looked-up-nfs3_fh_resolv.patch Normal file
View File

@ -0,0 +1,48 @@
From 4a3e8888d7e866137287fced284b71ba152a17ad Mon Sep 17 00:00:00 2001
From: Jiffin Tony Thottan <jthottan@redhat.com>
Date: Thu, 29 Nov 2018 19:22:40 +0530
Subject: [PATCH 469/493] nfs : set ctx for every inode looked up
nfs3_fh_resolve_inode_lookup_cbk()
The inode ctx for nfs xlator is set with help nfs_fix_generation.
But currently gnfs is crashing because inode_ctx is becoming null
nfs3_resolve_inode_hard() is used to perform a lookup on entire
path and looks like function is missing to set the ctx for inode.
This patch will set ctx for the inode which it looked on.
Upstream reference :
>url: https://review.gluster.org/#/c/glusterfs/+/21749/
>Change-Id: I464fa7f78df1bae990ebe97de8ccf6d5fb74fc9f
>fixes: bz#1651439
>Signed-off-by: Jiffin Tony Thottan <jthottan@redhat.com>
Change-Id: I464fa7f78df1bae990ebe97de8ccf6d5fb74fc9f
BUG: 1633177
Signed-off-by: Jiffin Tony Thottan <jthottan@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158676
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/nfs/server/src/nfs3-helpers.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/xlators/nfs/server/src/nfs3-helpers.c b/xlators/nfs/server/src/nfs3-helpers.c
index 0b97709..9bc8aff 100644
--- a/xlators/nfs/server/src/nfs3-helpers.c
+++ b/xlators/nfs/server/src/nfs3-helpers.c
@@ -3660,6 +3660,12 @@ nfs3_fh_resolve_entry_lookup_cbk (call_frame_t *frame, void *cookie,
inode_lookup (linked_inode);
inode_unref (cs->resolvedloc.inode);
cs->resolvedloc.inode = linked_inode;
+ } else {
+ /* nfs3_fh_resolve_entry_hard() use to resolve entire path if needed.
+ * So the ctx for inode obtained from here need to set properly,
+ * otherwise it may result in a crash.
+ */
+ nfs_fix_generation(this, inode);
}
err:
nfs3_call_resume (cs);
--
1.8.3.1

87
0470-dht-fix-use-after-free-in-dht_rmdir_readdirp_cbk.patch Normal file
View File

@ -0,0 +1,87 @@
From 870513f9bade449fa760a81e242102860a0fdc91 Mon Sep 17 00:00:00 2001
From: N Balachandran <nbalacha@redhat.com>
Date: Thu, 13 Dec 2018 10:54:15 +0530
Subject: [PATCH 470/493] dht: fix use after free in dht_rmdir_readdirp_cbk
The frame is freed when linkfile exist in dht_rmdir_is_subvol_empty(),
the following message use the freed local.
upstream patch: https://review.gluster.org/#/c/glusterfs/+/21446/
> Change-Id: I41191e8bd477f031a2444d5f15e578dc4f086e6b
> Updates: bz#1640489
> Signed-off-by: Kinglong Mee <mijinlong@open-fs.com>
Change-Id: Ia257e1da57cc486ab336e43f8e88187e984c70e2
BUG: 1654103
Author: Kinglong Mee <mijinlong@open-fs.com>
Signed-off-by: N Balachandran <nbalacha@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158486
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Raghavendra Gowdappa <rgowdapp@redhat.com>
---
xlators/cluster/dht/src/dht-common.c | 23 +++++++++++++----------
1 file changed, 13 insertions(+), 10 deletions(-)
diff --git a/xlators/cluster/dht/src/dht-common.c b/xlators/cluster/dht/src/dht-common.c
index 2e19036..ff0099c 100644
--- a/xlators/cluster/dht/src/dht-common.c
+++ b/xlators/cluster/dht/src/dht-common.c
@@ -10175,6 +10175,7 @@ dht_rmdir_readdirp_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
xlator_t *prev = NULL;
xlator_t *src = NULL;
int ret = 0;
+ char *path = NULL;
local = frame->local;
@@ -10182,6 +10183,11 @@ dht_rmdir_readdirp_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
src = prev;
if (op_ret > 2) {
+ /* dht_rmdir_is_subvol_empty() may free the frame,
+ * copy path for logging.
+ */
+ path = gf_strdup(local->loc.path);
+
ret = dht_rmdir_is_subvol_empty (frame, this, entries, src);
switch (ret) {
@@ -10192,27 +10198,24 @@ dht_rmdir_readdirp_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
local->loc.path, op_ret);
local->op_ret = -1;
local->op_errno = ENOTEMPTY;
- goto done;
+ break;
default:
/* @ret number of linkfiles are getting unlinked */
gf_msg_trace (this->name, 0,
"readdir on %s for %s found %d "
- "linkfiles", prev->name,
- local->loc.path, ret);
+ "linkfiles",
+ prev->name, path, ret);
break;
}
}
-
- if (ret) {
- return 0;
- }
-
-done:
/* readdirp failed or no linkto files were found on this subvol */
+ if (!ret) {
+ dht_rmdir_readdirp_done(frame, this);
+ }
+ GF_FREE(path);
- dht_rmdir_readdirp_done (frame, this);
return 0;
}
--
1.8.3.1

507
0471-glusterd-migrating-profile-commands-to-mgmt_v3-frame.patch Normal file
View File

@ -0,0 +1,507 @@
From e68845ff7018e5d81d7979684b18e6eda449b088 Mon Sep 17 00:00:00 2001
From: Sanju Rakonde <srakonde@redhat.com>
Date: Thu, 8 Nov 2018 18:50:18 +0530
Subject: [PATCH 471/493] glusterd: migrating profile commands to mgmt_v3
framework
Current profile commands use the op_state machine framework.
Porting it to use the mgmt_v3 framework.
The following tests were performed on the patch:
case 1:
1. On a 3 node cluster, created and started 3 volumes
2. Mounted all the three volumes and wrote some data
3. Started profile operation for all the volumes
4. Ran "gluster v status" from N1,
"gluster v profile <volname1> info" form N2,
"gluster v profile <volname2> info" from N3 simultaneously in a
loop for around 10000 times
5. Didn't find any cores generated.
case 2:
1. Repeat the steps 1,2 and 3 from case 1.
2. Ran "gluster v status" from N1,
"gluster v profile <volname1> info" form N2(terminal 1),
"gluster v profile <volname2> info" from N2(terminal 2)
simultaneously in a loop.
3. No cores were generated.
> fixes: bz#1654181
> Change-Id: I83044cf5aee3970ef94066c89fcc41783ed468a6
> Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
upstream patch: https://review.gluster.org/#/c/glusterfs/+/21736/
Change-Id: I83044cf5aee3970ef94066c89fcc41783ed468a6
BUG: 1639476
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158631
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Atin Mukherjee <amukherj@redhat.com>
---
libglusterfs/src/globals.h | 2 +
xlators/mgmt/glusterd/src/glusterd-handler.c | 18 +-
xlators/mgmt/glusterd/src/glusterd-mgmt.c | 240 +++++++++++++++++++++++++--
xlators/mgmt/glusterd/src/glusterd-mgmt.h | 6 +-
xlators/mgmt/glusterd/src/glusterd-op-sm.c | 4 +-
xlators/mgmt/glusterd/src/glusterd-op-sm.h | 6 +
6 files changed, 252 insertions(+), 24 deletions(-)
diff --git a/libglusterfs/src/globals.h b/libglusterfs/src/globals.h
index 1bede2e..d2b0964 100644
--- a/libglusterfs/src/globals.h
+++ b/libglusterfs/src/globals.h
@@ -111,6 +111,8 @@
#define GD_OP_VERSION_3_13_3 31303 /* Op-version for GlusterFS 3.13.3 */
+#define GD_OP_VERSION_6_0 60000 /* Op-version for GlusterFS 6.0 */
+
/* Downstream only change */
#define GD_OP_VERSION_3_11_2 31102 /* Op-version for RHGS 3.3.1-async */
#define GD_OP_VERSION_3_13_3 31303 /* Op-version for RHGS-3.4-Batch Update-1*/
diff --git a/xlators/mgmt/glusterd/src/glusterd-handler.c b/xlators/mgmt/glusterd/src/glusterd-handler.c
index 7486f51..90eaa95 100644
--- a/xlators/mgmt/glusterd/src/glusterd-handler.c
+++ b/xlators/mgmt/glusterd/src/glusterd-handler.c
@@ -28,6 +28,7 @@
#include "glusterd-sm.h"
#include "glusterd-op-sm.h"
#include "glusterd-utils.h"
+#include "glusterd-mgmt.h"
#include "glusterd-server-quorum.h"
#include "glusterd-store.h"
#include "glusterd-locks.h"
@@ -3065,10 +3066,13 @@ __glusterd_handle_cli_profile_volume (rpcsvc_request_t *req)
int32_t op = 0;
char err_str[2048] = {0,};
xlator_t *this = NULL;
+ glusterd_conf_t *conf = NULL;
GF_ASSERT (req);
this = THIS;
GF_ASSERT (this);
+ conf = this->private;
+ GF_VALIDATE_OR_GOTO(this->name, conf, out);
ret = xdr_to_generic (req->msg[0], &cli_req, (xdrproc_t)xdr_gf_cli_req);
if (ret < 0) {
@@ -3109,12 +3113,18 @@ __glusterd_handle_cli_profile_volume (rpcsvc_request_t *req)
goto out;
}
- ret = glusterd_op_begin (req, cli_op, dict, err_str, sizeof (err_str));
+ if (conf->op_version < GD_OP_VERSION_6_0) {
+ gf_msg_debug(this->name, 0, "The cluster is operating at "
+ "version less than %d. Falling back to op-sm "
+ "framework.", GD_OP_VERSION_6_0);
+ ret = glusterd_op_begin(req, cli_op, dict, err_str, sizeof(err_str));
+ glusterd_friend_sm();
+ glusterd_op_sm();
+ } else {
+ ret = glusterd_mgmt_v3_initiate_profile_phases(req, cli_op, dict);
+ }
out:
- glusterd_friend_sm ();
- glusterd_op_sm ();
-
free (cli_req.dict.dict_val);
if (ret) {
diff --git a/xlators/mgmt/glusterd/src/glusterd-mgmt.c b/xlators/mgmt/glusterd/src/glusterd-mgmt.c
index d7da3c1..751d6e4 100644
--- a/xlators/mgmt/glusterd/src/glusterd-mgmt.c
+++ b/xlators/mgmt/glusterd/src/glusterd-mgmt.c
@@ -19,6 +19,7 @@
#include "glusterd-locks.h"
#include "glusterd-mgmt.h"
#include "glusterd-op-sm.h"
+#include "glusterd-server-quorum.h"
#include "glusterd-volgen.h"
#include "glusterd-store.h"
#include "glusterd-snapshot-utils.h"
@@ -213,6 +214,16 @@ gd_mgmt_v3_pre_validate_fn (glusterd_op_t op, dict_t *dict,
}
break;
+ case GD_OP_PROFILE_VOLUME:
+ ret = glusterd_op_stage_stats_volume(dict, op_errstr);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_WARNING, 0,
+ GD_MSG_PRE_VALIDATION_FAIL,
+ "prevalidation failed for profile operation.");
+ goto out;
+ }
+ break;
+
case GD_OP_MAX_OPVERSION:
ret = 0;
break;
@@ -252,6 +263,16 @@ gd_mgmt_v3_brick_op_fn (glusterd_op_t op, dict_t *dict,
}
break;
}
+ case GD_OP_PROFILE_VOLUME:
+ {
+ ret = gd_brick_op_phase(op, rsp_dict, dict, op_errstr);
+ if (ret) {
+ gf_log(this->name, GF_LOG_WARNING, "%s brickop failed",
+ gd_op_list[op]);
+ goto out;
+ }
+ break;
+ }
default:
break;
}
@@ -406,6 +427,17 @@ gd_mgmt_v3_commit_fn (glusterd_op_t op, dict_t *dict,
break;
}
+ case GD_OP_PROFILE_VOLUME:
+ {
+ ret = glusterd_op_stats_volume(dict, op_errstr, rsp_dict);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0,
+ GD_MSG_COMMIT_OP_FAIL, "commit failed "
+ "volume profile operation.");
+ goto out;
+ }
+ break;
+ }
default:
break;
@@ -847,6 +879,7 @@ glusterd_pre_validate_aggr_rsp_dict (glusterd_op_t op,
case GD_OP_DETACH_TIER_STATUS:
case GD_OP_TIER_START_STOP:
case GD_OP_REMOVE_TIER_BRICK:
+ case GD_OP_PROFILE_VOLUME:
break;
case GD_OP_MAX_OPVERSION:
break;
@@ -1039,6 +1072,16 @@ glusterd_mgmt_v3_pre_validate (glusterd_op_t op, dict_t *req_dict,
goto out;
}
+ if (op == GD_OP_PROFILE_VOLUME) {
+ ret = glusterd_validate_quorum(this, op, req_dict, op_errstr);
+ if (ret) {
+ gf_msg (this->name, GF_LOG_ERROR, 0,
+ GD_MSG_SERVER_QUORUM_NOT_MET, "Server quorum "
+ "not met. Rejecting operation.");
+ goto out;
+ }
+ }
+
/* Pre Validation on local node */
ret = gd_mgmt_v3_pre_validate_fn (op, req_dict, op_errstr,
rsp_dict, op_errno);
@@ -1157,6 +1200,7 @@ glusterd_mgmt_v3_build_payload (dict_t **req, char **op_errstr, dict_t *dict,
case GD_OP_REPLACE_BRICK:
case GD_OP_RESET_BRICK:
case GD_OP_ADD_TIER_BRICK:
+ case GD_OP_PROFILE_VOLUME:
{
ret = dict_get_str (dict, "volname", &volname);
if (ret) {
@@ -1309,12 +1353,11 @@ out:
}
int
-glusterd_mgmt_v3_brick_op (glusterd_op_t op, dict_t *req_dict, char **op_errstr,
- uint32_t txn_generation)
+glusterd_mgmt_v3_brick_op (glusterd_op_t op, dict_t *rsp_dict, dict_t *req_dict,
+ char **op_errstr, uint32_t txn_generation)
{
int32_t ret = -1;
int32_t peer_cnt = 0;
- dict_t *rsp_dict = NULL;
glusterd_peerinfo_t *peerinfo = NULL;
struct syncargs args = {0};
uuid_t peer_uuid = {0};
@@ -1329,14 +1372,6 @@ glusterd_mgmt_v3_brick_op (glusterd_op_t op, dict_t *req_dict, char **op_errstr,
GF_ASSERT (req_dict);
GF_ASSERT (op_errstr);
- rsp_dict = dict_new ();
- if (!rsp_dict) {
- gf_msg (this->name, GF_LOG_ERROR, 0,
- GD_MSG_DICT_CREATE_FAIL,
- "Failed to create response dictionary");
- goto out;
- }
-
/* Perform brick op on local node */
ret = gd_mgmt_v3_brick_op_fn (op, req_dict, op_errstr,
rsp_dict);
@@ -1361,11 +1396,8 @@ glusterd_mgmt_v3_brick_op (glusterd_op_t op, dict_t *req_dict, char **op_errstr,
goto out;
}
- dict_unref (rsp_dict);
- rsp_dict = NULL;
-
/* Sending brick op req to other nodes in the cluster */
- gd_syncargs_init (&args, NULL);
+ gd_syncargs_init (&args, rsp_dict);
synctask_barrier_init((&args));
peer_cnt = 0;
@@ -2108,6 +2140,180 @@ out:
}
int32_t
+glusterd_mgmt_v3_initiate_profile_phases (rpcsvc_request_t *req,
+ glusterd_op_t op, dict_t *dict)
+{
+ int32_t ret = -1;
+ int32_t op_ret = -1;
+ dict_t *req_dict = NULL;
+ dict_t *tmp_dict = NULL;
+ glusterd_conf_t *conf = NULL;
+ char *op_errstr = NULL;
+ xlator_t *this = NULL;
+ gf_boolean_t is_acquired = _gf_false;
+ uuid_t *originator_uuid = NULL;
+ uint32_t txn_generation = 0;
+ uint32_t op_errno = 0;
+
+ this = THIS;
+ GF_ASSERT (this);
+ GF_ASSERT (req);
+ GF_ASSERT (dict);
+ conf = this->private;
+ GF_ASSERT (conf);
+
+ /* Save the peer list generation */
+ txn_generation = conf->generation;
+ cmm_smp_rmb ();
+ /* This read memory barrier makes sure that this assignment happens here
+ * only and is not reordered and optimized by either the compiler or the
+ * processor.
+ */
+
+ /* Save the MY_UUID as the originator_uuid. This originator_uuid
+ * will be used by is_origin_glusterd() to determine if a node
+ * is the originator node for a command. */
+ originator_uuid = GF_CALLOC (1, sizeof(uuid_t),
+ gf_common_mt_uuid_t);
+ if (!originator_uuid) {
+ ret = -1;
+ goto out;
+ }
+
+ gf_uuid_copy (*originator_uuid, MY_UUID);
+ ret = dict_set_bin (dict, "originator_uuid",
+ originator_uuid, sizeof (uuid_t));
+ if (ret) {
+ gf_msg (this->name, GF_LOG_ERROR, 0,
+ GD_MSG_DICT_SET_FAILED,
+ "Failed to set originator_uuid.");
+ GF_FREE (originator_uuid);
+ goto out;
+ }
+
+ /* Marking the operation as complete synctasked */
+ ret = dict_set_int32 (dict, "is_synctasked", _gf_true);
+ if (ret) {
+ gf_msg (this->name, GF_LOG_ERROR, 0,
+ GD_MSG_DICT_SET_FAILED,
+ "Failed to set synctasked flag.");
+ goto out;
+ }
+
+ /* Use a copy at local unlock as cli response will be sent before
+ * the unlock and the volname in the dict might be removed */
+ tmp_dict = dict_new();
+ if (!tmp_dict) {
+ gf_msg (this->name, GF_LOG_ERROR, 0,
+ GD_MSG_DICT_CREATE_FAIL, "Unable to create dict");
+ goto out;
+ }
+ dict_copy (dict, tmp_dict);
+
+ /* LOCKDOWN PHASE - Acquire mgmt_v3 locks */
+ ret = glusterd_mgmt_v3_initiate_lockdown (op, dict, &op_errstr,
+ &op_errno, &is_acquired,
+ txn_generation);
+ if (ret) {
+ gf_msg (this->name, GF_LOG_ERROR, 0,
+ GD_MSG_MGMTV3_LOCKDOWN_FAIL,
+ "mgmt_v3 lockdown failed.");
+ goto out;
+ }
+
+ /* BUILD PAYLOAD */
+ ret = glusterd_mgmt_v3_build_payload (&req_dict, &op_errstr, dict, op);
+ if (ret) {
+ gf_msg (this->name, GF_LOG_ERROR, 0,
+ GD_MSG_MGMTV3_PAYLOAD_BUILD_FAIL, LOGSTR_BUILD_PAYLOAD,
+ gd_op_list[op]);
+ if (op_errstr == NULL)
+ gf_asprintf (&op_errstr, OPERRSTR_BUILD_PAYLOAD);
+ goto out;
+ }
+
+ /* PRE-COMMIT VALIDATE PHASE */
+ ret = glusterd_mgmt_v3_pre_validate (op, req_dict, &op_errstr,
+ &op_errno, txn_generation);
+ if (ret) {
+ gf_msg (this->name, GF_LOG_ERROR, 0,
+ GD_MSG_PRE_VALIDATION_FAIL, "Pre Validation Failed");
+ goto out;
+ }
+
+ /* BRICK-OPS */
+ ret = glusterd_mgmt_v3_brick_op(op, dict, req_dict, &op_errstr,
+ txn_generation);
+ if (ret) {
+ gf_log(this->name, GF_LOG_ERROR, "Brick Op Failed");
+ goto out;
+ }
+
+ /* COMMIT OP PHASE */
+ ret = glusterd_mgmt_v3_commit (op, dict, req_dict, &op_errstr,
+ &op_errno, txn_generation);
+ if (ret) {
+ gf_msg (this->name, GF_LOG_ERROR, 0,
+ GD_MSG_COMMIT_OP_FAIL, "Commit Op Failed");
+ goto out;
+ }
+
+ /* POST-COMMIT VALIDATE PHASE */
+ /* As of now, post_validate is not trying to cleanup any failed
+ commands. So as of now, I am sending 0 (op_ret as 0).
+ */
+ ret = glusterd_mgmt_v3_post_validate (op, 0, dict, req_dict, &op_errstr,
+ txn_generation);
+ if (ret) {
+ gf_msg (this->name, GF_LOG_ERROR, 0,
+ GD_MSG_POST_VALIDATION_FAIL, "Post Validation Failed");
+ goto out;
+ }
+
+ ret = 0;
+out:
+ op_ret = ret;
+ /* UNLOCK PHASE FOR PEERS*/
+ (void) glusterd_mgmt_v3_release_peer_locks (op, dict, op_ret,
+ &op_errstr, is_acquired,
+ txn_generation);
+
+ /* LOCAL VOLUME(S) UNLOCK */
+ if (is_acquired) {
+ /* Trying to release multiple mgmt_v3 locks */
+ ret = glusterd_multiple_mgmt_v3_unlock (tmp_dict, MY_UUID);
+ if (ret) {
+ gf_msg (this->name, GF_LOG_ERROR, 0,
+ GD_MSG_MGMTV3_UNLOCK_FAIL,
+ "Failed to release mgmt_v3 locks on localhost");
+ op_ret = ret;
+ }
+ }
+
+ if (op_ret && (op_errno == 0))
+ op_errno = EG_INTRNL;
+
+ if (op != GD_OP_MAX_OPVERSION) {
+ /* SEND CLI RESPONSE */
+ glusterd_op_send_cli_response (op, op_ret, op_errno, req,
+ dict, op_errstr);
+ }
+
+ if (req_dict)
+ dict_unref (req_dict);
+
+ if (tmp_dict)
+ dict_unref (tmp_dict);
+
+ if (op_errstr) {
+ GF_FREE (op_errstr);
+ op_errstr = NULL;
+ }
+
+ return 0;
+}
+
+int32_t
glusterd_mgmt_v3_initiate_all_phases (rpcsvc_request_t *req, glusterd_op_t op,
dict_t *dict)
{
@@ -2465,7 +2671,7 @@ glusterd_mgmt_v3_initiate_snap_phases (rpcsvc_request_t *req, glusterd_op_t op,
goto out;
}
- ret = glusterd_mgmt_v3_brick_op (op, req_dict, &op_errstr,
+ ret = glusterd_mgmt_v3_brick_op (op, dict, req_dict, &op_errstr,
txn_generation);
if (ret) {
gf_msg (this->name, GF_LOG_ERROR, 0,
@@ -2526,7 +2732,7 @@ unbarrier:
goto out;
}
- ret = glusterd_mgmt_v3_brick_op (op, req_dict, &op_errstr,
+ ret = glusterd_mgmt_v3_brick_op (op, dict, req_dict, &op_errstr,
txn_generation);
if (ret) {
diff --git a/xlators/mgmt/glusterd/src/glusterd-mgmt.h b/xlators/mgmt/glusterd/src/glusterd-mgmt.h
index 2215f17..eff070d 100644
--- a/xlators/mgmt/glusterd/src/glusterd-mgmt.h
+++ b/xlators/mgmt/glusterd/src/glusterd-mgmt.h
@@ -37,7 +37,11 @@ glusterd_mgmt_v3_initiate_all_phases (rpcsvc_request_t *req, glusterd_op_t op,
dict_t *dict);
int32_t
-glusterd_mgmt_v3_initiate_snap_phases (rpcsvc_request_t *req, glusterd_op_t op,
+glusterd_mgmt_v3_initiate_profile_phases(rpcsvc_request_t *req,
+ glusterd_op_t op, dict_t *dict);
+
+int32_t
+glusterd_mgmt_v3_initiate_snap_phases(rpcsvc_request_t *req, glusterd_op_t op,
dict_t *dict);
int
diff --git a/xlators/mgmt/glusterd/src/glusterd-op-sm.c b/xlators/mgmt/glusterd/src/glusterd-op-sm.c
index 52a3db0..9f76ab3 100644
--- a/xlators/mgmt/glusterd/src/glusterd-op-sm.c
+++ b/xlators/mgmt/glusterd/src/glusterd-op-sm.c
@@ -2032,7 +2032,7 @@ glusterd_op_stage_status_volume (dict_t *dict, char **op_errstr)
return ret;
}
-static int
+int
glusterd_op_stage_stats_volume (dict_t *dict, char **op_errstr)
{
int ret = -1;
@@ -3322,7 +3322,7 @@ glusterd_remove_profile_volume_options (glusterd_volinfo_t *volinfo)
dict_del (volinfo->dict, fd_stats_key);
}
-static int
+int
glusterd_op_stats_volume (dict_t *dict, char **op_errstr,
dict_t *rsp_dict)
{
diff --git a/xlators/mgmt/glusterd/src/glusterd-op-sm.h b/xlators/mgmt/glusterd/src/glusterd-op-sm.h
index f2aee9c..e64d368 100644
--- a/xlators/mgmt/glusterd/src/glusterd-op-sm.h
+++ b/xlators/mgmt/glusterd/src/glusterd-op-sm.h
@@ -312,4 +312,10 @@ glusterd_dict_set_volid (dict_t *dict, char *volname, char **op_errstr);
int32_t
glusterd_tier_op (xlator_t *this, void *data);
+
+int
+glusterd_op_stats_volume (dict_t *dict, char **op_errstr, dict_t *rsp_dict);
+
+int
+glusterd_op_stage_stats_volume (dict_t *dict, char **op_errstr);
#endif
--
1.8.3.1

72
0472-glusterd-introduce-a-new-op-version-for-rhgs-3.4.3.patch Normal file
View File

@ -0,0 +1,72 @@
From ff97d22a02c02ad5edced2755a799a587b45cb79 Mon Sep 17 00:00:00 2001
From: Sanju Rakonde <srakonde@redhat.com>
Date: Mon, 17 Dec 2018 14:07:01 +0530
Subject: [PATCH 472/493] glusterd: introduce a new op-version for rhgs-3.4.3
This patch introduces a new op-version 31305 for rhgs-3.4.3 and
sets the max op-version to 31305.
For migrating profile commands (commit e68845ff7018e5d81d7979684b18e6eda449b088)
we used GD_OP_VERSION_6_0 in upstream. we are changing
it to GD_OP_VERSION_3_13_5 here.
Label: DOWNSTREAM ONLY
Change-Id: Ie3a05c70eb4e406889c468343f54e999b1218f19
BUG: 1639476
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158795
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Atin Mukherjee <amukherj@redhat.com>
---
libglusterfs/src/globals.h | 5 ++---
xlators/mgmt/glusterd/src/glusterd-handler.c | 4 ++--
2 files changed, 4 insertions(+), 5 deletions(-)
diff --git a/libglusterfs/src/globals.h b/libglusterfs/src/globals.h
index d2b0964..343263c 100644
--- a/libglusterfs/src/globals.h
+++ b/libglusterfs/src/globals.h
@@ -43,7 +43,7 @@
*/
#define GD_OP_VERSION_MIN 1 /* MIN is the fresh start op-version, mostly
should not change */
-#define GD_OP_VERSION_MAX GD_OP_VERSION_3_13_4 /* MAX VERSION is the maximum
+#define GD_OP_VERSION_MAX GD_OP_VERSION_3_13_5 /* MAX VERSION is the maximum
count in VME table, should
keep changing with
introduction of newer
@@ -111,12 +111,11 @@
#define GD_OP_VERSION_3_13_3 31303 /* Op-version for GlusterFS 3.13.3 */
-#define GD_OP_VERSION_6_0 60000 /* Op-version for GlusterFS 6.0 */
-
/* Downstream only change */
#define GD_OP_VERSION_3_11_2 31102 /* Op-version for RHGS 3.3.1-async */
#define GD_OP_VERSION_3_13_3 31303 /* Op-version for RHGS-3.4-Batch Update-1*/
#define GD_OP_VERSION_3_13_4 31304 /* Op-version for RHGS-3.4-Batch Update-2*/
+#define GD_OP_VERSION_3_13_5 31305 /* Op-version for RHGS-3.4-Batch Update-3*/
#include "xlator.h"
diff --git a/xlators/mgmt/glusterd/src/glusterd-handler.c b/xlators/mgmt/glusterd/src/glusterd-handler.c
index 90eaa95..c71bf3c 100644
--- a/xlators/mgmt/glusterd/src/glusterd-handler.c
+++ b/xlators/mgmt/glusterd/src/glusterd-handler.c
@@ -3113,10 +3113,10 @@ __glusterd_handle_cli_profile_volume (rpcsvc_request_t *req)
goto out;
}
- if (conf->op_version < GD_OP_VERSION_6_0) {
+ if (conf->op_version < GD_OP_VERSION_3_13_5) {
gf_msg_debug(this->name, 0, "The cluster is operating at "
"version less than %d. Falling back to op-sm "
- "framework.", GD_OP_VERSION_6_0);
+ "framework.", GD_OP_VERSION_3_13_5);
ret = glusterd_op_begin(req, cli_op, dict, err_str, sizeof(err_str));
glusterd_friend_sm();
glusterd_op_sm();
--
1.8.3.1

64
0473-rpc-bump-up-server.event-threads.patch Normal file
View File

@ -0,0 +1,64 @@
From 63e8fb2b6a3846c3f3a0e1db6275a8c03dbbc6ff Mon Sep 17 00:00:00 2001
From: Milind Changire <mchangir@redhat.com>
Date: Thu, 13 Dec 2018 10:00:45 +0530
Subject: [PATCH 473/493] rpc: bump up server.event-threads
Problem:
A single event-thread causes performance issues in the system.
Solution:
Bump up event-threads to 2 to make the system more performant.
This helps in making the system more responsive and helps avoid the
ping-timer-expiry problem as well. However, setting the event-threads
to 2 is not the only thing required to avoid ping-timer-expiry issues.
NOTE:
NFS xlator option nfs.event-threads does not yet exist here.
mainline:
> Change-Id: Idb0fd49e078db3bd5085dd083b0cdc77b59ddb00
> fixes: bz#1653277
> Signed-off-by: Milind Changire <mchangir@redhat.com>
> Reviewed-on: https://review.gluster.org/c/glusterfs/+/21719
Change-Id: Idb0fd49e078db3bd5085dd083b0cdc77b59ddb00
BUG: 1652537
Signed-off-by: Milind Changire <mchangir@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158482
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Raghavendra Gowdappa <rgowdapp@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
libglusterfs/src/glusterfs.h | 2 +-
xlators/protocol/server/src/server.c | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/libglusterfs/src/glusterfs.h b/libglusterfs/src/glusterfs.h
index 3e2f426..d06d8cf 100644
--- a/libglusterfs/src/glusterfs.h
+++ b/libglusterfs/src/glusterfs.h
@@ -249,7 +249,7 @@ enum gf_internal_fop_indicator {
#define GLUSTERFS_RPC_REPLY_SIZE 24
-#define STARTING_EVENT_THREADS 1
+#define STARTING_EVENT_THREADS 2
#define DEFAULT_VAR_RUN_DIRECTORY DATADIR "/run/gluster"
#define DEFAULT_GLUSTERFSD_MISC_DIRETORY DATADIR "/lib/misc/glusterfsd"
diff --git a/xlators/protocol/server/src/server.c b/xlators/protocol/server/src/server.c
index 6f510ea..1046152 100644
--- a/xlators/protocol/server/src/server.c
+++ b/xlators/protocol/server/src/server.c
@@ -1932,7 +1932,7 @@ struct volume_options options[] = {
.type = GF_OPTION_TYPE_INT,
.min = 1,
.max = 1024,
- .default_value = "1",
+ .default_value = "2",
.description = "Specifies the number of event threads to execute "
"in parallel. Larger values would help process"
" responses faster, depending on available processing"
--
1.8.3.1

45
0474-afr-open_ftruncate_cbk-should-read-fd-from-local-con.patch Normal file
View File

@ -0,0 +1,45 @@
From cdedd41ab825bfe59e8d1739fdea625a51f659f9 Mon Sep 17 00:00:00 2001
From: Soumya Koduri <skoduri@redhat.com>
Date: Fri, 9 Nov 2018 02:29:52 -0500
Subject: [PATCH 474/493] afr: open_ftruncate_cbk should read fd from
local->cont.open struct
afr_open stores the fd as part of its local->cont.open struct
but when it calls ftruncate (if open flags contain O_TRUNC), the
corresponding cbk function (afr_ open_ftruncate_cbk) is
incorrectly referencing uninitialized local->fd. This patch fixes
the same.
Upstream reference:
Change-Id: Icbdedbd1b8cfea11d8f41b6e5c4cb4b44d989aba
> updates: bz#1648687
> review-url: https://review.gluster.org/#/c/glusterfs/+/21617/
BUG: 1655578
updates: bz#1655578
Signed-off-by: Soumya Koduri <skoduri@redhat.com>
Change-Id: I9c26eadd811fdd32630227f3130dec28e4b6972b
Reviewed-on: https://code.engineering.redhat.com/gerrit/158799
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>
---
xlators/cluster/afr/src/afr-open.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/xlators/cluster/afr/src/afr-open.c b/xlators/cluster/afr/src/afr-open.c
index 6c625cc..d820462 100644
--- a/xlators/cluster/afr/src/afr-open.c
+++ b/xlators/cluster/afr/src/afr-open.c
@@ -62,7 +62,7 @@ afr_open_ftruncate_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
afr_local_t * local = frame->local;
AFR_STACK_UNWIND (open, frame, local->op_ret, local->op_errno,
- local->fd, xdata);
+ local->cont.open.fd, xdata);
return 0;
}
--
1.8.3.1

175
0475-glusterd-perform-store-operation-in-cleanup-lock.patch Normal file
View File

@ -0,0 +1,175 @@
From db15e8fe12b7148b2da975d915573cb24c4ee1c9 Mon Sep 17 00:00:00 2001
From: Atin Mukherjee <amukherj@redhat.com>
Date: Thu, 22 Nov 2018 09:58:52 +0530
Subject: [PATCH 475/493] glusterd: perform store operation in cleanup lock
All glusterd store operation and cleanup thread should work under a
critical section to avoid any partial store write.
> Change-Id: I4f12e738f597a1f925c87ea2f42565dcf9ecdb9d
> Fixes: bz#1652430
> Signed-off-by: Atin Mukherjee <amukherj@redhat.com>
upstream patch: https://review.gluster.org/#/c/glusterfs/+/21702/
Change-Id: I4f12e738f597a1f925c87ea2f42565dcf9ecdb9d
BUG: 1654161
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158804
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Atin Mukherjee <amukherj@redhat.com>
---
glusterfsd/src/glusterfsd.c | 73 ++++++++++++++++--------------
libglusterfs/src/glusterfs.h | 1 +
xlators/mgmt/glusterd/src/glusterd-store.c | 10 +++-
3 files changed, 49 insertions(+), 35 deletions(-)
diff --git a/glusterfsd/src/glusterfsd.c b/glusterfsd/src/glusterfsd.c
index 03bca24..57effbd 100644
--- a/glusterfsd/src/glusterfsd.c
+++ b/glusterfsd/src/glusterfsd.c
@@ -1395,43 +1395,46 @@ cleanup_and_exit (int signum)
if (ctx->cleanup_started)
return;
- ctx->cleanup_started = 1;
+ pthread_mutex_lock(&ctx->cleanup_lock);
+ {
+ ctx->cleanup_started = 1;
- /* signout should be sent to all the bricks in case brick mux is enabled
- * and multiple brick instances are attached to this process
- */
- if (ctx->active) {
- top = ctx->active->first;
- for (trav_p = &top->children; *trav_p;
- trav_p = &(*trav_p)->next) {
- victim = (*trav_p)->xlator;
- glusterfs_mgmt_pmap_signout (ctx, victim->name);
+ /* signout should be sent to all the bricks in case brick mux
+ * is enabled and multiple brick instances are attached to this
+ * process
+ */
+ if (ctx->active) {
+ top = ctx->active->first;
+ for (trav_p = &top->children; *trav_p;
+ trav_p = &(*trav_p)->next) {
+ victim = (*trav_p)->xlator;
+ glusterfs_mgmt_pmap_signout (ctx, victim->name);
+ }
+ } else {
+ glusterfs_mgmt_pmap_signout (ctx, NULL);
}
- } else {
- glusterfs_mgmt_pmap_signout (ctx, NULL);
- }
- /* below part is a racy code where the rpcsvc object is freed.
- * But in another thread (epoll thread), upon poll error in the
- * socket the transports are cleaned up where again rpcsvc object
- * is accessed (which is already freed by the below function).
- * Since the process is about to be killed dont execute the function
- * below.
- */
- /* if (ctx->listener) { */
- /* (void) glusterfs_listener_stop (ctx); */
- /* } */
+ /* below part is a racy code where the rpcsvc object is freed.
+ * But in another thread (epoll thread), upon poll error in the
+ * socket the transports are cleaned up where again rpcsvc object
+ * is accessed (which is already freed by the below function).
+ * Since the process is about to be killed dont execute the
+ * function below.
+ */
+ /* if (ctx->listener) { */
+ /* (void) glusterfs_listener_stop (ctx); */
+ /* } */
- /* Call fini() of FUSE xlator first:
- * so there are no more requests coming and
- * 'umount' of mount point is done properly */
- trav = ctx->master;
- if (trav && trav->fini) {
- THIS = trav;
- trav->fini (trav);
- }
+ /* Call fini() of FUSE xlator first:
+ * so there are no more requests coming and
+ * 'umount' of mount point is done properly */
+ trav = ctx->master;
+ if (trav && trav->fini) {
+ THIS = trav;
+ trav->fini (trav);
+ }
- glusterfs_pidfile_cleanup (ctx);
+ glusterfs_pidfile_cleanup (ctx);
#if 0
/* TODO: Properly do cleanup_and_exit(), with synchronization */
@@ -1442,8 +1445,9 @@ cleanup_and_exit (int signum)
}
#endif
- trav = NULL;
-
+ trav = NULL;
+ }
+ pthread_mutex_unlock(&ctx->cleanup_lock);
/* NOTE: Only the least significant 8 bits i.e (signum & 255)
will be available to parent process on calling exit() */
exit(abs(signum));
@@ -1598,6 +1602,7 @@ glusterfs_ctx_defaults_init (glusterfs_ctx_t *ctx)
goto out;
pthread_mutex_init (&ctx->notify_lock, NULL);
+ pthread_mutex_init(&ctx->cleanup_lock, NULL);
pthread_cond_init (&ctx->notify_cond, NULL);
ctx->clienttable = gf_clienttable_alloc();
diff --git a/libglusterfs/src/glusterfs.h b/libglusterfs/src/glusterfs.h
index d06d8cf..c12e94e 100644
--- a/libglusterfs/src/glusterfs.h
+++ b/libglusterfs/src/glusterfs.h
@@ -574,6 +574,7 @@ struct _glusterfs_ctx {
char btbuf[GF_BACKTRACE_LEN];
pthread_mutex_t notify_lock;
+ pthread_mutex_t cleanup_lock;
pthread_cond_t notify_cond;
int notifying;
diff --git a/xlators/mgmt/glusterd/src/glusterd-store.c b/xlators/mgmt/glusterd/src/glusterd-store.c
index f276fef..b3c4d9a 100644
--- a/xlators/mgmt/glusterd/src/glusterd-store.c
+++ b/xlators/mgmt/glusterd/src/glusterd-store.c
@@ -1792,10 +1792,17 @@ out:
int32_t
glusterd_store_volinfo (glusterd_volinfo_t *volinfo, glusterd_volinfo_ver_ac_t ac)
{
- int32_t ret = -1;
+ int32_t ret = -1;
+ glusterfs_ctx_t *ctx = NULL;
+ xlator_t *this = NULL;
+ this = THIS;
+ GF_ASSERT(this);
+ ctx = this->ctx;
+ GF_ASSERT(ctx);
GF_ASSERT (volinfo);
+ pthread_mutex_lock(&ctx->cleanup_lock);
pthread_mutex_lock(&volinfo->store_volinfo_lock);
{
glusterd_perform_volinfo_version_action(volinfo, ac);
@@ -1837,6 +1844,7 @@ glusterd_store_volinfo (glusterd_volinfo_t *volinfo, glusterd_volinfo_ver_ac_t a
}
unlock:
pthread_mutex_unlock(&volinfo->store_volinfo_lock);
+ pthread_mutex_unlock(&ctx->cleanup_lock);
if (ret)
glusterd_store_volume_cleanup_tmp(volinfo);
--
1.8.3.1

610
0476-afr-add-checks-for-allowing-lookups.patch Normal file
View File

@ -0,0 +1,610 @@
From e3f6fc1ccff95145b94aea9405cd136ada9000bc Mon Sep 17 00:00:00 2001
From: Ravishankar N <ravishankar@redhat.com>
Date: Wed, 16 Aug 2017 18:01:17 +0530
Subject: [PATCH 476/493] afr: add checks for allowing lookups
Patch in upstream master: https://review.gluster.org/#/c/glusterfs/+/17673/
Problem:
In an arbiter volume, lookup was being served from one of the sink
bricks (source brick was down). shard uses the iatt values from lookup cbk
to calculate the size and block count, which in this case were incorrect
values. shard_local_t->last_block was thus initialised to -1, resulting
in an infinite while loop in shard_common_resolve_shards().
Fix:
Use client quorum logic to allow or fail the lookups from afr if there
are no readable subvolumes. So in replica-3 or arbiter vols, if there is
no good copy or if quorum is not met, fail lookup with ENOTCONN.
With this fix, we are also removing support for quorum-reads xlator
option. So if quorum is not met, neither read nor write txns are allowed
and we fail the fop with ENOTCONN.
Change-Id: Ic65c00c24f77ece007328b421494eee62a505fa0
BUG: 1362129
Signed-off-by: Ravishankar N <ravishankar@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158650
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>
---
tests/basic/afr/quorum.t | 23 ----
tests/bugs/replicate/bug-977797.t | 2 +
xlators/cluster/afr/src/afr-common.c | 244 +++++++++++++++++++++------------
xlators/cluster/afr/src/afr-read-txn.c | 3 +-
xlators/cluster/afr/src/afr.c | 7 +-
xlators/cluster/afr/src/afr.h | 1 -
6 files changed, 164 insertions(+), 116 deletions(-)
diff --git a/tests/basic/afr/quorum.t b/tests/basic/afr/quorum.t
index 252e254..58116ba 100644
--- a/tests/basic/afr/quorum.t
+++ b/tests/basic/afr/quorum.t
@@ -31,11 +31,7 @@ TEST $CLI volume set $V0 cluster.quorum-count 2
TEST test_write
TEST kill_brick $V0 $H0 $B0/${V0}1
TEST ! test_write
-EXPECT "abc" cat $M0/b
-TEST $CLI volume set $V0 cluster.quorum-reads on
-EXPECT_WITHIN $CONFIG_UPDATE_TIMEOUT "1" mount_get_option_value $M0 $V0-replicate-0 quorum-reads
TEST ! cat $M0/b
-TEST $CLI volume reset $V0 cluster.quorum-reads
TEST $CLI volume set $V0 cluster.quorum-type auto
EXPECT auto volume_option $V0 cluster.quorum-type
@@ -44,11 +40,7 @@ EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 0
TEST test_write
TEST kill_brick $V0 $H0 $B0/${V0}1
TEST ! test_write
-EXPECT "abc" cat $M0/b
-TEST $CLI volume set $V0 cluster.quorum-reads on
-EXPECT_WITHIN $CONFIG_UPDATE_TIMEOUT "1" mount_get_option_value $M0 $V0-replicate-0 quorum-reads
TEST ! cat $M0/b
-TEST $CLI volume reset $V0 cluster.quorum-reads
TEST $CLI volume set $V0 cluster.quorum-type none
EXPECT none volume_option $V0 cluster.quorum-type
@@ -57,11 +49,6 @@ TEST test_write
TEST $CLI volume reset $V0 cluster.quorum-type
TEST test_write
EXPECT "abc" cat $M0/b
-TEST $CLI volume set $V0 cluster.quorum-reads on
-EXPECT_WITHIN $CONFIG_UPDATE_TIMEOUT "1" mount_get_option_value $M0 $V0-replicate-0 quorum-reads
-EXPECT "abc" cat $M0/b
-TEST $CLI volume reset $V0 cluster.quorum-reads
-
cleanup;
TEST glusterd;
@@ -86,24 +73,14 @@ TEST $CLI volume set $V0 cluster.quorum-count 3
TEST test_write
TEST kill_brick $V0 $H0 $B0/${V0}1
TEST ! test_write
-EXPECT "abc" cat $M0/b
-TEST $CLI volume set $V0 cluster.quorum-reads on
-EXPECT_WITHIN $CONFIG_UPDATE_TIMEOUT "1" mount_get_option_value $M0 $V0-replicate-0 quorum-reads
TEST ! cat $M0/b
-TEST $CLI volume reset $V0 cluster.quorum-reads
-
TEST $CLI volume set $V0 cluster.quorum-type auto
EXPECT auto volume_option $V0 cluster.quorum-type
TEST test_write
TEST kill_brick $V0 $H0 $B0/${V0}3
TEST ! test_write
-EXPECT "abc" cat $M0/b
-TEST $CLI volume set $V0 cluster.quorum-reads on
-EXPECT_WITHIN $CONFIG_UPDATE_TIMEOUT "1" mount_get_option_value $M0 $V0-replicate-0 quorum-reads
TEST ! cat $M0/b
-TEST $CLI volume reset $V0 cluster.quorum-reads
-
TEST $CLI volume set $V0 cluster.quorum-type none
EXPECT none volume_option $V0 cluster.quorum-type
diff --git a/tests/bugs/replicate/bug-977797.t b/tests/bugs/replicate/bug-977797.t
index ea9a98a..fee8205 100755
--- a/tests/bugs/replicate/bug-977797.t
+++ b/tests/bugs/replicate/bug-977797.t
@@ -53,6 +53,8 @@ TEST chmod 757 $M0/a/file
TEST $CLI volume start $V0 force
EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status $V0 1;
+#Trigger entry heal of $M0/a
+getfattr -n user.nosuchattr $M0/a
dd if=$M0/a/file of=/dev/null bs=1024k
#read fails, but heal is triggered.
TEST [ $? -ne 0 ]
diff --git a/xlators/cluster/afr/src/afr-common.c b/xlators/cluster/afr/src/afr-common.c
index 10d9620..231de9d 100644
--- a/xlators/cluster/afr/src/afr-common.c
+++ b/xlators/cluster/afr/src/afr-common.c
@@ -47,9 +47,7 @@
int32_t
afr_quorum_errno (afr_private_t *priv)
{
- if (priv->quorum_reads)
- return ENOTCONN;
- return EROFS;
+ return ENOTCONN;
}
int
@@ -1154,8 +1152,6 @@ afr_replies_interpret (call_frame_t *frame, xlator_t *this, inode_t *inode,
return ret;
}
-
-
int
afr_refresh_selfheal_done (int ret, call_frame_t *heal, void *opaque)
{
@@ -1726,6 +1722,29 @@ afr_inode_read_subvol_type_get (inode_t *inode, xlator_t *this,
return ret;
}
+void
+afr_readables_intersect_get (inode_t *inode, xlator_t *this, int *event,
+ unsigned char *intersection)
+{
+ afr_private_t *priv = NULL;
+ unsigned char *data_readable = NULL;
+ unsigned char *metadata_readable = NULL;
+ unsigned char *intersect = NULL;
+
+ priv = this->private;
+ data_readable = alloca0 (priv->child_count);
+ metadata_readable = alloca0 (priv->child_count);
+ intersect = alloca0 (priv->child_count);
+
+ afr_inode_read_subvol_get (inode, this, data_readable,
+ metadata_readable, event);
+
+ AFR_INTERSECT (intersect, data_readable, metadata_readable,
+ priv->child_count);
+ if (intersection)
+ memcpy (intersection, intersect,
+ sizeof (*intersection) * priv->child_count);
+}
int
afr_read_subvol_get (inode_t *inode, xlator_t *this, int *subvol_p,
@@ -1734,8 +1753,6 @@ afr_read_subvol_get (inode_t *inode, xlator_t *this, int *subvol_p,
afr_read_subvol_args_t *args)
{
afr_private_t *priv = NULL;
- unsigned char *data_readable = NULL;
- unsigned char *metadata_readable = NULL;
unsigned char *readable = NULL;
unsigned char *intersection = NULL;
int subvol = -1;
@@ -1744,17 +1761,11 @@ afr_read_subvol_get (inode_t *inode, xlator_t *this, int *subvol_p,
priv = this->private;
readable = alloca0 (priv->child_count);
- data_readable = alloca0 (priv->child_count);
- metadata_readable = alloca0 (priv->child_count);
intersection = alloca0 (priv->child_count);
afr_inode_read_subvol_type_get (inode, this, readable, &event, type);
- afr_inode_read_subvol_get (inode, this, data_readable, metadata_readable,
- &event);
-
- AFR_INTERSECT (intersection, data_readable, metadata_readable,
- priv->child_count);
+ afr_readables_intersect_get (inode, this, &event, intersection);
if (AFR_COUNT (intersection, priv->child_count) > 0)
subvol = afr_read_subvol_select_by_policy (inode, this,
@@ -2188,18 +2199,28 @@ afr_get_parent_read_subvol (xlator_t *this, inode_t *parent,
int
afr_read_subvol_decide (inode_t *inode, xlator_t *this,
- afr_read_subvol_args_t *args)
+ afr_read_subvol_args_t *args, unsigned char *readable)
{
- int data_subvol = -1;
- int mdata_subvol = -1;
+ int event = 0;
+ afr_private_t *priv = NULL;
+ unsigned char *intersection = NULL;
+
+ priv = this->private;
+ intersection = alloca0 (priv->child_count);
+
+ afr_readables_intersect_get (inode, this, &event, intersection);
- data_subvol = afr_data_subvol_get (inode, this, NULL, NULL, NULL, args);
- mdata_subvol = afr_metadata_subvol_get (inode, this,
- NULL, NULL, NULL, args);
- if (data_subvol == -1 || mdata_subvol == -1)
+ if (AFR_COUNT (intersection, priv->child_count) <= 0) {
+ /* TODO: If we have one brick with valid data_readable and
+ * another with metadata_readable, try to send an iatt with
+ * valid bits from both.*/
return -1;
+ }
- return data_subvol;
+ memcpy (readable, intersection, sizeof (*readable) * priv->child_count);
+
+ return afr_read_subvol_select_by_policy (inode, this, intersection,
+ args);
}
static inline int
@@ -2216,7 +2237,49 @@ afr_first_up_child (call_frame_t *frame, xlator_t *this)
if (local->replies[i].valid &&
local->replies[i].op_ret == 0)
return i;
- return 0;
+ return -1;
+}
+
+static void
+afr_attempt_readsubvol_set (call_frame_t *frame, xlator_t *this,
+ unsigned char *success_replies,
+ unsigned char *data_readable, int *read_subvol)
+{
+ afr_private_t *priv = NULL;
+ afr_local_t *local = NULL;
+ int spb_choice = -1;
+ int child_count = -1;
+
+ if (*read_subvol != -1)
+ return;
+
+ priv = this->private;
+ local = frame->local;
+ child_count = priv->child_count;
+
+ afr_inode_split_brain_choice_get (local->inode, this,
+ &spb_choice);
+ if ((spb_choice >= 0) &&
+ (AFR_COUNT(success_replies, child_count) == child_count)) {
+ *read_subvol = spb_choice;
+ } else if (!priv->quorum_count) {
+ *read_subvol = afr_first_up_child (frame, this);
+ } else if (priv->quorum_count &&
+ afr_has_quorum (data_readable, this)) {
+ /* read_subvol is guaranteed to be valid if we hit this path. */
+ *read_subvol = afr_first_up_child (frame, this);
+ } else {
+ /* If quorum is enabled and we do not have a
+ readable yet, it means all good copies are down.
+ */
+ local->op_ret = -1;
+ local->op_errno = ENOTCONN;
+ gf_msg (this->name, GF_LOG_WARNING, 0,
+ AFR_MSG_READ_SUBVOL_ERROR, "no read "
+ "subvols for %s", local->loc.path);
+ }
+ if (*read_subvol >= 0)
+ dict_del (local->replies[*read_subvol].xdata, GF_CONTENT_KEY);
}
static void
@@ -2230,13 +2293,13 @@ afr_lookup_done (call_frame_t *frame, xlator_t *this)
int par_read_subvol = 0;
int ret = -1;
unsigned char *readable = NULL;
+ unsigned char *success_replies = NULL;
int event = 0;
struct afr_reply *replies = NULL;
uuid_t read_gfid = {0, };
gf_boolean_t locked_entry = _gf_false;
gf_boolean_t can_interpret = _gf_true;
inode_t *parent = NULL;
- int spb_choice = -1;
ia_type_t ia_type = IA_INVAL;
afr_read_subvol_args_t args = {0,};
char *gfid_heal_msg = NULL;
@@ -2250,11 +2313,12 @@ afr_lookup_done (call_frame_t *frame, xlator_t *this)
this);
readable = alloca0 (priv->child_count);
+ success_replies = alloca0 (priv->child_count);
afr_inode_read_subvol_get (parent, this, readable, NULL, &event);
+ par_read_subvol = afr_get_parent_read_subvol (this, parent, replies,
+ readable);
- afr_inode_split_brain_choice_get (local->inode, this,
- &spb_choice);
/* First, check if we have a gfid-change from somewhere,
If so, propagate that so that a fresh lookup can be
issued
@@ -2262,13 +2326,17 @@ afr_lookup_done (call_frame_t *frame, xlator_t *this)
if (local->cont.lookup.needs_fresh_lookup) {
local->op_ret = -1;
local->op_errno = ESTALE;
- goto unwind;
+ goto error;
}
op_errno = afr_final_errno (frame->local, this->private);
local->op_errno = op_errno;
read_subvol = -1;
+ for (i = 0; i < priv->child_count; i++)
+ if (replies[i].valid && replies[i].op_ret == 0)
+ success_replies[i] = 1;
+
for (i = 0; i < priv->child_count; i++) {
if (!replies[i].valid)
continue;
@@ -2277,9 +2345,9 @@ afr_lookup_done (call_frame_t *frame, xlator_t *this)
replies[i].op_errno == ENOENT) {
/* Second, check entry is still
"underway" in creation */
- local->op_ret = -1;
- local->op_errno = ENOENT;
- goto unwind;
+ local->op_ret = -1;
+ local->op_errno = ENOENT;
+ goto error;
}
if (replies[i].op_ret == -1)
@@ -2293,8 +2361,8 @@ afr_lookup_done (call_frame_t *frame, xlator_t *this)
}
}
- if (read_subvol == -1)
- goto unwind;
+ if (read_subvol == -1)
+ goto error;
/* We now have a read_subvol, which is readable[] (if there
were any). Next we look for GFID mismatches. We don't
consider a GFID mismatch as an error if read_subvol is
@@ -2318,58 +2386,61 @@ afr_lookup_done (call_frame_t *frame, xlator_t *this)
if (readable[read_subvol] && !readable[i])
continue;
+ /* If we were called from glfsheal and there is still a gfid
+ * mismatch, succeed the lookup and let glfsheal print the
+ * response via gfid-heal-msg.*/
+ if (!dict_get_str (local->xattr_req, "gfid-heal-msg",
+ &gfid_heal_msg))
+ goto cant_interpret;
+
/* LOG ERROR */
local->op_ret = -1;
local->op_errno = EIO;
- goto unwind;
+ goto error;
}
/* Forth, for the finalized GFID, pick the best subvolume
to return stats from.
*/
+ read_subvol = -1;
+ memset (readable, 0, sizeof (*readable) * priv->child_count);
if (can_interpret) {
+ if (!afr_has_quorum (success_replies, this))
+ goto cant_interpret;
/* It is safe to call afr_replies_interpret() because we have
a response from all the UP subvolumes and all of them resolved
to the same GFID
*/
gf_uuid_copy (args.gfid, read_gfid);
args.ia_type = ia_type;
- if (afr_replies_interpret (frame, this, local->inode, NULL)) {
- read_subvol = afr_read_subvol_decide (local->inode,
- this, &args);
+ ret = afr_replies_interpret (frame, this, local->inode, NULL);
+ read_subvol = afr_read_subvol_decide (local->inode, this, &args,
+ readable);
+ if (read_subvol == -1)
+ goto cant_interpret;
+ if (ret) {
afr_inode_event_gen_reset (local->inode, this);
- goto cant_interpret;
- } else {
- read_subvol = afr_data_subvol_get (local->inode, this,
- NULL, NULL, NULL, &args);
- }
+ dict_del (local->replies[read_subvol].xdata,
+ GF_CONTENT_KEY);
+ }
} else {
cant_interpret:
+ afr_attempt_readsubvol_set (frame, this, success_replies,
+ readable, &read_subvol);
if (read_subvol == -1) {
- if (spb_choice >= 0)
- read_subvol = spb_choice;
- else
- read_subvol = afr_first_up_child (frame, this);
+ goto error;
}
- dict_del (replies[read_subvol].xdata, GF_CONTENT_KEY);
}
afr_handle_quota_size (frame, this);
-unwind:
afr_set_need_heal (this, local);
- if (read_subvol == -1) {
- if (spb_choice >= 0)
- read_subvol = spb_choice;
- else
- read_subvol = afr_first_up_child (frame, this);
-
- }
- par_read_subvol = afr_get_parent_read_subvol (this, parent, replies,
- readable);
if (AFR_IS_ARBITER_BRICK (priv, read_subvol) && local->op_ret == 0) {
- local->op_ret = -1;
- local->op_errno = ENOTCONN;
+ local->op_ret = -1;
+ local->op_errno = ENOTCONN;
+ gf_msg_debug(this->name, 0, "Arbiter cannot be a read subvol "
+ "for %s", local->loc.path);
+ goto error;
}
ret = dict_get_str (local->xattr_req, "gfid-heal-msg", &gfid_heal_msg);
@@ -2389,6 +2460,11 @@ unwind:
local->inode, &local->replies[read_subvol].poststat,
local->replies[read_subvol].xdata,
&local->replies[par_read_subvol].postparent);
+ return;
+
+error:
+ AFR_STACK_UNWIND (lookup, frame, local->op_ret, local->op_errno, NULL,
+ NULL, NULL, NULL);
}
/*
@@ -2904,55 +2980,54 @@ afr_discover_done (call_frame_t *frame, xlator_t *this)
afr_local_t *local = NULL;
int i = -1;
int op_errno = 0;
- int spb_choice = -1;
int read_subvol = -1;
+ unsigned char *data_readable = NULL;
+ unsigned char *success_replies = NULL;
priv = this->private;
local = frame->local;
-
- afr_inode_split_brain_choice_get (local->inode, this,
- &spb_choice);
+ data_readable = alloca0 (priv->child_count);
+ success_replies = alloca0 (priv->child_count);
for (i = 0; i < priv->child_count; i++) {
if (!local->replies[i].valid)
continue;
- if (local->replies[i].op_ret == 0)
+ if (local->replies[i].op_ret == 0) {
+ success_replies[i] = 1;
local->op_ret = 0;
+ }
}
op_errno = afr_final_errno (frame->local, this->private);
if (local->op_ret < 0) {
- local->op_errno = op_errno;
- local->op_ret = -1;
- goto unwind;
+ AFR_STACK_UNWIND (lookup, frame, -1, op_errno, NULL, NULL,
+ NULL, NULL);
+ return;
}
- afr_replies_interpret (frame, this, local->inode, NULL);
+ if (!afr_has_quorum (success_replies, this))
+ goto unwind;
- read_subvol = afr_read_subvol_decide (local->inode, this, NULL);
- if (read_subvol == -1) {
- gf_msg (this->name, GF_LOG_WARNING, 0,
- AFR_MSG_READ_SUBVOL_ERROR, "no read subvols for %s",
- local->loc.path);
+ afr_replies_interpret (frame, this, local->inode, NULL);
- if (spb_choice >= 0) {
- read_subvol = spb_choice;
- } else {
- read_subvol = afr_first_up_child (frame, this);
- }
- }
+ read_subvol = afr_read_subvol_decide (local->inode, this, NULL,
+ data_readable);
unwind:
+ afr_attempt_readsubvol_set (frame, this, success_replies, data_readable,
+ &read_subvol);
if (read_subvol == -1) {
- if (spb_choice >= 0)
- read_subvol = spb_choice;
- else
- read_subvol = afr_first_up_child (frame, this);
+ AFR_STACK_UNWIND (lookup, frame, local->op_ret, local->op_errno,
+ NULL, NULL, NULL, NULL);
+ return;
}
+
if (AFR_IS_ARBITER_BRICK (priv, read_subvol) && local->op_ret == 0) {
- local->op_ret = -1;
- local->op_errno = ENOTCONN;
+ local->op_ret = -1;
+ local->op_errno = ENOTCONN;
+ gf_msg_debug (this->name, 0, "Arbiter cannot be a read subvol "
+ "for %s", local->loc.path);
}
AFR_STACK_UNWIND (lookup, frame, local->op_ret, local->op_errno,
@@ -4646,7 +4721,6 @@ afr_priv_dump (xlator_t *this)
gf_proc_dump_write("read_child", "%d", priv->read_child);
gf_proc_dump_write("favorite_child", "%d", priv->favorite_child);
gf_proc_dump_write("wait_count", "%u", priv->wait_count);
- gf_proc_dump_write("quorum-reads", "%d", priv->quorum_reads);
gf_proc_dump_write("heal-wait-queue-length", "%d",
priv->heal_wait_qlen);
gf_proc_dump_write("heal-waiters", "%d", priv->heal_waiters);
diff --git a/xlators/cluster/afr/src/afr-read-txn.c b/xlators/cluster/afr/src/afr-read-txn.c
index 50e8040..f6c491b 100644
--- a/xlators/cluster/afr/src/afr-read-txn.c
+++ b/xlators/cluster/afr/src/afr-read-txn.c
@@ -193,8 +193,7 @@ afr_read_txn (call_frame_t *frame, xlator_t *this, inode_t *inode,
local->inode = inode_ref (inode);
local->is_read_txn = _gf_true;
- if (priv->quorum_reads &&
- priv->quorum_count && !afr_has_quorum (priv->child_up, this)) {
+ if (priv->quorum_count && !afr_has_quorum (local->child_up, this)) {
local->op_ret = -1;
local->op_errno = ENOTCONN;
read_subvol = -1;
diff --git a/xlators/cluster/afr/src/afr.c b/xlators/cluster/afr/src/afr.c
index 0122b7f..1b738c0 100644
--- a/xlators/cluster/afr/src/afr.c
+++ b/xlators/cluster/afr/src/afr.c
@@ -267,8 +267,6 @@ reconfigure (xlator_t *this, dict_t *options)
GF_OPTION_RECONF ("heal-timeout", priv->shd.timeout, options,
int32, out);
- GF_OPTION_RECONF ("quorum-reads", priv->quorum_reads, options,
- bool, out);
GF_OPTION_RECONF ("consistent-metadata", priv->consistent_metadata,
options, bool, out);
@@ -531,7 +529,6 @@ init (xlator_t *this)
GF_OPTION_INIT ("iam-self-heal-daemon", priv->shd.iamshd, bool, out);
GF_OPTION_INIT ("heal-timeout", priv->shd.timeout, int32, out);
- GF_OPTION_INIT ("quorum-reads", priv->quorum_reads, bool, out);
GF_OPTION_INIT ("consistent-metadata", priv->consistent_metadata, bool,
out);
GF_OPTION_INIT ("consistent-io", priv->consistent_io, bool, out);
@@ -965,8 +962,8 @@ struct volume_options options[] = {
{ .key = {"quorum-reads"},
.type = GF_OPTION_TYPE_BOOL,
.default_value = "no",
- .description = "If quorum-reads is \"true\" only allow reads if "
- "quorum is met when quorum is enabled.",
+ .description = "This option has been removed. Reads are not allowed "
+ "if quorum is not met.",
},
{ .key = {"node-uuid"},
.type = GF_OPTION_TYPE_STR,
diff --git a/xlators/cluster/afr/src/afr.h b/xlators/cluster/afr/src/afr.h
index af9dbc8..7010e9b 100644
--- a/xlators/cluster/afr/src/afr.h
+++ b/xlators/cluster/afr/src/afr.h
@@ -131,7 +131,6 @@ typedef struct _afr_private {
gf_boolean_t pre_op_compat; /* on/off */
uint32_t post_op_delay_secs;
unsigned int quorum_count;
- gf_boolean_t quorum_reads;
char vol_uuid[UUID_SIZE + 1];
int32_t *last_event;
--
1.8.3.1

1808
0477-glusterd-perform-rcu_read_lock-unlock-under-cleanup_.patch Normal file
View File

File diff suppressed because it is too large Load Diff

289
0478-libglusterfs-fix-memory-corruption-caused-by-per-thr.patch Normal file
View File

@ -0,0 +1,289 @@
From a5471a84069631ab0d0605cf7b68f16285f5079f Mon Sep 17 00:00:00 2001
From: Xavi Hernandez <xhernandez@redhat.com>
Date: Fri, 14 Dec 2018 11:26:36 +0100
Subject: [PATCH 478/493] libglusterfs: fix memory corruption caused by
per-thread mem pools
There was a race in the per-thread memory pool management that could lead
to memory corruption. The race appeared when the following sequence of
events happened:
1. Thread T1 allocated a memory object O1 from its own private pool P1
2. T1 terminates and P1 is marked to be destroyed
3. The mem-sweeper thread is woken up and scans all private pools
4. It detects that P1 needs to be destroyed and starts releasing the
objects from hot and cold lists.
5. Thread T2 releases O1
6. O1 is added to the hot list of P1
The problem happens because steps 4 and 6 are protected by diferent locks,
so they can run concurrently. This means that both T1 and T2 are modifying
the same list at the same time, potentially causing corruption.
This patch fixes the problem using the following approach:
1. When an object is released, it's only returned to the hot list of the
corresponding memory pool if it's not marked to be destroyed. Otherwise
the memory is released to the system.
2. Object release and mem-sweeper thread synchronize access to the deletion
mark of the memory pool to prevent simultaneous access to the list.
Some other minor adjustments are made to reduce the lengths of the locked
regions.
This patch is not 100% identical to upstream version because changes
coming from https://github.com/gluster/glusterfs/issues/307 are not
backported.
Upstream patch: https://review.gluster.org/c/glusterfs/+/21583
> Fixes: bz#1651165
> Change-Id: I63be3893f92096e57f54a6150e0461340084ddde
> Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
Upstream patch: https://review.gluster.org/c/glusterfs/+/21727
> Change-Id: Idbf23bda7f9228d60c644a1bea4b6c2cfc582090
> updates: bz#1193929
> Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
Change-Id: I63be3893f92096e57f54a6150e0461340084ddde
BUG: 1647499
Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158658
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Pranith Kumar Karampuri <pkarampu@redhat.com>
---
libglusterfs/src/mem-pool.c | 137 ++++++++++++++++++++++++++------------------
1 file changed, 81 insertions(+), 56 deletions(-)
diff --git a/libglusterfs/src/mem-pool.c b/libglusterfs/src/mem-pool.c
index ba29137..8ff261c 100644
--- a/libglusterfs/src/mem-pool.c
+++ b/libglusterfs/src/mem-pool.c
@@ -411,37 +411,34 @@ static unsigned int init_count = 0;
static pthread_t sweeper_tid;
-void
+gf_boolean_t
collect_garbage (sweep_state_t *state, per_thread_pool_list_t *pool_list)
{
unsigned int i;
per_thread_pool_t *pt_pool;
-
- if (pool_list->poison) {
- list_del (&pool_list->thr_list);
- list_add (&pool_list->thr_list, &state->death_row);
- return;
- }
-
- if (state->n_cold_lists >= N_COLD_LISTS) {
- return;
- }
+ gf_boolean_t poisoned;
(void) pthread_spin_lock (&pool_list->lock);
- for (i = 0; i < NPOOLS; ++i) {
- pt_pool = &pool_list->pools[i];
- if (pt_pool->cold_list) {
- state->cold_lists[state->n_cold_lists++]
- = pt_pool->cold_list;
- }
- pt_pool->cold_list = pt_pool->hot_list;
- pt_pool->hot_list = NULL;
- if (state->n_cold_lists >= N_COLD_LISTS) {
- /* We'll just catch up on a future pass. */
- break;
+
+ poisoned = pool_list->poison != 0;
+ if (!poisoned) {
+ for (i = 0; i < NPOOLS; ++i) {
+ pt_pool = &pool_list->pools[i];
+ if (pt_pool->cold_list) {
+ if (state->n_cold_lists >= N_COLD_LISTS) {
+ break;
+ }
+ state->cold_lists[state->n_cold_lists++]
+ = pt_pool->cold_list;
+ }
+ pt_pool->cold_list = pt_pool->hot_list;
+ pt_pool->hot_list = NULL;
}
}
+
(void) pthread_spin_unlock (&pool_list->lock);
+
+ return poisoned;
}
@@ -469,6 +466,7 @@ pool_sweeper (void *arg)
struct timeval begin_time;
struct timeval end_time;
struct timeval elapsed;
+ gf_boolean_t poisoned;
/*
* This is all a bit inelegant, but the point is to avoid doing
@@ -488,7 +486,13 @@ pool_sweeper (void *arg)
(void) pthread_mutex_lock (&pool_lock);
list_for_each_entry_safe (pool_list, next_pl,
&pool_threads, thr_list) {
- collect_garbage (&state, pool_list);
+ (void) pthread_mutex_unlock (&pool_lock);
+ poisoned = collect_garbage (&state, pool_list);
+ (void) pthread_mutex_lock (&pool_lock);
+ if (poisoned) {
+ list_move(&pool_list->thr_list,
+ &state.death_row);
+ }
}
(void) pthread_mutex_unlock (&pool_lock);
(void) gettimeofday (&end_time, NULL);
@@ -525,7 +529,15 @@ pool_destructor (void *arg)
{
per_thread_pool_list_t *pool_list = arg;
- /* The pool-sweeper thread will take it from here. */
+ /* The pool-sweeper thread will take it from here.
+ *
+ * We can change 'poison' here without taking locks because the change
+ * itself doesn't interact with other parts of the code and a simple
+ * write is already atomic from the point of view of the processor.
+ *
+ * This change can modify what mem_put() does, but both possibilities
+ * are fine until the sweeper thread kicks in. The real synchronization
+ * must be between mem_put() and the sweeper thread. */
pool_list->poison = 1;
}
@@ -736,7 +748,7 @@ mem_get_pool_list (void)
(void) pthread_mutex_unlock (&pool_free_lock);
if (!pool_list) {
- pool_list = CALLOC (pool_list_size, 1);
+ pool_list = MALLOC (pool_list_size);
if (!pool_list) {
return NULL;
}
@@ -761,26 +773,47 @@ mem_get_pool_list (void)
}
pooled_obj_hdr_t *
-mem_get_from_pool (per_thread_pool_t *pt_pool)
+mem_get_from_pool (struct mem_pool *mem_pool)
{
+ per_thread_pool_list_t *pool_list;
+ per_thread_pool_t *pt_pool;
pooled_obj_hdr_t *retval;
+ pool_list = mem_get_pool_list ();
+ if (!pool_list || pool_list->poison) {
+ return NULL;
+ }
+
+ pt_pool = &pool_list->pools[mem_pool->power_of_two-POOL_SMALLEST];
+
+ (void) pthread_spin_lock (&pool_list->lock);
+
retval = pt_pool->hot_list;
if (retval) {
- GF_ATOMIC_INC (pt_pool->parent->allocs_hot);
pt_pool->hot_list = retval->next;
- return retval;
+ (void) pthread_spin_unlock (&pool_list->lock);
+ GF_ATOMIC_INC (pt_pool->parent->allocs_hot);
+ } else {
+ retval = pt_pool->cold_list;
+ if (retval) {
+ pt_pool->cold_list = retval->next;
+ (void) pthread_spin_unlock (&pool_list->lock);
+ GF_ATOMIC_INC (pt_pool->parent->allocs_cold);
+ } else {
+ (void) pthread_spin_unlock (&pool_list->lock);
+ GF_ATOMIC_INC (pt_pool->parent->allocs_stdc);
+ retval = malloc (1 << mem_pool->power_of_two);
+ }
}
- retval = pt_pool->cold_list;
if (retval) {
- GF_ATOMIC_INC (pt_pool->parent->allocs_cold);
- pt_pool->cold_list = retval->next;
- return retval;
+ retval->magic = GF_MEM_HEADER_MAGIC;
+ retval->next = NULL;
+ retval->pool_list = pool_list;
+ retval->power_of_two = mem_pool->power_of_two;
}
- GF_ATOMIC_INC (pt_pool->parent->allocs_stdc);
- return malloc (1 << pt_pool->parent->power_of_two);
+ return retval;
}
@@ -791,8 +824,6 @@ mem_get (struct mem_pool *mem_pool)
return GF_CALLOC (1, AVAILABLE_SIZE (mem_pool->power_of_two),
gf_common_mt_mem_pool);
#else
- per_thread_pool_list_t *pool_list;
- per_thread_pool_t *pt_pool;
pooled_obj_hdr_t *retval;
if (!mem_pool) {
@@ -801,25 +832,11 @@ mem_get (struct mem_pool *mem_pool)
return NULL;
}
- pool_list = mem_get_pool_list ();
- if (!pool_list || pool_list->poison) {
- return NULL;
- }
-
- (void) pthread_spin_lock (&pool_list->lock);
- pt_pool = &pool_list->pools[mem_pool->power_of_two-POOL_SMALLEST];
- retval = mem_get_from_pool (pt_pool);
- (void) pthread_spin_unlock (&pool_list->lock);
-
+ retval = mem_get_from_pool (mem_pool);
if (!retval) {
return NULL;
}
- retval->magic = GF_MEM_HEADER_MAGIC;
- retval->next = NULL;
- retval->pool_list = pool_list;;
- retval->power_of_two = mem_pool->power_of_two;
-
return retval + 1;
#endif /* GF_DISABLE_MEMPOOL */
}
@@ -849,12 +866,20 @@ mem_put (void *ptr)
pool_list = hdr->pool_list;
pt_pool = &pool_list->pools[hdr->power_of_two-POOL_SMALLEST];
- (void) pthread_spin_lock (&pool_list->lock);
hdr->magic = GF_MEM_INVALID_MAGIC;
- hdr->next = pt_pool->hot_list;
- pt_pool->hot_list = hdr;
- GF_ATOMIC_INC (pt_pool->parent->frees_to_list);
- (void) pthread_spin_unlock (&pool_list->lock);
+
+ (void) pthread_spin_lock (&pool_list->lock);
+ if (!pool_list->poison) {
+ hdr->next = pt_pool->hot_list;
+ pt_pool->hot_list = hdr;
+ (void) pthread_spin_unlock (&pool_list->lock);
+ GF_ATOMIC_INC (pt_pool->parent->frees_to_list);
+ } else {
+ (void) pthread_spin_unlock (&pool_list->lock);
+ /* If the owner thread of this element has terminated, we
+ * simply release its memory. */
+ free(hdr);
+ }
#endif /* GF_DISABLE_MEMPOOL */
}
--
1.8.3.1

51
0479-ganesha-ha-ensure-pacemaker-is-enabled-after-setup.patch Normal file
View File

@ -0,0 +1,51 @@
From e42fcda7ca4becd4e14b36c6318ed6c3a3068783 Mon Sep 17 00:00:00 2001
From: "Kaleb S. KEITHLEY" <kkeithle@redhat.com>
Date: Tue, 11 Dec 2018 10:09:42 -0500
Subject: [PATCH 479/493] ganesha-ha: ensure pacemaker is enabled after setup
There appears to be a race between `pcs cluster setup ...` early
in the setup and the `systemctl enable pacemaker` at the end. The
`pcs cluster setup ...` disables pacemaker and corosync. (Now, in
pacemaker-1.1.18. Was it always the case?)
I am not able to reproduce this on my devel system. I speculate that
on a busy system that the `pcs cluster setup ...` disable may, under
the right conditions, not run until after the setup script enables
it. It must require the right alignment of the Sun, Moon, and all
the planets.
Regardless, we'll use the --enable option to `pcs cluster setup ...`
to ensure that the cluster (re)starts pacemaker.
Label: DOWNSTREAM ONLY
Change-Id: I771ff62c37426438b80e61651a8b4ecaf2d549c3
BUG: 1637564
Signed-off-by: Kaleb S. KEITHLEY <kkeithle@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158294
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Soumya Koduri <skoduri@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
extras/ganesha/scripts/ganesha-ha.sh | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/extras/ganesha/scripts/ganesha-ha.sh b/extras/ganesha/scripts/ganesha-ha.sh
index 5cdafad..5a7f5ae 100644
--- a/extras/ganesha/scripts/ganesha-ha.sh
+++ b/extras/ganesha/scripts/ganesha-ha.sh
@@ -195,9 +195,9 @@ setup_cluster()
pcs cluster auth ${servers}
# pcs cluster setup --name ${name} ${servers}
- pcs cluster setup ${RHEL6_PCS_CNAME_OPTION} ${name} --transport udpu ${servers}
+ pcs cluster setup ${RHEL6_PCS_CNAME_OPTION} ${name} --enable --transport udpu ${servers}
if [ $? -ne 0 ]; then
- logger "pcs cluster setup ${RHEL6_PCS_CNAME_OPTION} ${name} ${servers} failed"
+ logger "pcs cluster setup ${RHEL6_PCS_CNAME_OPTION} ${name} --enable --transport udpu ${servers} failed"
#set up failed stop all ganesha process and clean up symlinks in cluster
stop_ganesha_all "${servers}"
exit 1;
--
1.8.3.1

101
0480-geo-rep-Make-slave-volume-read-only-by-default.patch Normal file
View File

@ -0,0 +1,101 @@
From ecb5c63d41daf98c3bca73d2ab24aa77e1b34886 Mon Sep 17 00:00:00 2001
From: Sunny Kumar <sunkumar@redhat.com>
Date: Mon, 17 Dec 2018 12:56:33 +0530
Subject: [PATCH 480/493] geo-rep: Make slave volume read-only (by default)
Added a command to set "features.read-only" option
to a default value "on" for slave volume.
Changes are made in:
$SRC//extras/hook-scripts/S56glusterd-geo-rep-create-post.sh
for root geo-rep and
$SRC/geo-replication/src/set_geo_rep_pem_keys.sh
for non-root geo-rep.
Upstream patch : https://review.gluster.org/#/c/glusterfs/+/21739
>Fixes: bz#1654187
>Signed-off-by: Harpreet Kaur <hlalwani@redhat.com>
Change-Id: I15beeae3506f3f6b1dcba0a5c50b6344fd468c7c
BUG: 1643370
Signed-off-by: Sunny Kumar <sunkumar@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158790
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Kotresh Hiremath Ravishankar <khiremat@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
extras/hook-scripts/S56glusterd-geo-rep-create-post.sh | 1 +
geo-replication/src/set_geo_rep_pem_keys.sh | 1 +
tests/geo-rep.rc | 7 +++++++
tests/geo-rep/georep-basic-dr-rsync.t | 3 +++
tests/geo-rep/georep-basic-dr-tarssh.t | 3 +++
5 files changed, 15 insertions(+)
diff --git a/extras/hook-scripts/S56glusterd-geo-rep-create-post.sh b/extras/hook-scripts/S56glusterd-geo-rep-create-post.sh
index a5e472e..589c263 100755
--- a/extras/hook-scripts/S56glusterd-geo-rep-create-post.sh
+++ b/extras/hook-scripts/S56glusterd-geo-rep-create-post.sh
@@ -90,5 +90,6 @@ if [ -f $pub_file ]; then
ssh -p ${SSH_PORT} $slave_ip "mv $pub_file_tmp ${pub_file_dname}/${mastervol}_${slavevol}_${pub_file_bname}"
ssh -p ${SSH_PORT} $slave_ip "gluster system:: copy file /geo-replication/${mastervol}_${slavevol}_common_secret.pem.pub > /dev/null"
ssh -p ${SSH_PORT} $slave_ip "gluster system:: execute add_secret_pub root geo-replication/${mastervol}_${slavevol}_common_secret.pem.pub > /dev/null"
+ ssh -p ${SSH_PORT} $slave_ip "gluster vol set ${slavevol} features.read-only on"
fi
fi
diff --git a/geo-replication/src/set_geo_rep_pem_keys.sh b/geo-replication/src/set_geo_rep_pem_keys.sh
index ae23f4f..8a43fa3 100755
--- a/geo-replication/src/set_geo_rep_pem_keys.sh
+++ b/geo-replication/src/set_geo_rep_pem_keys.sh
@@ -47,6 +47,7 @@ function main()
cp $home_dir/${COMMON_SECRET_PEM_PUB} ${GLUSTERD_WORKDIR}/geo-replication/
gluster system:: copy file /geo-replication/${COMMON_SECRET_PEM_PUB}
gluster system:: execute add_secret_pub $user geo-replication/${master_vol}_${slave_vol}_common_secret.pem.pub
+ gluster vol set ${slave_vol} features.read-only on
else
echo "$home_dir/common_secret.pem.pub not present. Please run geo-replication command on master with push-pem option to generate the file"
exit 1;
diff --git a/tests/geo-rep.rc b/tests/geo-rep.rc
index 1a44b4a..f9ab3fc 100644
--- a/tests/geo-rep.rc
+++ b/tests/geo-rep.rc
@@ -230,3 +230,10 @@ function hardlink_rename_ok()
fi
return 0
}
+
+function check_slave_read_only()
+{
+ volum=$1
+ gluster volume info $1 | grep 'features.read-only: on'
+ echo $?
+}
diff --git a/tests/geo-rep/georep-basic-dr-rsync.t b/tests/geo-rep/georep-basic-dr-rsync.t
index 39da524..fd0c4aa 100644
--- a/tests/geo-rep/georep-basic-dr-rsync.t
+++ b/tests/geo-rep/georep-basic-dr-rsync.t
@@ -57,6 +57,9 @@ EXPECT 4 check_status_num_rows "Created" #15
#Enable_metavolume
TEST $GEOREP_CLI $master $slave config use_meta_volume true
+#Verify "features.read-only" Option
+EXPECT_WITHIN $GEO_REP_TIMEOUT 0 check_slave_read_only $GSV0
+
#Start_georep
TEST $GEOREP_CLI $master $slave start
diff --git a/tests/geo-rep/georep-basic-dr-tarssh.t b/tests/geo-rep/georep-basic-dr-tarssh.t
index 5f879db..5331df9 100644
--- a/tests/geo-rep/georep-basic-dr-tarssh.t
+++ b/tests/geo-rep/georep-basic-dr-tarssh.t
@@ -62,6 +62,9 @@ TEST $GEOREP_CLI $master $slave config use_meta_volume true
#Config tarssh as sync-engine
TEST $GEOREP_CLI $master $slave config use_tarssh true
+#Verify "features.read-only" Option
+EXPECT_WITHIN $GEO_REP_TIMEOUT 0 check_slave_read_only $GSV0
+
#Start_georep
TEST $GEOREP_CLI $master $slave start
--
1.8.3.1

102
0481-extras-hooks-Do-not-blindly-remove-volume-share-from.patch Normal file
View File

@ -0,0 +1,102 @@
From 644bcd954000b77fc8f49e3a7941de23ca869427 Mon Sep 17 00:00:00 2001
From: Anoop C S <anoopcs@redhat.com>
Date: Wed, 21 Mar 2018 13:09:01 +0530
Subject: [PATCH 481/493] extras/hooks: Do not blindly remove volume share from
smb.conf
When Gluster volumes are shared via Samba, any extra smb.conf parameter
setting done by administrator to those shares are lost during restart
of the volume. Instead of removing the whole share completely from
smb.conf(via hook scripts during volume stop) it is better to make it
temporarily unavailable to end-users till the volume is started again.
Therefore we make use of a smb.conf parameter named 'available'[1] to
achieve the above intend.
[1] https://www.samba.org/samba/docs/current/man-html/smb.conf.5.html
upstream ref: https://review.gluster.org/c/glusterfs/+/19750
Change-Id: I68a9055b50791f6ffd3b95a3c13d858a75fa6530
BUG: 1164778
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158495
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 | 11 ++++++-----
extras/hook-scripts/start/post/S30samba-start.sh | 4 +++-
extras/hook-scripts/stop/pre/S30samba-stop.sh | 6 +++---
3 files changed, 12 insertions(+), 9 deletions(-)
diff --git a/extras/hook-scripts/set/post/S30samba-set.sh b/extras/hook-scripts/set/post/S30samba-set.sh
index b93415b..c596db0 100755
--- a/extras/hook-scripts/set/post/S30samba-set.sh
+++ b/extras/hook-scripts/set/post/S30samba-set.sh
@@ -103,9 +103,9 @@ function sighup_samba () {
fi
}
-function del_samba_share () {
+function deactivate_samba_share () {
volname=$1
- sed -i "/\[gluster-$volname\]/,/^$/d" ${CONFIGFILE}
+ sed -i -e '/^\[gluster-'"$volname"'\]/{ :a' -e 'n; /available = no/H; /^$/!{$!ba;}; x; /./!{ s/^/available = no/; $!{G;x}; $H; }; s/.*//; x; };' ${CONFIGFILE}
}
function is_volume_started () {
@@ -140,12 +140,13 @@ if [ "$USERCIFS_SET" = "YES" ] || [ "$USERSMB_SET" = "YES" ]; then
find_config_info
if [ "$(get_smb "$VOL")" = "disable" ]; then
- del_samba_share $VOL
- sighup_samba
+ deactivate_samba_share $VOL
else
if ! grep --quiet "\[gluster-$VOL\]" ${CONFIGFILE} ; then
add_samba_share $VOL
- sighup_samba
+ else
+ sed -i '/\[gluster-'"$VOL"'\]/,/^$/!b;/available = no/d' ${CONFIGFILE}
fi
fi
+ sighup_samba
fi
diff --git a/extras/hook-scripts/start/post/S30samba-start.sh b/extras/hook-scripts/start/post/S30samba-start.sh
index 92ddaf4..5d586ee 100755
--- a/extras/hook-scripts/start/post/S30samba-start.sh
+++ b/extras/hook-scripts/start/post/S30samba-start.sh
@@ -127,5 +127,7 @@ find_config_info
if ! grep --quiet "\[gluster-$VOL\]" ${CONFIGFILE} ; then
add_samba_share $VOL
- sighup_samba
+else
+ sed -i '/\[gluster-'"$VOL"'\]/,/^$/!b;/available = no/d' ${CONFIGFILE}
fi
+sighup_samba
diff --git a/extras/hook-scripts/stop/pre/S30samba-stop.sh b/extras/hook-scripts/stop/pre/S30samba-stop.sh
index 5e87845..ea79938 100755
--- a/extras/hook-scripts/stop/pre/S30samba-stop.sh
+++ b/extras/hook-scripts/stop/pre/S30samba-stop.sh
@@ -56,9 +56,9 @@ function find_config_info () {
PIDDIR=`smbd -b | grep PIDDIR | awk '{print $2}'`
}
-function del_samba_share () {
+function deactivate_samba_share () {
volname=$1
- sed -i "/\[gluster-$volname\]/,/^$/d" ${CONFIGFILE}
+ sed -i -e '/^\[gluster-'"$volname"'\]/{ :a' -e 'n; /available = no/H; /^$/!{$!ba;}; x; /./!{ s/^/available = no/; $!{G;x}; $H; }; s/.*//; x; };' ${CONFIGFILE}
}
function sighup_samba () {
@@ -73,5 +73,5 @@ function sighup_samba () {
parse_args "$@"
find_config_info
-del_samba_share $VOL
+deactivate_samba_share $VOL
sighup_samba
--
1.8.3.1

91
0482-extras-hooks-General-improvements-to-S30samba-start..patch Normal file
View File

@ -0,0 +1,91 @@
From 121180edc218432a782a153e94b9f884d4c56a7c Mon Sep 17 00:00:00 2001
From: Anoop C S <anoopcs@redhat.com>
Date: Sun, 9 Sep 2018 11:39:47 +0530
Subject: [PATCH 482/493] extras/hooks: General improvements to
S30samba-start.sh
Based on https://review.gluster.org/c/glusterfs/+/19204 from
Milan Zink <mzink@redhat.com>
upstream ref: https://review.gluster.org/c/glusterfs/+/21124
Change-Id: I61f62407a62475a74a6cc046c24748a31c66e6cd
BUG: 1541568
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158496
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
extras/hook-scripts/start/post/S30samba-start.sh | 26 ++++++++++++------------
1 file changed, 13 insertions(+), 13 deletions(-)
diff --git a/extras/hook-scripts/start/post/S30samba-start.sh b/extras/hook-scripts/start/post/S30samba-start.sh
index 5d586ee..dfd9c1b 100755
--- a/extras/hook-scripts/start/post/S30samba-start.sh
+++ b/extras/hook-scripts/start/post/S30samba-start.sh
@@ -68,14 +68,14 @@ function parse_args () {
}
function find_config_info () {
- cmdout=`smbd -b | grep smb.conf`
- if [ $? -ne 0 ];then
+ cmdout=$(smbd -b 2> /dev/null)
+ CONFIGFILE=$(echo "$cmdout" | grep CONFIGFILE | awk '{print $2}')
+ if [ -z "$CONFIGFILE" ]; then
echo "Samba is not installed"
exit 1
fi
- CONFIGFILE=`echo $cmdout | awk '{print $2}'`
- PIDDIR=`smbd -b | grep PIDDIR | awk '{print $2}'`
- LOGFILEBASE=`smbd -b | grep 'LOGFILEBASE' | awk '{print $2}'`
+ PIDDIR=$(echo "$cmdout" | grep PIDDIR | awk '{print $2}')
+ LOGFILEBASE=$(echo "$cmdout" | grep 'LOGFILEBASE' | awk '{print $2}')
}
function add_samba_share () {
@@ -89,11 +89,11 @@ function add_samba_share () {
STRING+="path = /\n"
STRING+="read only = no\n"
STRING+="guest ok = yes\n"
- printf "$STRING" >> ${CONFIGFILE}
+ printf "$STRING" >> "${CONFIGFILE}"
}
function sighup_samba () {
- pid=`cat ${PIDDIR}/smbd.pid`
+ pid=$(cat "${PIDDIR}/smbd.pid" 2> /dev/null)
if [ "x$pid" != "x" ]
then
kill -HUP "$pid";
@@ -106,12 +106,12 @@ function get_smb () {
volname=$1
uservalue=
- usercifsvalue=$(grep user.cifs $GLUSTERD_WORKDIR/vols/"$volname"/info |\
+ usercifsvalue=$(grep user.cifs "$GLUSTERD_WORKDIR"/vols/"$volname"/info |\
cut -d"=" -f2)
- usersmbvalue=$(grep user.smb $GLUSTERD_WORKDIR/vols/"$volname"/info |\
+ usersmbvalue=$(grep user.smb "$GLUSTERD_WORKDIR"/vols/"$volname"/info |\
cut -d"=" -f2)
- if [ $usercifsvalue = "disable" ] || [ $usersmbvalue = "disable" ]; then
+ if [ "$usercifsvalue" = "disable" ] || [ "$usersmbvalue" = "disable" ]; then
uservalue="disable"
fi
echo "$uservalue"
@@ -125,9 +125,9 @@ fi
#Find smb.conf, smbd pid directory and smbd logfile path
find_config_info
-if ! grep --quiet "\[gluster-$VOL\]" ${CONFIGFILE} ; then
- add_samba_share $VOL
+if ! grep --quiet "\[gluster-$VOL\]" "${CONFIGFILE}" ; then
+ add_samba_share "$VOL"
else
- sed -i '/\[gluster-'"$VOL"'\]/,/^$/!b;/available = no/d' ${CONFIGFILE}
+ sed -i '/\[gluster-'"$VOL"'\]/,/^$/!b;/available = no/d' "${CONFIGFILE}"
fi
sighup_samba
--
1.8.3.1

86
0483-Do-not-blindly-add-volume-share-section-to-smb.conf.patch Normal file
View File

@ -0,0 +1,86 @@
From c807ba5a11364d8eb83b86b0e4262a32b6834267 Mon Sep 17 00:00:00 2001
From: Anoop C S <anoopcs@redhat.com>
Date: Thu, 13 Dec 2018 12:05:37 +0530
Subject: [PATCH 483/493] Do not blindly add volume share section to smb.conf
With this change, by default GlusterFS volume share section will
no longer be added to smb.conf for client access unless user.cifs
or user.smb volume set options are enabled. This also fixes the
hook script to check for presence of all configuration possibilities
for those volume set options like 'enable' or 'on'.
upstream ref: https://review.gluster.org/c/glusterfs/+/19204
Change-Id: Ibecf7fffb4507d7255d963c3b1482afb0d0db984
BUG: 1541568
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158497
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 | 13 +++++++++++--
extras/hook-scripts/start/post/S30samba-start.sh | 18 +++++++++++++++---
2 files changed, 26 insertions(+), 5 deletions(-)
diff --git a/extras/hook-scripts/set/post/S30samba-set.sh b/extras/hook-scripts/set/post/S30samba-set.sh
index c596db0..d2a62d3 100755
--- a/extras/hook-scripts/set/post/S30samba-set.sh
+++ b/extras/hook-scripts/set/post/S30samba-set.sh
@@ -123,9 +123,18 @@ function get_smb () {
usersmbvalue=$(grep user.smb $GLUSTERD_WORKDIR/vols/"$volname"/info |\
cut -d"=" -f2)
- if [ $usercifsvalue = "disable" ] || [ $usersmbvalue = "disable" ]; then
- uservalue="disable"
+ if [ -n "$usercifsvalue" ]; then
+ if [ "$usercifsvalue" = "disable" ] || [ "$usercifsvalue" = "off" ]; then
+ uservalue="disable"
+ fi
fi
+
+ if [ -n "$usersmbvalue" ]; then
+ if [ "$usersmbvalue" = "disable" ] || [ "$usersmbvalue" = "off" ]; then
+ uservalue="disable"
+ fi
+ fi
+
echo "$uservalue"
}
diff --git a/extras/hook-scripts/start/post/S30samba-start.sh b/extras/hook-scripts/start/post/S30samba-start.sh
index dfd9c1b..2854bdd 100755
--- a/extras/hook-scripts/start/post/S30samba-start.sh
+++ b/extras/hook-scripts/start/post/S30samba-start.sh
@@ -111,14 +111,26 @@ function get_smb () {
usersmbvalue=$(grep user.smb "$GLUSTERD_WORKDIR"/vols/"$volname"/info |\
cut -d"=" -f2)
- if [ "$usercifsvalue" = "disable" ] || [ "$usersmbvalue" = "disable" ]; then
- uservalue="disable"
+ if [ -n "$usercifsvalue" ]; then
+ if [ "$usercifsvalue" = "enable" ] || [ "$usercifsvalue" = "on" ]; then
+ uservalue="enable"
+ fi
fi
+
+ if [ -n "$usersmbvalue" ]; then
+ if [ "$usersmbvalue" = "enable" ] || [ "$usersmbvalue" = "on" ]; then
+ uservalue="enable"
+ fi
+ fi
+
echo "$uservalue"
}
parse_args "$@"
-if [ "$(get_smb "$VOL")" = "disable" ]; then
+
+value=$(get_smb "$VOL")
+
+if [ -z "$value" ] || [ "$value" != "enable" ]; then
exit 0
fi
--
1.8.3.1

84
0484-extras-New-group-volume-set-command-for-Samba-integr.patch Normal file
View File

@ -0,0 +1,84 @@
From bf8ca8c73df0a114b9728150934d6b7ecd3cbd6f Mon Sep 17 00:00:00 2001
From: Anoop C S <anoopcs@redhat.com>
Date: Thu, 6 Dec 2018 15:05:20 +0530
Subject: [PATCH 484/493] extras: New group volume set command for Samba
integration
# gluster volume set <VOLNAME> group samba
List of volume set options from group-samba are aimed at improving the below
workloads which consumes time and network hops in SMB world:
* Listing of large directories
* Negative lookups during creation of files
Caching the necessary metadata required for these workloads saves us time and
network hops. On the other side we have to ensure correctness(avoiding stale
cache) in caching(via md-cache) with the help of cache invalidation in an
environment where multiple client access is expected.
upstream ref: https://review.gluster.org/c/glusterfs/+/21814
Change-Id: Icdd2d8e5eb290e12bc509105418c668f432f4eae
BUG: 1655385
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158723
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
extras/Makefile.am | 4 +++-
extras/group-samba | 10 ++++++++++
glusterfs.spec.in | 1 +
3 files changed, 14 insertions(+), 1 deletion(-)
create mode 100644 extras/group-samba
diff --git a/extras/Makefile.am b/extras/Makefile.am
index e0e05b5..f898245 100644
--- a/extras/Makefile.am
+++ b/extras/Makefile.am
@@ -14,7 +14,7 @@ confdir = $(sysconfdir)/glusterfs
conf_DATA = glusterfs-logrotate gluster-rsyslog-7.2.conf gluster-rsyslog-5.8.conf \
logger.conf.example glusterfs-georep-logrotate group-virt.example \
group-metadata-cache group-gluster-block group-nl-cache group-db-workload \
- group-distributed-virt
+ group-distributed-virt group-samba
voldir = $(sysconfdir)/glusterfs
vol_DATA = glusterd.vol
@@ -53,3 +53,5 @@ install-data-local:
$(DESTDIR)$(GLUSTERD_WORKDIR)/groups/db-workload
$(INSTALL_DATA) $(top_srcdir)/extras/group-distributed-virt \
$(DESTDIR)$(GLUSTERD_WORKDIR)/groups/distributed-virt
+ $(INSTALL_DATA) $(top_srcdir)/extras/group-samba \
+ $(DESTDIR)$(GLUSTERD_WORKDIR)/groups/samba
diff --git a/extras/group-samba b/extras/group-samba
new file mode 100644
index 0000000..ee39202
--- /dev/null
+++ b/extras/group-samba
@@ -0,0 +1,10 @@
+features.cache-invalidation=on
+features.cache-invalidation-timeout=600
+performance.cache-samba-metadata=on
+performance.stat-prefetch=on
+performance.cache-invalidation=on
+performance.md-cache-timeout=600
+network.inode-lru-limit=200000
+performance.nl-cache=on
+performance.nl-cache-timeout=600
+performance.parallel-readdir=on
diff --git a/glusterfs.spec.in b/glusterfs.spec.in
index a4accd9..16dc5d7 100644
--- a/glusterfs.spec.in
+++ b/glusterfs.spec.in
@@ -1524,6 +1524,7 @@ exit 0
%attr(0644,-,-) %{_sharedstatedir}/glusterd/groups/db-workload
%attr(0644,-,-) %{_sharedstatedir}/glusterd/groups/distributed-virt
%attr(0644,-,-) %{_sharedstatedir}/glusterd/groups/nl-cache
+ %attr(0644,-,-) %{_sharedstatedir}/glusterd/groups/samba
%dir %attr(0755,-,-) %{_sharedstatedir}/glusterd/glusterfind
%dir %attr(0755,-,-) %{_sharedstatedir}/glusterd/glusterfind/.keys
%ghost %dir %attr(0755,-,-) %{_sharedstatedir}/glusterd/glustershd
--
1.8.3.1

60
0485-cluster-ec-Prevent-volume-create-without-redundant-b.patch Normal file
View File

@ -0,0 +1,60 @@
From f3982cf5bf3250ee29da3ea4477591c298780043 Mon Sep 17 00:00:00 2001
From: Sunil Kumar Acharya <sheggodu@redhat.com>
Date: Mon, 17 Dec 2018 16:37:21 +0530
Subject: [PATCH 485/493] cluster/ec : Prevent volume create without redundant
brick
Problem:
EC volumes can be created without any redundant brick.
Solution:
Updated the conditional check to avoid volume create
without redundant brick.
>fixes: bz#1642448
>Change-Id: I0cb334b1b9378d67fcb8abf793dbe312c3179c0b
>Signed-off-by: Sunil Kumar Acharya <sheggodu@redhat.com>
Upstream Patch: https://review.gluster.org/#/c/glusterfs/+/21478/
BUG: 1597252
Change-Id: I0cb334b1b9378d67fcb8abf793dbe312c3179c0b
Signed-off-by: Sunil Kumar Acharya <sheggodu@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158910
Tested-by: RHGS Build Bot <nigelb@redhat.com>
---
cli/src/cli-cmd-parser.c | 6 +++---
tests/basic/glusterd/disperse-create.t | 1 +
2 files changed, 4 insertions(+), 3 deletions(-)
diff --git a/cli/src/cli-cmd-parser.c b/cli/src/cli-cmd-parser.c
index 3745fb4..e790d79 100644
--- a/cli/src/cli-cmd-parser.c
+++ b/cli/src/cli-cmd-parser.c
@@ -275,9 +275,9 @@ cli_cmd_create_disperse_check (struct cli_state *state, int *disperse,
tmp = *disperse - *redundancy;
}
- if (*redundancy > (*disperse - 1) / 2) {
- cli_err ("redundancy must be less than %d for a "
- "disperse %d volume",
+ if ((*redundancy < 1) || (*redundancy > (*disperse - 1) / 2)) {
+ cli_err ("redundancy must be greater than or equal to 1 and"
+ " less than %d for a disperse %d volume",
(*disperse + 1) / 2, *disperse);
return -1;
diff --git a/tests/basic/glusterd/disperse-create.t b/tests/basic/glusterd/disperse-create.t
index e5ce74c..5b3ed13 100644
--- a/tests/basic/glusterd/disperse-create.t
+++ b/tests/basic/glusterd/disperse-create.t
@@ -48,6 +48,7 @@ TEST ! $CLI volume create $V0 redundancy 1 redundancy 1 $H0:$B0/b20 $H0:$B0/b21
#Minimum counts test
TEST ! $CLI volume create $V0 disperse 2 $H0:$B0/b20 $H0:$B0/b22
TEST ! $CLI volume create $V0 disperse-data 1 redundancy 0 $H0:$B0/b20 $H0:$B0/b22
+TEST ! $CLI volume create $V0 disperse 4 disperse-data 4 $H0:$B0/b20 $H0:$B0/b21 $H0:$B0/b23 $H0:$B0/b24
TEST ! $CLI volume create $V0 redundancy 0 $H0:$B0/b20 $H0:$B0/b22
#Wrong count n != k+m
--
1.8.3.1

51
0486-performance-rda-Fixed-dict_t-memory-leak.patch Normal file
View File

@ -0,0 +1,51 @@
From 64cfb0126d88bd6b841777fefb63e413b587a7b1 Mon Sep 17 00:00:00 2001
From: N Balachandran <nbalacha@redhat.com>
Date: Tue, 18 Dec 2018 14:08:04 +0530
Subject: [PATCH 486/493] performance/rda: Fixed dict_t memory leak
Removed all references to dict_t xdata_from_req which is
allocated but not used anywhere. It is also not cleaned up
and hence causes a memory leak.
upstream patch: https://review.gluster.org/#/c/glusterfs/+/21859/
> Change-Id: I2edb857696191e872ad12a12efc36999626bacc7
> fixes: bz#1659432
> Signed-off-by: N Balachandran <nbalacha@redhat.com>
Change-Id: Ic07ecbefef3140aeb24a2afaf97c80ee65768e7e
BUG: 1659439
Signed-off-by: N Balachandran <nbalacha@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158915
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Raghavendra Gowdappa <rgowdapp@redhat.com>
---
xlators/performance/readdir-ahead/src/readdir-ahead.c | 8 --------
1 file changed, 8 deletions(-)
diff --git a/xlators/performance/readdir-ahead/src/readdir-ahead.c b/xlators/performance/readdir-ahead/src/readdir-ahead.c
index 6501a6b..3e0951c 100644
--- a/xlators/performance/readdir-ahead/src/readdir-ahead.c
+++ b/xlators/performance/readdir-ahead/src/readdir-ahead.c
@@ -505,18 +505,10 @@ rda_opendir(call_frame_t *frame, xlator_t *this, loc_t *loc, fd_t *fd,
{
int op_errno = 0;
struct rda_local *local = NULL;
- dict_t *xdata_from_req = NULL;
if (xdata) {
- xdata_from_req = dict_new();
- if (!xdata_from_req) {
- op_errno = ENOMEM;
- goto unwind;
- }
-
local = mem_get0(this->local_pool);
if (!local) {
- dict_unref(xdata_from_req);
op_errno = ENOMEM;
goto unwind;
}
--
1.8.3.1

270
0487-mem-pool-add-tracking-of-mem_pool-that-requested-the.patch Normal file
View File

@ -0,0 +1,270 @@
From 3cc901acea41632df0c342639c4292c10bd90964 Mon Sep 17 00:00:00 2001
From: Mohit Agrawal <moagrawa@redhat.com>
Date: Tue, 18 Dec 2018 15:39:14 +0530
Subject: [PATCH 487/493] mem-pool: add tracking of mem_pool that requested the
allocation
This renames the current 'struct mem_pool' to 'struct mem_pool_shared'.
The mem_pool_shared is globally allocated and not specific for
particular objects.
A new 'struct mem_pool' gets allocated when mem_pool_new() is called. It
points to the mem_pool_shared that handles the actual allocation
requests. The 'struct mem_pool' is only used for accounting of the
objects that the caller requested and free'd.
All of these changes will be used to collect all the memory pools a
glusterfs_ctx_t is consuming, so that statedumps can be collected per
context.
> Updates: #307
> Change-Id: I6355d3f0251c928e0bbfc71be3431307c6f3a3da
> Signed-off-by: Niels de Vos <ndevos@redhat.com>
> Reviewed-on: https://review.gluster.org/18073
> Smoke: Gluster Build System <jenkins@build.gluster.org>
> CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
> Reviewed-by: Amar Tumballi <amarts@redhat.com>
> Reviewed-by: Jeff Darcy <jeff@pl.atyp.us>
> Cherry picked from commit 2645e730b79b44fc035170657e43bb52f3e855c5
Change-Id: I6cce6284e4553c6ca59a90ad124c23c950db3148
BUG: 1648893
Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
Change-Id: I363d71152b1dd17eca53d9c327fcdf2f26c0fb61
Reviewed-on: https://code.engineering.redhat.com/gerrit/158930
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
libglusterfs/src/mem-pool.c | 69 +++++++++++++++++++++++++++-----------------
libglusterfs/src/mem-pool.h | 20 +++++++++++--
libglusterfs/src/mem-types.h | 2 --
3 files changed, 60 insertions(+), 31 deletions(-)
diff --git a/libglusterfs/src/mem-pool.c b/libglusterfs/src/mem-pool.c
index 8ff261c..a8a9347 100644
--- a/libglusterfs/src/mem-pool.c
+++ b/libglusterfs/src/mem-pool.c
@@ -14,15 +14,6 @@
#include <stdlib.h>
#include <stdarg.h>
-#define GF_MEM_POOL_LIST_BOUNDARY (sizeof(struct list_head))
-#define GF_MEM_POOL_PTR (sizeof(struct mem_pool*))
-#define GF_MEM_POOL_PAD_BOUNDARY (GF_MEM_POOL_LIST_BOUNDARY + GF_MEM_POOL_PTR + sizeof(int))
-#define mem_pool_chunkhead2ptr(head) ((head) + GF_MEM_POOL_PAD_BOUNDARY)
-#define mem_pool_ptr2chunkhead(ptr) ((ptr) - GF_MEM_POOL_PAD_BOUNDARY)
-#define is_mem_chunk_in_use(ptr) (*ptr == 1)
-#define mem_pool_from_ptr(ptr) ((ptr) + GF_MEM_POOL_LIST_BOUNDARY)
-
-#define GLUSTERFS_ENV_MEM_ACCT_STR "GLUSTERFS_DISABLE_MEM_ACCT"
#include "unittest/unittest.h"
#include "libglusterfs-messages.h"
@@ -380,7 +371,7 @@ static pthread_mutex_t pool_lock = PTHREAD_MUTEX_INITIALIZER;
static struct list_head pool_threads;
static pthread_mutex_t pool_free_lock = PTHREAD_MUTEX_INITIALIZER;
static struct list_head pool_free_threads;
-static struct mem_pool pools[NPOOLS];
+static struct mem_pool_shared pools[NPOOLS];
static size_t pool_list_size;
#if !defined(GF_DISABLE_MEMPOOL)
@@ -689,6 +680,8 @@ mem_pool_new_fn (unsigned long sizeof_type,
unsigned long count, char *name)
{
unsigned int i;
+ struct mem_pool *new = NULL;
+ struct mem_pool_shared *pool = NULL;
if (!sizeof_type) {
gf_msg_callingfn ("mem-pool", GF_LOG_ERROR, EINVAL,
@@ -698,13 +691,27 @@ mem_pool_new_fn (unsigned long sizeof_type,
for (i = 0; i < NPOOLS; ++i) {
if (sizeof_type <= AVAILABLE_SIZE(pools[i].power_of_two)) {
- return &pools[i];
+ pool = &pools[i];
+ break;
}
}
- gf_msg_callingfn ("mem-pool", GF_LOG_ERROR, EINVAL,
- LG_MSG_INVALID_ARG, "invalid argument");
- return NULL;
+ if (!pool) {
+ gf_msg_callingfn ("mem-pool", GF_LOG_ERROR, EINVAL,
+ LG_MSG_INVALID_ARG, "invalid argument");
+ return NULL;
+ }
+
+ new = GF_CALLOC (sizeof (struct mem_pool), 1, gf_common_mt_mem_pool);
+ if (!new)
+ return NULL;
+
+ new->sizeof_type = sizeof_type;
+ new->count = count;
+ new->name = name;
+ new->pool = pool;
+
+ return new;
}
void*
@@ -721,7 +728,7 @@ mem_get0 (struct mem_pool *mem_pool)
ptr = mem_get(mem_pool);
if (ptr) {
- memset (ptr, 0, AVAILABLE_SIZE(mem_pool->power_of_two));
+ memset (ptr, 0, AVAILABLE_SIZE(mem_pool->pool->power_of_two));
}
return ptr;
@@ -784,7 +791,7 @@ mem_get_from_pool (struct mem_pool *mem_pool)
return NULL;
}
- pt_pool = &pool_list->pools[mem_pool->power_of_two-POOL_SMALLEST];
+ pt_pool = &pool_list->pools[mem_pool->pool->power_of_two-POOL_SMALLEST];
(void) pthread_spin_lock (&pool_list->lock);
@@ -802,7 +809,7 @@ mem_get_from_pool (struct mem_pool *mem_pool)
} else {
(void) pthread_spin_unlock (&pool_list->lock);
GF_ATOMIC_INC (pt_pool->parent->allocs_stdc);
- retval = malloc (1 << mem_pool->power_of_two);
+ retval = malloc (1 << mem_pool->pool->power_of_two);
}
}
@@ -810,7 +817,7 @@ mem_get_from_pool (struct mem_pool *mem_pool)
retval->magic = GF_MEM_HEADER_MAGIC;
retval->next = NULL;
retval->pool_list = pool_list;
- retval->power_of_two = mem_pool->power_of_two;
+ retval->power_of_two = mem_pool->pool->power_of_two;
}
return retval;
@@ -821,9 +828,10 @@ void *
mem_get (struct mem_pool *mem_pool)
{
#if defined(GF_DISABLE_MEMPOOL)
- return GF_CALLOC (1, AVAILABLE_SIZE (mem_pool->power_of_two),
+ return GF_CALLOC (1, AVAILABLE_SIZE (mem_pool->pool->power_of_two),
gf_common_mt_mem_pool);
#else
+ per_thread_pool_list_t *pool_list;
pooled_obj_hdr_t *retval;
if (!mem_pool) {
@@ -832,11 +840,22 @@ mem_get (struct mem_pool *mem_pool)
return NULL;
}
+ pool_list = mem_get_pool_list ();
+ if (!pool_list || pool_list->poison) {
+ return NULL;
+ }
+
retval = mem_get_from_pool (mem_pool);
+
if (!retval) {
return NULL;
}
+ retval->magic = GF_MEM_HEADER_MAGIC;
+ retval->pool = mem_pool;
+ retval->pool_list = pool_list;
+ retval->power_of_two = mem_pool->pool->power_of_two;
+
return retval + 1;
#endif /* GF_DISABLE_MEMPOOL */
}
@@ -886,14 +905,12 @@ mem_put (void *ptr)
void
mem_pool_destroy (struct mem_pool *pool)
{
- if (!pool)
- return;
+ GF_FREE (pool);
/*
- * Pools are now permanent, so this does nothing. Yes, this means we
- * can keep allocating from a pool after calling mem_destroy on it, but
- * that's kind of OK. All of the objects *in* the pool will eventually
- * be freed via the pool-sweeper thread, and this way we don't have to
- * add a lot of reference-counting complexity.
+ * Pools are now permanent, so the mem_pool->pool is kept around. All
+ * of the objects *in* the pool will eventually be freed via the
+ * pool-sweeper thread, and this way we don't have to add a lot of
+ * reference-counting complexity.
*/
}
diff --git a/libglusterfs/src/mem-pool.h b/libglusterfs/src/mem-pool.h
index dfe1f9a..057d957 100644
--- a/libglusterfs/src/mem-pool.h
+++ b/libglusterfs/src/mem-pool.h
@@ -204,18 +204,31 @@ out:
return dup_mem;
}
+/* kind of 'header' for the actual mem_pool_shared structure, this might make
+ * it possible to dump some more details in a statedump */
+struct mem_pool {
+ unsigned long sizeof_type;
+ unsigned long count;
+ char *name;
+
+ struct mem_pool_shared *pool;
+};
+
typedef struct pooled_obj_hdr {
unsigned long magic;
struct pooled_obj_hdr *next;
struct per_thread_pool_list *pool_list;
unsigned int power_of_two;
+
+ /* track the pool that was used to request this object */
+ struct mem_pool *pool;
} pooled_obj_hdr_t;
#define AVAILABLE_SIZE(p2) ((1 << (p2)) - sizeof(pooled_obj_hdr_t))
typedef struct per_thread_pool {
- /* This never changes, so doesn't need a lock. */
- struct mem_pool *parent;
+ /* the pool that was used to request this allocation */
+ struct mem_pool_shared *parent;
/* Everything else is protected by our own lock. */
pooled_obj_hdr_t *hot_list;
pooled_obj_hdr_t *cold_list;
@@ -243,7 +256,8 @@ typedef struct per_thread_pool_list {
per_thread_pool_t pools[1];
} per_thread_pool_list_t;
-struct mem_pool {
+/* actual pool structure, shared between different mem_pools */
+struct mem_pool_shared {
unsigned int power_of_two;
/*
* Updates to these are *not* protected by a global lock, so races
diff --git a/libglusterfs/src/mem-types.h b/libglusterfs/src/mem-types.h
index 85cb5d2..64d0e90 100644
--- a/libglusterfs/src/mem-types.h
+++ b/libglusterfs/src/mem-types.h
@@ -61,9 +61,7 @@ enum gf_common_mem_types_ {
gf_common_mt_char,
gf_common_mt_rbthash_table_t,
gf_common_mt_rbthash_bucket,
-#if defined(GF_DISABLE_MEMPOOL)
gf_common_mt_mem_pool,
-#endif
gf_common_mt_long,
gf_common_mt_rpcsvc_auth_list,
gf_common_mt_rpcsvc_t,
--
1.8.3.1

459
0488-cluster-afr-Allow-lookup-on-root-if-it-is-from-ADD_R.patch Normal file
View File

@ -0,0 +1,459 @@
From f398b6b9705f1b75d17d965ffcd72157d5be3daf Mon Sep 17 00:00:00 2001
From: karthik-us <ksubrahm@redhat.com>
Date: Tue, 18 Dec 2018 16:04:42 +0530
Subject: [PATCH 488/493] cluster/afr: Allow lookup on root if it is from
ADD_REPLICA_MOUNT
Problem: When trying to convert a plain distribute volume to replica-3
or arbiter type it is failing with ENOTCONN error as the lookup on
the root will fail as there is no quorum.
Fix: Allow lookup on root if it is coming from the ADD_REPLICA_MOUNT
which is used while adding bricks to a volume. It will try to set the
pending xattrs for the newly added bricks to allow the heal to happen
in the right direction and avoid data loss scenarios.
Note: This fix will solve the problem of type conversion only in the
case where the volume was mounted at least once. The conversion of
non mounted volumes will still fail since the dht selfheal tries to
set the directory layout will fail as they do that with the PID
GF_CLIENT_PID_NO_ROOT_SQUASH set in the frame->root.
Backport of: https://review.gluster.org/#/c/glusterfs/+/21791/
Change-Id: Ie31d429dfebbfb0f60610c9c5739595c54b19c46
BUG: 1645480
Signed-off-by: karthik-us <ksubrahm@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158932
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Ravishankar Narayanankutty <ravishankar@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
libglusterfs/src/common-utils.h | 3 +-
...g-1655854-support-dist-to-rep3-arb-conversion.t | 95 ++++++++++++++++++++++
xlators/cluster/afr/src/afr-common.c | 80 +++++++++++++-----
xlators/cluster/afr/src/afr-inode-write.c | 2 +-
xlators/cluster/afr/src/afr-read-txn.c | 3 +-
xlators/cluster/afr/src/afr-transaction.c | 11 ++-
xlators/cluster/afr/src/afr-transaction.h | 3 +-
xlators/cluster/afr/src/afr.c | 2 +-
xlators/cluster/afr/src/afr.h | 4 +
xlators/mgmt/glusterd/src/glusterd-utils.c | 2 +-
10 files changed, 175 insertions(+), 30 deletions(-)
create mode 100644 tests/bugs/replicate/bug-1655854-support-dist-to-rep3-arb-conversion.t
diff --git a/libglusterfs/src/common-utils.h b/libglusterfs/src/common-utils.h
index c804ed5..50c1f9a 100644
--- a/libglusterfs/src/common-utils.h
+++ b/libglusterfs/src/common-utils.h
@@ -162,7 +162,8 @@ enum _gf_special_pid
GF_CLIENT_PID_BITD = -8,
GF_CLIENT_PID_SCRUB = -9,
GF_CLIENT_PID_TIER_DEFRAG = -10,
- GF_SERVER_PID_TRASH = -11
+ GF_SERVER_PID_TRASH = -11,
+ GF_CLIENT_PID_ADD_REPLICA_MOUNT = -12
};
enum _gf_xlator_ipc_targets {
diff --git a/tests/bugs/replicate/bug-1655854-support-dist-to-rep3-arb-conversion.t b/tests/bugs/replicate/bug-1655854-support-dist-to-rep3-arb-conversion.t
new file mode 100644
index 0000000..783016d
--- /dev/null
+++ b/tests/bugs/replicate/bug-1655854-support-dist-to-rep3-arb-conversion.t
@@ -0,0 +1,95 @@
+#!/bin/bash
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+cleanup;
+
+TEST glusterd
+TEST pidof glusterd
+
+# Conversion from 2x1 to 2x3
+
+TEST $CLI volume create $V0 $H0:$B0/${V0}{0,1}
+EXPECT 'Created' volinfo_field $V0 'Status';
+TEST $CLI volume start $V0
+EXPECT 'Started' volinfo_field $V0 'Status';
+
+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
+
+TEST $GFS --volfile-id=$V0 --volfile-server=$H0 $M0;
+TEST mkdir $M0/dir
+TEST dd if=/dev/urandom of=$M0/dir/file bs=100K count=5
+file_md5sum=$(md5sum $M0/dir/file | awk '{print $1}')
+
+TEST $CLI volume add-brick $V0 replica 3 $H0:$B0/${V0}{2..5}
+
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V0 $H0 $B0/${V0}2
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V0 $H0 $B0/${V0}3
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V0 $H0 $B0/${V0}4
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V0 $H0 $B0/${V0}5
+
+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 $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 3
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 4
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 5
+
+# Trigger heal and wait for for it to complete
+TEST $CLI volume heal $V0
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" get_pending_heal_count $V0
+
+# Check whether the directory & file are healed to the newly added bricks
+TEST ls $B0/${V0}2/dir
+TEST ls $B0/${V0}3/dir
+TEST ls $B0/${V0}4/dir
+TEST ls $B0/${V0}5/dir
+
+TEST [ $file_md5sum == $(md5sum $B0/${V0}4/dir/file | awk '{print $1}') ]
+TEST [ $file_md5sum == $(md5sum $B0/${V0}5/dir/file | awk '{print $1}') ]
+
+
+# Conversion from 2x1 to 2x(2+1)
+
+TEST $CLI volume create $V1 $H0:$B0/${V1}{0,1}
+EXPECT 'Created' volinfo_field $V1 'Status';
+TEST $CLI volume start $V1
+EXPECT 'Started' volinfo_field $V1 'Status';
+
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V1 $H0 $B0/${V1}0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V1 $H0 $B0/${V1}1
+
+TEST $GFS --volfile-id=$V1 --volfile-server=$H0 $M1;
+TEST mkdir $M1/dir
+TEST dd if=/dev/urandom of=$M1/dir/file bs=100K count=5
+file_md5sum=$(md5sum $M1/dir/file | awk '{print $1}')
+
+TEST $CLI volume add-brick $V1 replica 3 arbiter 1 $H0:$B0/${V1}{2..5}
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V1 $H0 $B0/${V1}2
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V1 $H0 $B0/${V1}3
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V1 $H0 $B0/${V1}4
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V1 $H0 $B0/${V1}5
+
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "Y" glustershd_up_status
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V1 0
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V1 1
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V1 2
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V1 3
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V1 4
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V1 5
+
+# Trigger heal and wait for for it to complete
+TEST $CLI volume heal $V1
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" get_pending_heal_count $V1
+
+# Check whether the directory & file are healed to the newly added bricks
+TEST ls $B0/${V1}2/dir
+TEST ls $B0/${V1}3/dir
+TEST ls $B0/${V1}4/dir
+TEST ls $B0/${V1}5/dir
+
+EXPECT "0" stat -c %s $B0/${V1}5/dir/file
+TEST [ $file_md5sum == $(md5sum $B0/${V1}4/dir/file | awk '{print $1}') ]
+
+cleanup;
diff --git a/xlators/cluster/afr/src/afr-common.c b/xlators/cluster/afr/src/afr-common.c
index 231de9d..322dfbe 100644
--- a/xlators/cluster/afr/src/afr-common.c
+++ b/xlators/cluster/afr/src/afr-common.c
@@ -2265,7 +2265,7 @@ afr_attempt_readsubvol_set (call_frame_t *frame, xlator_t *this,
} else if (!priv->quorum_count) {
*read_subvol = afr_first_up_child (frame, this);
} else if (priv->quorum_count &&
- afr_has_quorum (data_readable, this)) {
+ afr_has_quorum (data_readable, this, NULL)) {
/* read_subvol is guaranteed to be valid if we hit this path. */
*read_subvol = afr_first_up_child (frame, this);
} else {
@@ -2405,7 +2405,7 @@ afr_lookup_done (call_frame_t *frame, xlator_t *this)
read_subvol = -1;
memset (readable, 0, sizeof (*readable) * priv->child_count);
if (can_interpret) {
- if (!afr_has_quorum (success_replies, this))
+ if (!afr_has_quorum (success_replies, this, NULL))
goto cant_interpret;
/* It is safe to call afr_replies_interpret() because we have
a response from all the UP subvolumes and all of them resolved
@@ -2887,7 +2887,7 @@ afr_lookup_entry_heal (call_frame_t *frame, xlator_t *this)
if (name_state_mismatch) {
if (!priv->quorum_count)
goto name_heal;
- if (!afr_has_quorum (success, this))
+ if (!afr_has_quorum (success, this, NULL))
goto name_heal;
if (op_errno)
goto name_heal;
@@ -2979,7 +2979,6 @@ afr_discover_done (call_frame_t *frame, xlator_t *this)
afr_private_t *priv = NULL;
afr_local_t *local = NULL;
int i = -1;
- int op_errno = 0;
int read_subvol = -1;
unsigned char *data_readable = NULL;
unsigned char *success_replies = NULL;
@@ -2998,15 +2997,13 @@ afr_discover_done (call_frame_t *frame, xlator_t *this)
}
}
- op_errno = afr_final_errno (frame->local, this->private);
-
if (local->op_ret < 0) {
- AFR_STACK_UNWIND (lookup, frame, -1, op_errno, NULL, NULL,
- NULL, NULL);
- return;
- }
+ local->op_ret = -1;
+ local->op_errno = afr_final_errno (frame->local, this->private);
+ goto error;
+ }
- if (!afr_has_quorum (success_replies, this))
+ if (!afr_has_quorum (success_replies, this, frame))
goto unwind;
afr_replies_interpret (frame, this, local->inode, NULL);
@@ -3017,11 +3014,8 @@ afr_discover_done (call_frame_t *frame, xlator_t *this)
unwind:
afr_attempt_readsubvol_set (frame, this, success_replies, data_readable,
&read_subvol);
- if (read_subvol == -1) {
- AFR_STACK_UNWIND (lookup, frame, local->op_ret, local->op_errno,
- NULL, NULL, NULL, NULL);
- return;
- }
+ if (read_subvol == -1)
+ goto error;
if (AFR_IS_ARBITER_BRICK (priv, read_subvol) && local->op_ret == 0) {
local->op_ret = -1;
@@ -3034,6 +3028,11 @@ unwind:
local->inode, &local->replies[read_subvol].poststat,
local->replies[read_subvol].xdata,
&local->replies[read_subvol].postparent);
+ return;
+
+error:
+ AFR_STACK_UNWIND (lookup, frame, local->op_ret, local->op_errno, NULL,
+ NULL, NULL, NULL);
}
@@ -3977,7 +3976,8 @@ afr_fop_lock_done (call_frame_t *frame, xlator_t *this)
if (afr_is_conflicting_lock_present (local->op_ret, local->op_errno)) {
afr_unlock_locks_and_proceed (frame, this, lock_count);
- } else if (priv->quorum_count && !afr_has_quorum (success, this)) {
+ } else if (priv->quorum_count &&
+ !afr_has_quorum (success, this, NULL)) {
local->fop_lock_state = AFR_FOP_LOCK_QUORUM_FAILED;
local->op_ret = -1;
local->op_errno = afr_final_errno (local, priv);
@@ -4485,7 +4485,7 @@ afr_lk_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
&local->cont.lk.user_flock,
local->xdata_req);
} else if (priv->quorum_count &&
- !afr_has_quorum (local->cont.lk.locked_nodes, this)) {
+ !afr_has_quorum (local->cont.lk.locked_nodes, this, NULL)) {
local->op_ret = -1;
local->op_errno = afr_final_errno (local, priv);
@@ -5199,7 +5199,7 @@ afr_notify (xlator_t *this, int32_t event,
}
had_quorum = priv->quorum_count && afr_has_quorum (priv->child_up,
- this);
+ this, NULL);
if (priv->halo_enabled) {
halo_max_latency_msec = afr_get_halo_latency (this);
@@ -5328,7 +5328,7 @@ afr_notify (xlator_t *this, int32_t event,
UNLOCK (&priv->lock);
if (priv->quorum_count) {
- has_quorum = afr_has_quorum (priv->child_up, this);
+ has_quorum = afr_has_quorum (priv->child_up, this, NULL);
if (!had_quorum && has_quorum) {
gf_msg (this->name, GF_LOG_INFO, 0, AFR_MSG_QUORUM_MET,
"Client-quorum is met");
@@ -6543,3 +6543,43 @@ afr_set_inode_local (xlator_t *this, afr_local_t *local, inode_t *inode)
}
return ret;
}
+
+gf_boolean_t
+afr_is_add_replica_mount_lookup_on_root (call_frame_t *frame)
+{
+ afr_local_t *local = NULL;
+
+ local = frame->local;
+
+ if (frame->root->pid != GF_CLIENT_PID_ADD_REPLICA_MOUNT)
+ return _gf_false;
+
+ if (local->op != GF_FOP_LOOKUP)
+ /* TODO:If the replica count is being increased on a plain
+ * distribute volume that was never mounted, we need to allow
+ * setxattr on '/' with GF_CLIENT_PID_NO_ROOT_SQUASH to
+ * accomodate for DHT layout setting */
+ return _gf_false;
+
+ if (local->inode == NULL)
+ return _gf_false;
+
+ if (!__is_root_gfid (local->inode->gfid))
+ return _gf_false;
+
+ return _gf_true;
+}
+
+gf_boolean_t
+afr_lookup_has_quorum (call_frame_t *frame, xlator_t *this,
+ unsigned char *subvols)
+{
+ afr_private_t *priv = this->private;
+
+ if (frame && afr_is_add_replica_mount_lookup_on_root (frame)) {
+ if (AFR_COUNT (subvols, priv->child_count) > 0)
+ return _gf_true;
+ }
+
+ return _gf_false;
+}
diff --git a/xlators/cluster/afr/src/afr-inode-write.c b/xlators/cluster/afr/src/afr-inode-write.c
index 8b1dcfd..ea4755a 100644
--- a/xlators/cluster/afr/src/afr-inode-write.c
+++ b/xlators/cluster/afr/src/afr-inode-write.c
@@ -1539,7 +1539,7 @@ afr_handle_empty_brick (xlator_t *this, call_frame_t *frame, loc_t *loc,
if (ret && ab_ret)
goto out;
- if (frame->root->pid != GF_CLIENT_PID_SELF_HEALD) {
+ if (frame->root->pid != GF_CLIENT_PID_ADD_REPLICA_MOUNT) {
gf_msg (this->name, GF_LOG_ERROR, EPERM,
afr_get_msg_id (op_type),
"'%s' is an internal extended attribute.",
diff --git a/xlators/cluster/afr/src/afr-read-txn.c b/xlators/cluster/afr/src/afr-read-txn.c
index f6c491b..ec322ae 100644
--- a/xlators/cluster/afr/src/afr-read-txn.c
+++ b/xlators/cluster/afr/src/afr-read-txn.c
@@ -193,7 +193,8 @@ afr_read_txn (call_frame_t *frame, xlator_t *this, inode_t *inode,
local->inode = inode_ref (inode);
local->is_read_txn = _gf_true;
- if (priv->quorum_count && !afr_has_quorum (local->child_up, this)) {
+ if (priv->quorum_count &&
+ !afr_has_quorum (local->child_up, this, NULL)) {
local->op_ret = -1;
local->op_errno = ENOTCONN;
read_subvol = -1;
diff --git a/xlators/cluster/afr/src/afr-transaction.c b/xlators/cluster/afr/src/afr-transaction.c
index 0a67a83..0e58e02 100644
--- a/xlators/cluster/afr/src/afr-transaction.c
+++ b/xlators/cluster/afr/src/afr-transaction.c
@@ -160,7 +160,7 @@ afr_changelog_has_quorum (afr_local_t *local, xlator_t *this)
}
}
- if (afr_has_quorum (success_children, this)) {
+ if (afr_has_quorum (success_children, this, NULL)) {
return _gf_true;
}
@@ -690,7 +690,7 @@ afr_handle_symmetric_errors (call_frame_t *frame, xlator_t *this)
}
gf_boolean_t
-afr_has_quorum (unsigned char *subvols, xlator_t *this)
+afr_has_quorum (unsigned char *subvols, xlator_t *this, call_frame_t *frame)
{
unsigned int quorum_count = 0;
afr_private_t *priv = NULL;
@@ -699,6 +699,9 @@ afr_has_quorum (unsigned char *subvols, xlator_t *this)
priv = this->private;
up_children_count = AFR_COUNT (subvols, priv->child_count);
+ if (afr_lookup_has_quorum (frame, this, subvols))
+ return _gf_true;
+
if (priv->quorum_count == AFR_QUORUM_AUTO) {
/*
* Special case for auto-quorum with an even number of nodes.
@@ -753,7 +756,7 @@ afr_has_fop_quorum (call_frame_t *frame)
locked_nodes = afr_locked_nodes_get (local->transaction.type,
&local->internal_lock);
- return afr_has_quorum (locked_nodes, this);
+ return afr_has_quorum (locked_nodes, this, NULL);
}
static gf_boolean_t
@@ -771,7 +774,7 @@ afr_has_fop_cbk_quorum (call_frame_t *frame)
success[i] = 1;
}
- return afr_has_quorum (success, this);
+ return afr_has_quorum (success, this, NULL);
}
gf_boolean_t
diff --git a/xlators/cluster/afr/src/afr-transaction.h b/xlators/cluster/afr/src/afr-transaction.h
index a27e9a3..1c42c66 100644
--- a/xlators/cluster/afr/src/afr-transaction.h
+++ b/xlators/cluster/afr/src/afr-transaction.h
@@ -38,7 +38,8 @@ int afr_read_txn (call_frame_t *frame, xlator_t *this, inode_t *inode,
int afr_read_txn_continue (call_frame_t *frame, xlator_t *this, int subvol);
call_frame_t *afr_transaction_detach_fop_frame (call_frame_t *frame);
-gf_boolean_t afr_has_quorum (unsigned char *subvols, xlator_t *this);
+gf_boolean_t afr_has_quorum (unsigned char *subvols, xlator_t *this,
+ call_frame_t *frame);
gf_boolean_t afr_needs_changelog_update (afr_local_t *local);
void afr_zero_fill_stat (afr_local_t *local);
diff --git a/xlators/cluster/afr/src/afr.c b/xlators/cluster/afr/src/afr.c
index 1b738c0..d5347e9 100644
--- a/xlators/cluster/afr/src/afr.c
+++ b/xlators/cluster/afr/src/afr.c
@@ -244,7 +244,7 @@ reconfigure (xlator_t *this, dict_t *options)
GF_OPTION_RECONF ("quorum-count", priv->quorum_count, options,
uint32, out);
fix_quorum_options (this, priv, qtype, options);
- if (priv->quorum_count && !afr_has_quorum (priv->child_up, this))
+ if (priv->quorum_count && !afr_has_quorum (priv->child_up, this, NULL))
gf_msg (this->name, GF_LOG_WARNING, 0, AFR_MSG_QUORUM_FAIL,
"Client-quorum is not met");
diff --git a/xlators/cluster/afr/src/afr.h b/xlators/cluster/afr/src/afr.h
index 7010e9b..febc509 100644
--- a/xlators/cluster/afr/src/afr.h
+++ b/xlators/cluster/afr/src/afr.h
@@ -1227,4 +1227,8 @@ afr_is_symmetric_error (call_frame_t *frame, xlator_t *this);
gf_boolean_t
afr_is_pending_set (xlator_t *this, dict_t *xdata, int type);
+
+gf_boolean_t
+afr_lookup_has_quorum (call_frame_t *frame, xlator_t *this,
+ unsigned char *subvols);
#endif /* __AFR_H__ */
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index 2290343..6468ecb 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -13788,7 +13788,7 @@ glusterd_handle_replicate_brick_ops (glusterd_volinfo_t *volinfo,
goto out;
}
- ret = gf_asprintf (&pid, "%d", GF_CLIENT_PID_SELF_HEALD);
+ ret = gf_asprintf (&pid, "%d", GF_CLIENT_PID_ADD_REPLICA_MOUNT);
if (ret < 0)
goto out;
--
1.8.3.1

91
0489-cluster-afr-Do-not-update-read_subvol-in-inode_ctx-a.patch Normal file
View File

@ -0,0 +1,91 @@
From adeeec1fbe7241d18903ae3830a1fabb1061be21 Mon Sep 17 00:00:00 2001
From: karthik-us <ksubrahm@redhat.com>
Date: Mon, 17 Dec 2018 14:28:07 +0530
Subject: [PATCH 489/493] cluster/afr: Do not update read_subvol in inode_ctx
after rename/link fop
Since rename/link fops on a file will not change any data in it, it should
not update the read_subvol values in the inode_ctx, which interprets the
data & metadata readable subvols for that file. The old read_subvol values
should be retained even after the rename/link operations.
Backport of: https://review.gluster.org/#/c/glusterfs/+/21841
Change-Id: I22787938b3f8295a48162ab4f498e4010d66a1ab
BUG: 1645480
Signed-off-by: karthik-us <ksubrahm@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158803
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Tested-by: Ravishankar Narayanankutty <ravishankar@redhat.com>
Reviewed-by: Ravishankar Narayanankutty <ravishankar@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
...7783-do-not-update-read-subvol-on-rename-link.t | 40 ++++++++++++++++++++++
xlators/cluster/afr/src/afr-dir-write.c | 4 ++-
2 files changed, 43 insertions(+), 1 deletion(-)
create mode 100644 tests/bugs/replicate/bug-1657783-do-not-update-read-subvol-on-rename-link.t
diff --git a/tests/bugs/replicate/bug-1657783-do-not-update-read-subvol-on-rename-link.t b/tests/bugs/replicate/bug-1657783-do-not-update-read-subvol-on-rename-link.t
new file mode 100644
index 0000000..b180f0e
--- /dev/null
+++ b/tests/bugs/replicate/bug-1657783-do-not-update-read-subvol-on-rename-link.t
@@ -0,0 +1,40 @@
+#!/bin/bash
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+cleanup;
+
+TEST glusterd
+TEST pidof glusterd
+TEST $CLI volume create $V0 replica 3 $H0:$B0/${V0}{0..2}
+TEST $CLI volume set $V0 self-heal-daemon off
+TEST $CLI volume set $V0 cluster.data-self-heal off
+TEST $CLI volume set $V0 cluster.metadata-self-heal off
+TEST $CLI volume set $V0 cluster.entry-self-heal off
+TEST $CLI volume set $V0 performance.write-behind off
+
+TEST $CLI volume start $V0
+TEST $GFS --volfile-id=$V0 --volfile-server=$H0 $M0;
+TEST mkdir $M0/dir
+TEST "echo abc > $M0/file1"
+TEST "echo uvw > $M0/file2"
+
+TEST kill_brick $V0 $H0 $B0/${V0}0
+TEST "echo def > $M0/file1"
+TEST "echo xyz > $M0/file2"
+
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V0 $H0 $B0/${V0}0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 0
+
+TEST kill_brick $V0 $H0 $B0/${V0}1
+
+# Rename file1 and read it. Read should be served from the 3rd brick
+TEST mv $M0/file1 $M0/file3
+EXPECT "def" cat $M0/file3
+
+# Create a link to file2 and read it. Read should be served from the 3rd brick
+TEST ln $M0/file2 $M0/dir/file4
+EXPECT "xyz" cat $M0/dir/file4
+EXPECT "xyz" cat $M0/file2
+
+cleanup
diff --git a/xlators/cluster/afr/src/afr-dir-write.c b/xlators/cluster/afr/src/afr-dir-write.c
index 75889de..c508034 100644
--- a/xlators/cluster/afr/src/afr-dir-write.c
+++ b/xlators/cluster/afr/src/afr-dir-write.c
@@ -99,7 +99,9 @@ __afr_dir_write_finalize (call_frame_t *frame, xlator_t *this)
}
if (local->inode) {
- afr_replies_interpret (frame, this, local->inode, NULL);
+ if (local->op != GF_FOP_RENAME && local->op != GF_FOP_LINK)
+ afr_replies_interpret (frame, this, local->inode, NULL);
+
inode_read_subvol = afr_data_subvol_get (local->inode, this,
NULL, NULL, NULL, &args);
}
--
1.8.3.1

921
0490-glusterd-migrating-rebalance-commands-to-mgmt_v3-fra.patch Normal file
View File

@ -0,0 +1,921 @@
From 783c36e573a9c937422e63af038bb35648483b9e Mon Sep 17 00:00:00 2001
From: Sanju Rakonde <srakonde@redhat.com>
Date: Fri, 30 Nov 2018 16:16:55 +0530
Subject: [PATCH 490/493] glusterd: migrating rebalance commands to mgmt_v3
framework
Current rebalance commands use the op_state machine framework.
Porting it to use the mgmt_v3 framework.
> Change-Id: I6faf4a6335c2e2f3d54bbde79908a7749e4613e7
> fixes: bz#1655827
> Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
upstream patch: https://review.gluster.org/#/c/glusterfs/+/21762/
Change-Id: I6faf4a6335c2e2f3d54bbde79908a7749e4613e7
BUG: 1652466
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158917
Reviewed-by: Atin Mukherjee <amukherj@redhat.com>
Tested-by: RHGS Build Bot <nigelb@redhat.com>
---
libglusterfs/src/globals.h | 2 +
xlators/mgmt/glusterd/src/glusterd-handler.c | 4 +-
xlators/mgmt/glusterd/src/glusterd-mgmt.c | 128 ++++++-
xlators/mgmt/glusterd/src/glusterd-mgmt.h | 5 +-
xlators/mgmt/glusterd/src/glusterd-op-sm.h | 3 +
xlators/mgmt/glusterd/src/glusterd-rebalance.c | 495 ++++++++++++++++++++++++-
xlators/mgmt/glusterd/src/glusterd-syncop.c | 9 +
xlators/mgmt/glusterd/src/glusterd-utils.c | 4 +-
xlators/mgmt/glusterd/src/glusterd.h | 9 +
9 files changed, 637 insertions(+), 22 deletions(-)
diff --git a/libglusterfs/src/globals.h b/libglusterfs/src/globals.h
index 343263c..5e3b180 100644
--- a/libglusterfs/src/globals.h
+++ b/libglusterfs/src/globals.h
@@ -111,6 +111,8 @@
#define GD_OP_VERSION_3_13_3 31303 /* Op-version for GlusterFS 3.13.3 */
+#define GD_OP_VERSION_6_0 60000 /* Op-version for GlusterFS 6.0 */
+
/* Downstream only change */
#define GD_OP_VERSION_3_11_2 31102 /* Op-version for RHGS 3.3.1-async */
#define GD_OP_VERSION_3_13_3 31303 /* Op-version for RHGS-3.4-Batch Update-1*/
diff --git a/xlators/mgmt/glusterd/src/glusterd-handler.c b/xlators/mgmt/glusterd/src/glusterd-handler.c
index d40de89..d8e3335 100644
--- a/xlators/mgmt/glusterd/src/glusterd-handler.c
+++ b/xlators/mgmt/glusterd/src/glusterd-handler.c
@@ -3120,7 +3120,9 @@ __glusterd_handle_cli_profile_volume (rpcsvc_request_t *req)
glusterd_friend_sm();
glusterd_op_sm();
} else {
- ret = glusterd_mgmt_v3_initiate_profile_phases(req, cli_op, dict);
+ ret = glusterd_mgmt_v3_initiate_all_phases_with_brickop_phase(req,
+ cli_op,
+ dict);
}
out:
diff --git a/xlators/mgmt/glusterd/src/glusterd-mgmt.c b/xlators/mgmt/glusterd/src/glusterd-mgmt.c
index d98c6bc..ef8a2d9 100644
--- a/xlators/mgmt/glusterd/src/glusterd-mgmt.c
+++ b/xlators/mgmt/glusterd/src/glusterd-mgmt.c
@@ -224,6 +224,16 @@ gd_mgmt_v3_pre_validate_fn (glusterd_op_t op, dict_t *dict,
}
break;
+ case GD_OP_REBALANCE:
+ case GD_OP_DEFRAG_BRICK_VOLUME:
+ ret = glusterd_mgmt_v3_op_stage_rebalance(dict, op_errstr);
+ if (ret) {
+ gf_log(this->name, GF_LOG_WARNING,
+ "Rebalance Prevalidate Failed");
+ goto out;
+ }
+ break;
+
case GD_OP_MAX_OPVERSION:
ret = 0;
break;
@@ -264,6 +274,8 @@ gd_mgmt_v3_brick_op_fn (glusterd_op_t op, dict_t *dict,
break;
}
case GD_OP_PROFILE_VOLUME:
+ case GD_OP_REBALANCE:
+ case GD_OP_DEFRAG_BRICK_VOLUME:
{
ret = gd_brick_op_phase(op, rsp_dict, dict, op_errstr);
if (ret) {
@@ -438,6 +450,19 @@ gd_mgmt_v3_commit_fn (glusterd_op_t op, dict_t *dict,
}
break;
}
+ case GD_OP_REBALANCE:
+ case GD_OP_DEFRAG_BRICK_VOLUME:
+ {
+ ret = glusterd_mgmt_v3_op_rebalance(dict, op_errstr,
+ rsp_dict);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0,
+ GD_MSG_COMMIT_OP_FAIL,
+ "Rebalance Commit Failed");
+ goto out;
+ }
+ break;
+ }
default:
break;
@@ -880,6 +905,8 @@ glusterd_pre_validate_aggr_rsp_dict (glusterd_op_t op,
case GD_OP_TIER_START_STOP:
case GD_OP_REMOVE_TIER_BRICK:
case GD_OP_PROFILE_VOLUME:
+ case GD_OP_DEFRAG_BRICK_VOLUME:
+ case GD_OP_REBALANCE:
break;
case GD_OP_MAX_OPVERSION:
break;
@@ -1197,6 +1224,7 @@ glusterd_mgmt_v3_build_payload (dict_t **req, char **op_errstr, dict_t *dict,
break;
case GD_OP_START_VOLUME:
case GD_OP_ADD_BRICK:
+ case GD_OP_DEFRAG_BRICK_VOLUME:
case GD_OP_REPLACE_BRICK:
case GD_OP_RESET_BRICK:
case GD_OP_ADD_TIER_BRICK:
@@ -1221,6 +1249,30 @@ glusterd_mgmt_v3_build_payload (dict_t **req, char **op_errstr, dict_t *dict,
dict_copy (dict, req_dict);
}
break;
+
+ case GD_OP_REBALANCE: {
+ if (gd_set_commit_hash(dict) != 0) {
+ ret = -1;
+ goto out;
+ }
+ ret = dict_get_str (dict, "volname", &volname);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_CRITICAL, errno,
+ GD_MSG_DICT_GET_FAILED,
+ "volname is not present in "
+ "operation ctx");
+ goto out;
+ }
+
+ if (strcasecmp(volname, "all")) {
+ ret = glusterd_dict_set_volid(dict, volname, op_errstr);
+ if (ret)
+ goto out;
+ }
+ dict_copy(dict, req_dict);
+ }
+ break;
+
case GD_OP_TIER_START_STOP:
case GD_OP_REMOVE_TIER_BRICK:
case GD_OP_DETACH_TIER_STATUS:
@@ -1247,6 +1299,7 @@ gd_mgmt_v3_brick_op_cbk_fn (struct rpc_req *req, struct iovec *iov,
call_frame_t *frame = NULL;
int32_t op_ret = -1;
int32_t op_errno = -1;
+ dict_t *rsp_dict = NULL;
xlator_t *this = NULL;
uuid_t *peerid = NULL;
@@ -1278,10 +1331,45 @@ gd_mgmt_v3_brick_op_cbk_fn (struct rpc_req *req, struct iovec *iov,
if (ret < 0)
goto out;
+ if (rsp.dict.dict_len) {
+ /* Unserialize the dictionary */
+ rsp_dict = dict_new ();
+
+ ret = dict_unserialize (rsp.dict.dict_val,
+ rsp.dict.dict_len,
+ &rsp_dict);
+ if (ret < 0) {
+ free (rsp.dict.dict_val);
+ goto out;
+ } else {
+ rsp_dict->extra_stdfree = rsp.dict.dict_val;
+ }
+ }
+
gf_uuid_copy (args->uuid, rsp.uuid);
+ pthread_mutex_lock (&args->lock_dict);
+ {
+ if (rsp.op == GD_OP_DEFRAG_BRICK_VOLUME)
+ ret = glusterd_syncop_aggr_rsp_dict (rsp.op, args->dict,
+ rsp_dict);
+ }
+ pthread_mutex_unlock (&args->lock_dict);
- op_ret = rsp.op_ret;
- op_errno = rsp.op_errno;
+ if (ret) {
+ gf_msg (this->name, GF_LOG_ERROR, 0,
+ GD_MSG_RESP_AGGR_FAIL, "%s",
+ "Failed to aggregate response from "
+ " node/brick");
+ if (!rsp.op_ret)
+ op_ret = ret;
+ else {
+ op_ret = rsp.op_ret;
+ op_errno = rsp.op_errno;
+ }
+ } else {
+ op_ret = rsp.op_ret;
+ op_errno = rsp.op_errno;
+ }
out:
gd_mgmt_v3_collate_errors (args, op_ret, op_errno, rsp.op_errstr,
@@ -1353,11 +1441,12 @@ out:
}
int
-glusterd_mgmt_v3_brick_op (glusterd_op_t op, dict_t *rsp_dict, dict_t *req_dict,
+glusterd_mgmt_v3_brick_op (glusterd_op_t op, dict_t *op_ctx, dict_t *req_dict,
char **op_errstr, uint32_t txn_generation)
{
int32_t ret = -1;
int32_t peer_cnt = 0;
+ dict_t *rsp_dict = NULL;
glusterd_peerinfo_t *peerinfo = NULL;
struct syncargs args = {0};
uuid_t peer_uuid = {0};
@@ -1372,6 +1461,13 @@ glusterd_mgmt_v3_brick_op (glusterd_op_t op, dict_t *rsp_dict, dict_t *req_dict,
GF_ASSERT (req_dict);
GF_ASSERT (op_errstr);
+ rsp_dict = dict_new();
+ if (!rsp_dict) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_CREATE_FAIL,
+ "Failed to create response dictionary");
+ goto out;
+ }
+
/* Perform brick op on local node */
ret = gd_mgmt_v3_brick_op_fn (op, req_dict, op_errstr,
rsp_dict);
@@ -1395,9 +1491,21 @@ glusterd_mgmt_v3_brick_op (glusterd_op_t op, dict_t *rsp_dict, dict_t *req_dict,
}
goto out;
}
+ if (op == GD_OP_DEFRAG_BRICK_VOLUME || op == GD_OP_PROFILE_VOLUME) {
+ ret = glusterd_syncop_aggr_rsp_dict(op, op_ctx, rsp_dict);
+ if (ret) {
+ gf_log(this->name, GF_LOG_ERROR, "%s",
+ "Failed to aggregate response from "
+ " node/brick");
+ goto out;
+ }
+ }
+
+ dict_unref(rsp_dict);
+ rsp_dict = NULL;
/* Sending brick op req to other nodes in the cluster */
- gd_syncargs_init (&args, rsp_dict);
+ gd_syncargs_init (&args, op_ctx);
synctask_barrier_init((&args));
peer_cnt = 0;
@@ -1616,6 +1724,13 @@ glusterd_mgmt_v3_commit (glusterd_op_t op, dict_t *op_ctx, dict_t *req_dict,
GF_ASSERT (op_errstr);
GF_VALIDATE_OR_GOTO (this->name, op_errno, out);
+ if (op == GD_OP_REBALANCE || op == GD_OP_DEFRAG_BRICK_VOLUME) {
+ ret = glusterd_set_rebalance_id_in_rsp_dict(req_dict, op_ctx);
+ if (ret) {
+ gf_log(this->name, GF_LOG_WARNING,
+ "Failed to set rebalance id in dict.");
+ }
+ }
rsp_dict = dict_new ();
if (!rsp_dict) {
gf_msg (this->name, GF_LOG_ERROR, 0,
@@ -2140,8 +2255,9 @@ out:
}
int32_t
-glusterd_mgmt_v3_initiate_profile_phases (rpcsvc_request_t *req,
- glusterd_op_t op, dict_t *dict)
+glusterd_mgmt_v3_initiate_all_phases_with_brickop_phase (rpcsvc_request_t *req,
+ glusterd_op_t op,
+ dict_t *dict)
{
int32_t ret = -1;
int32_t op_ret = -1;
diff --git a/xlators/mgmt/glusterd/src/glusterd-mgmt.h b/xlators/mgmt/glusterd/src/glusterd-mgmt.h
index eff070d..ef0fe10 100644
--- a/xlators/mgmt/glusterd/src/glusterd-mgmt.h
+++ b/xlators/mgmt/glusterd/src/glusterd-mgmt.h
@@ -37,8 +37,9 @@ glusterd_mgmt_v3_initiate_all_phases (rpcsvc_request_t *req, glusterd_op_t op,
dict_t *dict);
int32_t
-glusterd_mgmt_v3_initiate_profile_phases(rpcsvc_request_t *req,
- glusterd_op_t op, dict_t *dict);
+glusterd_mgmt_v3_initiate_all_phases_with_brickop_phase(rpcsvc_request_t *req,
+ glusterd_op_t op,
+ dict_t *dict);
int32_t
glusterd_mgmt_v3_initiate_snap_phases(rpcsvc_request_t *req, glusterd_op_t op,
diff --git a/xlators/mgmt/glusterd/src/glusterd-op-sm.h b/xlators/mgmt/glusterd/src/glusterd-op-sm.h
index e64d368..cf1e61c 100644
--- a/xlators/mgmt/glusterd/src/glusterd-op-sm.h
+++ b/xlators/mgmt/glusterd/src/glusterd-op-sm.h
@@ -318,4 +318,7 @@ glusterd_op_stats_volume (dict_t *dict, char **op_errstr, dict_t *rsp_dict);
int
glusterd_op_stage_stats_volume (dict_t *dict, char **op_errstr);
+
+int
+gd_set_commit_hash(dict_t *dict);
#endif
diff --git a/xlators/mgmt/glusterd/src/glusterd-rebalance.c b/xlators/mgmt/glusterd/src/glusterd-rebalance.c
index 5ab828c..7ba5f65 100644
--- a/xlators/mgmt/glusterd/src/glusterd-rebalance.c
+++ b/xlators/mgmt/glusterd/src/glusterd-rebalance.c
@@ -23,6 +23,7 @@
#include "glusterd.h"
#include "glusterd-sm.h"
#include "glusterd-op-sm.h"
+#include "glusterd-mgmt.h"
#include "glusterd-utils.h"
#include "glusterd-messages.h"
#include "glusterd-store.h"
@@ -501,6 +502,7 @@ __glusterd_handle_defrag_volume (rpcsvc_request_t *req)
int32_t ret = -1;
gf_cli_req cli_req = {{0,}};
glusterd_conf_t *priv = NULL;
+ int32_t op = GD_OP_NONE;
dict_t *dict = NULL;
char *volname = NULL;
gf_cli_defrag_type cmd = 0;
@@ -563,17 +565,25 @@ __glusterd_handle_defrag_volume (rpcsvc_request_t *req)
(cmd == GF_DEFRAG_CMD_STOP_DETACH_TIER) ||
(cmd == GF_DEFRAG_CMD_STOP) ||
(cmd == GF_DEFRAG_CMD_DETACH_STATUS)) {
- ret = glusterd_op_begin (req, GD_OP_DEFRAG_BRICK_VOLUME,
- dict, msg, sizeof (msg));
+ op = GD_OP_DEFRAG_BRICK_VOLUME;
} else
- ret = glusterd_op_begin (req, GD_OP_REBALANCE, dict,
- msg, sizeof (msg));
-
+ op = GD_OP_REBALANCE;
+
+ if (priv->op_version < GD_OP_VERSION_6_0) {
+ gf_msg_debug(this->name, 0,
+ "The cluster is operating at "
+ "version less than %d. Falling back "
+ "to op-sm framework.",
+ GD_OP_VERSION_6_0);
+ ret = glusterd_op_begin(req, op, dict, msg, sizeof(msg));
+ glusterd_friend_sm();
+ glusterd_op_sm();
+ } else {
+ ret = glusterd_mgmt_v3_initiate_all_phases_with_brickop_phase(req,
+ op,
+ dict);
+ }
out:
-
- glusterd_friend_sm ();
- glusterd_op_sm ();
-
if (ret) {
if (msg[0] == '\0')
snprintf (msg, sizeof (msg), "Operation failed");
@@ -583,8 +593,8 @@ out:
}
free (cli_req.dict.dict_val);//malloced by xdr
-
- return 0;
+ gf_msg_debug(this->name, 0, "Returning %d", ret);
+ return ret;
}
int
@@ -628,6 +638,469 @@ glusterd_brick_validation (dict_t *dict, char *key, data_t *value,
}
int
+glusterd_set_rebalance_id_in_rsp_dict(dict_t *req_dict, dict_t *rsp_dict)
+{
+ int ret = -1;
+ int32_t cmd = 0;
+ char *volname = NULL;
+ glusterd_volinfo_t *volinfo = NULL;
+ char msg[2048] = {0};
+ char *task_id_str = NULL;
+ xlator_t *this = NULL;
+
+ this = THIS;
+ GF_ASSERT(this);
+
+ GF_ASSERT(rsp_dict);
+ GF_ASSERT(req_dict);
+
+ ret = dict_get_str(rsp_dict, "volname", &volname);
+ if (ret) {
+ gf_msg_debug(this->name, 0, "volname not found");
+ goto out;
+ }
+
+ ret = dict_get_int32(rsp_dict, "rebalance-command", &cmd);
+ if (ret) {
+ gf_msg_debug(this->name, 0, "cmd not found");
+ goto out;
+ }
+
+ ret = glusterd_rebalance_cmd_validate(cmd, volname, &volinfo, msg,
+ sizeof(msg));
+ if (ret) {
+ gf_msg_debug(this->name, 0, "failed to validate");
+ goto out;
+ }
+
+ /* reblance id is generted in glusterd_mgmt_v3_op_stage_rebalance(), but
+ * rsp_dict is unavailable there. So copying it to rsp_dict from req_dict
+ * here. So that cli can display the rebalance id.*/
+ if ((cmd == GF_DEFRAG_CMD_START) ||
+ (cmd == GF_DEFRAG_CMD_START_LAYOUT_FIX) ||
+ (cmd == GF_DEFRAG_CMD_START_FORCE) ||
+ (cmd == GF_DEFRAG_CMD_START_TIER)) {
+ if (is_origin_glusterd(rsp_dict)) {
+ ret = dict_get_str(req_dict, GF_REBALANCE_TID_KEY,
+ &task_id_str);
+ if (ret) {
+ snprintf(msg, sizeof(msg), "Missing rebalance-id");
+ gf_msg(this->name, GF_LOG_WARNING, 0,
+ GD_MSG_REBALANCE_ID_MISSING, "%s", msg);
+ ret = 0;
+ } else {
+ gf_uuid_parse(task_id_str,
+ volinfo->rebal.rebalance_id);
+ ret = glusterd_copy_uuid_to_dict(
+ volinfo->rebal.rebalance_id, rsp_dict,
+ GF_REBALANCE_TID_KEY);
+ if (ret) {
+ snprintf(msg, sizeof(msg),
+ "Failed to set rebalance id for volume %s",
+ volname);
+ gf_msg(this->name, GF_LOG_WARNING, 0,
+ GD_MSG_DICT_SET_FAILED, "%s",
+ msg);
+ }
+ }
+ }
+ }
+
+ /* Set task-id, if available, in rsp_dict for operations other than
+ * start. This is needed when we want rebalance id in xml output
+ */
+ if (cmd == GF_DEFRAG_CMD_STATUS || cmd == GF_DEFRAG_CMD_STOP ||
+ cmd == GF_DEFRAG_CMD_STATUS_TIER) {
+ if (!gf_uuid_is_null(volinfo->rebal.rebalance_id)) {
+ if (GD_OP_REMOVE_BRICK == volinfo->rebal.op)
+ ret = glusterd_copy_uuid_to_dict(
+ volinfo->rebal.rebalance_id, rsp_dict,
+ GF_REMOVE_BRICK_TID_KEY);
+ else
+ ret = glusterd_copy_uuid_to_dict(
+ volinfo->rebal.rebalance_id,
+ rsp_dict, GF_REBALANCE_TID_KEY);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0,
+ GD_MSG_DICT_SET_FAILED,
+ "Failed to set task-id for volume %s",
+ volname);
+ goto out;
+ }
+ }
+ }
+out:
+ return ret;
+}
+
+int
+glusterd_mgmt_v3_op_stage_rebalance(dict_t *dict, char **op_errstr)
+{
+ char *volname = NULL;
+ char *cmd_str = NULL;
+ int ret = 0;
+ int32_t cmd = 0;
+ char msg[2048] = {0};
+ glusterd_volinfo_t *volinfo = NULL;
+ char *task_id_str = NULL;
+ xlator_t *this = 0;
+ int32_t is_force = 0;
+
+ this = THIS;
+ GF_ASSERT(this);
+
+ ret = dict_get_str (dict, "volname", &volname);
+ if (ret) {
+ gf_msg_debug(this->name, 0, "volname not found");
+ goto out;
+ }
+
+ ret = dict_get_int32 (dict, "rebalance-command", &cmd);
+ if (ret) {
+ gf_msg_debug(this->name, 0, "cmd not found");
+ goto out;
+ }
+
+ ret = glusterd_rebalance_cmd_validate(cmd, volname, &volinfo, msg,
+ sizeof(msg));
+ if (ret) {
+ gf_msg_debug(this->name, 0, "failed to validate");
+ goto out;
+ }
+ switch (cmd) {
+ case GF_DEFRAG_CMD_START_TIER:
+ ret = dict_get_int32 (dict, "force", &is_force);
+ if (ret)
+ is_force = 0;
+
+ if (volinfo->type != GF_CLUSTER_TYPE_TIER) {
+ gf_asprintf(op_errstr,
+ "volume %s is not a tier "
+ "volume.",
+ volinfo->volname);
+ ret = -1;
+ goto out;
+ }
+ if ((!is_force) && glusterd_is_tier_daemon_running(volinfo)) {
+ ret = gf_asprintf(op_errstr,
+ "A Tier daemon is "
+ "already running on volume %s",
+ volname);
+ ret = -1;
+ goto out;
+ }
+ /* Fall through */
+ case GF_DEFRAG_CMD_START:
+ case GF_DEFRAG_CMD_START_LAYOUT_FIX:
+ /* Check if the connected clients are all of version
+ * glusterfs-3.6 and higher. This is needed to prevent some data
+ * loss issues that could occur when older clients are connected
+ * when rebalance is run. This check can be bypassed by using
+ * 'force'
+ */
+ ret = glusterd_check_client_op_version_support(volname,
+ GD_OP_VERSION_RHS_3_0,
+ NULL);
+ if (ret) {
+ ret = gf_asprintf(op_errstr,
+ "Volume %s has one or "
+ "more connected clients of a version"
+ " lower than GlusterFS-v3.6.0. "
+ "Starting rebalance in this state "
+ "could lead to data loss.\nPlease "
+ "disconnect those clients before "
+ "attempting this command again.",
+ volname);
+ goto out;
+ }
+ /* Fall through */
+ case GF_DEFRAG_CMD_START_FORCE:
+ if (is_origin_glusterd(dict)) {
+ ret = glusterd_generate_and_set_task_id(dict,
+ GF_REBALANCE_TID_KEY);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_TASKID_GEN_FAIL,
+ "Failed to generate task-id");
+ goto out;
+ }
+ } else {
+ ret = dict_get_str(dict, GF_REBALANCE_TID_KEY, &task_id_str);
+ if (ret) {
+ snprintf(msg, sizeof(msg), "Missing rebalance-id");
+ gf_msg(this->name, GF_LOG_WARNING, 0,
+ GD_MSG_REBALANCE_ID_MISSING, "%s", msg);
+ ret = 0;
+ }
+ }
+ ret = glusterd_defrag_start_validate(volinfo, msg, sizeof(msg),
+ GD_OP_REBALANCE);
+ if (ret) {
+ gf_msg_debug(this->name, 0,
+ "defrag start validate "
+ "failed for volume %s.",
+ volinfo->volname);
+ goto out;
+ }
+ break;
+ case GF_DEFRAG_CMD_STATUS_TIER:
+ case GF_DEFRAG_CMD_STATUS:
+ case GF_DEFRAG_CMD_STOP:
+
+ ret = dict_get_str(dict, "cmd-str", &cmd_str);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, GD_MSG_DICT_GET_FAILED,
+ "Failed to get "
+ "command string");
+ ret = -1;
+ goto out;
+ }
+ if ((strstr(cmd_str, "rebalance") != NULL) &&
+ (volinfo->rebal.op != GD_OP_REBALANCE)) {
+ snprintf(msg, sizeof(msg),
+ "Rebalance not started "
+ "for volume %s.", volinfo->volname);
+ ret = -1;
+ goto out;
+ }
+
+ if (strstr(cmd_str, "remove-brick") != NULL) {
+ if (volinfo->rebal.op != GD_OP_REMOVE_BRICK) {
+ snprintf(msg, sizeof(msg),
+ "remove-brick not "
+ "started for volume %s.",
+ volinfo->volname);
+ ret = -1;
+ goto out;
+ }
+
+ /* For remove-brick status/stop command check whether
+ * given input brick is part of volume or not.*/
+
+ ret = dict_foreach_fnmatch(dict, "brick*",
+ glusterd_brick_validation, volinfo);
+ if (ret == -1) {
+ snprintf(msg, sizeof(msg),
+ "Incorrect brick for volume %s",
+ volinfo->volname);
+ goto out;
+ }
+ }
+ if (cmd == GF_DEFRAG_CMD_STATUS_TIER) {
+ if (volinfo->type != GF_CLUSTER_TYPE_TIER) {
+ snprintf(msg, sizeof(msg),
+ "volume %s is not "
+ "a tier volume.",
+ volinfo->volname);
+ ret = -1;
+ goto out;
+ }
+ }
+
+ break;
+
+ case GF_DEFRAG_CMD_STOP_DETACH_TIER:
+ case GF_DEFRAG_CMD_DETACH_STATUS:
+ if (volinfo->type != GF_CLUSTER_TYPE_TIER) {
+ snprintf(msg, sizeof(msg),
+ "volume %s is not "
+ "a tier volume.",
+ volinfo->volname);
+ ret = -1;
+ goto out;
+ }
+
+ if (volinfo->rebal.op != GD_OP_REMOVE_BRICK) {
+ snprintf(msg, sizeof(msg),
+ "Detach-tier "
+ "not started");
+ ret = -1;
+ goto out;
+ }
+ break;
+ default:
+ break;
+ }
+
+ ret = 0;
+out:
+ if (ret && op_errstr && msg[0])
+ *op_errstr = gf_strdup(msg);
+
+ return ret;
+}
+
+int
+glusterd_mgmt_v3_op_rebalance(dict_t *dict, char **op_errstr, dict_t *rsp_dict)
+{
+ char *volname = NULL;
+ int ret = 0;
+ int32_t cmd = 0;
+ char msg[2048] = {0};
+ glusterd_volinfo_t *volinfo = NULL;
+ glusterd_brickinfo_t *brickinfo = NULL;
+ glusterd_brickinfo_t *tmp = NULL;
+ gf_boolean_t volfile_update = _gf_false;
+ char *task_id_str = NULL;
+ xlator_t *this = NULL;
+ uint32_t commit_hash;
+ int32_t is_force = 0;
+
+ this = THIS;
+ GF_ASSERT(this);
+
+ ret = dict_get_str(dict, "volname", &volname);
+ if (ret) {
+ gf_msg_debug(this->name, 0, "volname not given");
+ goto out;
+ }
+
+ ret = dict_get_int32(dict, "rebalance-command", &cmd);
+ if (ret) {
+ gf_msg_debug(this->name, 0, "command not given");
+ goto out;
+ }
+
+ ret = glusterd_rebalance_cmd_validate(cmd, volname, &volinfo, msg,
+ sizeof(msg));
+ if (ret) {
+ gf_msg_debug(this->name, 0, "cmd validate failed");
+ goto out;
+ }
+
+ switch (cmd) {
+ case GF_DEFRAG_CMD_START:
+ case GF_DEFRAG_CMD_START_LAYOUT_FIX:
+ case GF_DEFRAG_CMD_START_FORCE:
+ case GF_DEFRAG_CMD_START_TIER:
+
+ ret = dict_get_int32(dict, "force", &is_force);
+ if (ret)
+ is_force = 0;
+ if (!is_force) {
+ /* Reset defrag status to 'NOT STARTED' whenever a
+ * remove-brick/rebalance command is issued to remove
+ * stale information from previous run.
+ */
+ volinfo->rebal.defrag_status = GF_DEFRAG_STATUS_NOT_STARTED;
+
+ ret = dict_get_str(dict, GF_REBALANCE_TID_KEY, &task_id_str);
+ if (ret) {
+ gf_msg_debug(this->name, 0,
+ "Missing rebalance id");
+ ret = 0;
+ } else {
+ gf_uuid_parse(task_id_str, volinfo->rebal.rebalance_id);
+ volinfo->rebal.op = GD_OP_REBALANCE;
+ }
+ if (!gd_should_i_start_rebalance(volinfo)) {
+ /* Store the rebalance-id and rebalance command
+ * even if the peer isn't starting a rebalance
+ * process. On peers where a rebalance process
+ * is started, glusterd_handle_defrag_start
+ * performs the storing.
+ * Storing this is needed for having
+ * 'volume status' work correctly.
+ */
+ glusterd_store_perform_node_state_store(volinfo);
+ break;
+ }
+ if (dict_get_uint32(dict, "commit-hash", &commit_hash) == 0) {
+ volinfo->rebal.commit_hash = commit_hash;
+ }
+ ret = glusterd_handle_defrag_start(volinfo, msg, sizeof(msg),
+ cmd, NULL, GD_OP_REBALANCE);
+ break;
+ } else {
+ /* Reset defrag status to 'STARTED' so that the
+ * pid is checked and restarted accordingly.
+ * If the pid is not running it executes the
+ * "NOT_STARTED" case and restarts the process
+ */
+ volinfo->rebal.defrag_status = GF_DEFRAG_STATUS_STARTED;
+ volinfo->rebal.defrag_cmd = cmd;
+ volinfo->rebal.op = GD_OP_REBALANCE;
+
+ ret = dict_get_str(dict, GF_REBALANCE_TID_KEY, &task_id_str);
+ if (ret) {
+ gf_msg_debug(this->name, 0,
+ "Missing rebalance id");
+ ret = 0;
+ } else {
+ gf_uuid_parse(task_id_str, volinfo->rebal.rebalance_id);
+ volinfo->rebal.op = GD_OP_REBALANCE;
+ }
+ if (dict_get_uint32(dict, "commit-hash", &commit_hash) == 0) {
+ volinfo->rebal.commit_hash = commit_hash;
+ }
+ ret = glusterd_restart_rebalance_for_volume(volinfo);
+ break;
+ }
+ case GF_DEFRAG_CMD_STOP:
+ case GF_DEFRAG_CMD_STOP_DETACH_TIER:
+ /* Clear task-id only on explicitly stopping rebalance.
+ * Also clear the stored operation, so it doesn't cause trouble
+ * with future rebalance/remove-brick starts
+ */
+ gf_uuid_clear(volinfo->rebal.rebalance_id);
+ volinfo->rebal.op = GD_OP_NONE;
+
+ /* Fall back to the old volume file in case of decommission*/
+ cds_list_for_each_entry_safe(brickinfo, tmp, &volinfo->bricks,
+ brick_list)
+ {
+ if (!brickinfo->decommissioned)
+ continue;
+ brickinfo->decommissioned = 0;
+ volfile_update = _gf_true;
+ }
+
+ if (volfile_update == _gf_false) {
+ ret = 0;
+ break;
+ }
+
+ ret = glusterd_create_volfiles_and_notify_services(volinfo);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_WARNING, 0,
+ GD_MSG_VOLFILE_CREATE_FAIL, "failed to create volfiles");
+ goto out;
+ }
+
+ ret = glusterd_store_volinfo(volinfo,
+ GLUSTERD_VOLINFO_VER_AC_INCREMENT);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_WARNING, 0, GD_MSG_VOLINFO_SET_FAIL,
+ "failed to store volinfo");
+ goto out;
+ }
+
+ if (volinfo->type == GF_CLUSTER_TYPE_TIER &&
+ cmd == GF_OP_CMD_STOP_DETACH_TIER) {
+ glusterd_defrag_info_set(volinfo, dict,
+ GF_DEFRAG_CMD_START_TIER,
+ GF_DEFRAG_CMD_START, GD_OP_REBALANCE);
+ glusterd_restart_rebalance_for_volume(volinfo);
+ }
+
+ ret = 0;
+ break;
+
+ case GF_DEFRAG_CMD_START_DETACH_TIER:
+ case GF_DEFRAG_CMD_STATUS:
+ case GF_DEFRAG_CMD_STATUS_TIER:
+ break;
+ default:
+ break;
+ }
+
+out:
+ if (ret && op_errstr && msg[0])
+ *op_errstr = gf_strdup(msg);
+
+ return ret;
+}
+
+int
glusterd_op_stage_rebalance (dict_t *dict, char **op_errstr)
{
char *volname = NULL;
diff --git a/xlators/mgmt/glusterd/src/glusterd-syncop.c b/xlators/mgmt/glusterd/src/glusterd-syncop.c
index 9a67d1c..7baef64 100644
--- a/xlators/mgmt/glusterd/src/glusterd-syncop.c
+++ b/xlators/mgmt/glusterd/src/glusterd-syncop.c
@@ -317,6 +317,15 @@ glusterd_syncop_aggr_rsp_dict (glusterd_op_t op, dict_t *aggr, dict_t *rsp)
ret = glusterd_max_opversion_use_rsp_dict (aggr, rsp);
break;
+ case GD_OP_PROFILE_VOLUME:
+ ret = glusterd_profile_volume_use_rsp_dict(aggr, rsp);
+ break;
+
+ case GD_OP_REBALANCE:
+ case GD_OP_DEFRAG_BRICK_VOLUME:
+ ret = glusterd_volume_rebalance_use_rsp_dict(aggr, rsp);
+ break;
+
case GD_OP_TIER_STATUS:
case GD_OP_DETACH_TIER_STATUS:
case GD_OP_REMOVE_TIER_BRICK:
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index 6468ecb..0fe56eb 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -10884,7 +10884,7 @@ glusterd_volume_rebalance_use_rsp_dict (dict_t *aggr, dict_t *rsp_dict)
int ret = 0;
int32_t index = 0;
int32_t count = 0;
- int32_t current_index = 2;
+ int32_t current_index = 1;
int32_t value32 = 0;
uint64_t value = 0;
char *peer_uuid_str = NULL;
@@ -10925,7 +10925,7 @@ glusterd_volume_rebalance_use_rsp_dict (dict_t *aggr, dict_t *rsp_dict)
if (ret)
gf_msg ("glusterd", GF_LOG_ERROR, 0,
GD_MSG_DICT_GET_FAILED,
- "failed to get index");
+ "failed to get index from rsp dict");
memset (key, 0, 256);
snprintf (key, 256, "node-uuid-%d", index);
diff --git a/xlators/mgmt/glusterd/src/glusterd.h b/xlators/mgmt/glusterd/src/glusterd.h
index 42c8821..f1e41be 100644
--- a/xlators/mgmt/glusterd/src/glusterd.h
+++ b/xlators/mgmt/glusterd/src/glusterd.h
@@ -1223,6 +1223,15 @@ int glusterd_op_stage_add_brick (dict_t *dict, char **op_errstr,
dict_t *rsp_dict);
int glusterd_op_stage_remove_brick (dict_t *dict, char **op_errstr);
+int
+glusterd_set_rebalance_id_in_rsp_dict(dict_t *req_dict, dict_t *rsp_dict);
+
+int
+glusterd_mgmt_v3_op_stage_rebalance(dict_t *dict, char **op_errstr);
+
+int
+glusterd_mgmt_v3_op_rebalance(dict_t *dict, char **op_errstr, dict_t *rsp_dict);
+
int glusterd_op_stage_rebalance (dict_t *dict, char **op_errstr);
int glusterd_op_rebalance (dict_t *dict, char **op_errstr, dict_t *rsp_dict);
--
1.8.3.1

56
0491-glusterd-tag-rebalance-mgmt_v3-command-to-op-version.patch Normal file
View File

@ -0,0 +1,56 @@
From d12c2c34484f92e968dea4a52538205a69d5484f Mon Sep 17 00:00:00 2001
From: Atin Mukherjee <amukherj@redhat.com>
Date: Tue, 18 Dec 2018 17:57:25 +0530
Subject: [PATCH 491/493] glusterd: tag rebalance mgmt_v3 command to op-version
31305
In upstream migrating rebalance command is tagged to op-version 60000
but in downstream the latest new op-version is 31305.
Label: DOWNSTREAM ONLY
Change-Id: I30bbad3efca29bf42b9a750581eb1aebc8a30ff9
BUG: 1652466
Signed-off-by: Atin Mukherjee <amukherj@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158943
Tested-by: RHGS Build Bot <nigelb@redhat.com>
---
libglusterfs/src/globals.h | 2 --
xlators/mgmt/glusterd/src/glusterd-rebalance.c | 4 ++--
2 files changed, 2 insertions(+), 4 deletions(-)
diff --git a/libglusterfs/src/globals.h b/libglusterfs/src/globals.h
index 5e3b180..343263c 100644
--- a/libglusterfs/src/globals.h
+++ b/libglusterfs/src/globals.h
@@ -111,8 +111,6 @@
#define GD_OP_VERSION_3_13_3 31303 /* Op-version for GlusterFS 3.13.3 */
-#define GD_OP_VERSION_6_0 60000 /* Op-version for GlusterFS 6.0 */
-
/* Downstream only change */
#define GD_OP_VERSION_3_11_2 31102 /* Op-version for RHGS 3.3.1-async */
#define GD_OP_VERSION_3_13_3 31303 /* Op-version for RHGS-3.4-Batch Update-1*/
diff --git a/xlators/mgmt/glusterd/src/glusterd-rebalance.c b/xlators/mgmt/glusterd/src/glusterd-rebalance.c
index 7ba5f65..ba32241 100644
--- a/xlators/mgmt/glusterd/src/glusterd-rebalance.c
+++ b/xlators/mgmt/glusterd/src/glusterd-rebalance.c
@@ -569,12 +569,12 @@ __glusterd_handle_defrag_volume (rpcsvc_request_t *req)
} else
op = GD_OP_REBALANCE;
- if (priv->op_version < GD_OP_VERSION_6_0) {
+ if (priv->op_version < GD_OP_VERSION_3_13_5) {
gf_msg_debug(this->name, 0,
"The cluster is operating at "
"version less than %d. Falling back "
"to op-sm framework.",
- GD_OP_VERSION_6_0);
+ GD_OP_VERSION_3_13_5);
ret = glusterd_op_begin(req, op, dict, msg, sizeof(msg));
glusterd_friend_sm();
glusterd_op_sm();
--
1.8.3.1

282
0492-mem-pool-track-glusterfs_ctx_t-in-struct-mem_pool.patch Normal file
View File

@ -0,0 +1,282 @@
From c8e58e3a577e70a64df77fe885847285f682d9fb Mon Sep 17 00:00:00 2001
From: Niels de Vos <ndevos@redhat.com>
Date: Tue, 29 Aug 2017 00:16:22 +0200
Subject: [PATCH 492/493] mem-pool: track glusterfs_ctx_t in struct mem_pool
In order to generate statedumps per glusterfs_ctx_t, it is needed to
place all the memory pools in a structure that the context can reach.
The 'struct mem_pool' has been extended with a 'list_head owner' that is
linked with the glusterfs_ctx_t->mempool_list.
All callers of mem_pool_new() have been updated to pass the current
glusterfs_ctx_t along. This context is needed to add the new memory pool
to the list and for grabbing the ctx->lock while updating the
glusterfs_ctx_t->mempool_list.
> Updates: #307
> Change-Id: Ia9384424d8d1630ef3efc9d5d523bf739c356c6e
> Signed-off-by: Niels de Vos <ndevos@redhat.com>
> Reviewed-on: https://review.gluster.org/18075
> Smoke: Gluster Build System <jenkins@build.gluster.org>
> CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
> Reviewed-by: Jeff Darcy <jeff@pl.atyp.us>
Change-Id: Id3b193f366f7c46f91b77bced8729a4eb538837b
BUG: 1648893
Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158937
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
libglusterfs/src/ctx.c | 2 --
libglusterfs/src/glusterfs.h | 2 --
libglusterfs/src/mem-pool.c | 18 +++++++++++++++++-
libglusterfs/src/mem-pool.h | 16 +++++++++++++---
libglusterfs/src/rbthash.c | 5 +++--
libglusterfs/src/rbthash.h | 2 +-
libglusterfs/src/statedump.c | 16 +++++++++++++++-
xlators/cluster/ec/src/ec-method.c | 2 +-
xlators/nfs/server/src/nfs3.c | 3 ++-
xlators/performance/io-cache/src/io-cache.c | 2 +-
10 files changed, 53 insertions(+), 15 deletions(-)
diff --git a/libglusterfs/src/ctx.c b/libglusterfs/src/ctx.c
index 94c56ac..90480d0 100644
--- a/libglusterfs/src/ctx.c
+++ b/libglusterfs/src/ctx.c
@@ -31,9 +31,7 @@ glusterfs_ctx_new ()
ctx->mem_acct_enable = gf_global_mem_acct_enable_get();
INIT_LIST_HEAD (&ctx->graphs);
-#if defined(OLD_MEM_POOLS)
INIT_LIST_HEAD (&ctx->mempool_list);
-#endif
INIT_LIST_HEAD (&ctx->volfile_list);
ctx->daemon_pipe[0] = -1;
diff --git a/libglusterfs/src/glusterfs.h b/libglusterfs/src/glusterfs.h
index c12e94e..157437c 100644
--- a/libglusterfs/src/glusterfs.h
+++ b/libglusterfs/src/glusterfs.h
@@ -526,11 +526,9 @@ struct _glusterfs_ctx {
int process_mode; /*mode in which process is runninng*/
struct syncenv *env; /* The env pointer to the synctasks */
-#if defined(OLD_MEM_POOLS)
struct list_head mempool_list; /* used to keep a global list of
mempools, used to log details of
mempool in statedump */
-#endif
char *statedump_path;
struct mem_pool *dict_pool;
diff --git a/libglusterfs/src/mem-pool.c b/libglusterfs/src/mem-pool.c
index a8a9347..999a83f 100644
--- a/libglusterfs/src/mem-pool.c
+++ b/libglusterfs/src/mem-pool.c
@@ -676,7 +676,7 @@ void mem_pools_fini (void) {}
#endif
struct mem_pool *
-mem_pool_new_fn (unsigned long sizeof_type,
+mem_pool_new_fn (glusterfs_ctx_t *ctx, unsigned long sizeof_type,
unsigned long count, char *name)
{
unsigned int i;
@@ -706,10 +706,18 @@ mem_pool_new_fn (unsigned long sizeof_type,
if (!new)
return NULL;
+ new->ctx = ctx;
new->sizeof_type = sizeof_type;
new->count = count;
new->name = name;
new->pool = pool;
+ INIT_LIST_HEAD (&new->owner);
+
+ LOCK (&ctx->lock);
+ {
+ list_add (&new->owner, &ctx->mempool_list);
+ }
+ UNLOCK (&ctx->lock);
return new;
}
@@ -905,6 +913,14 @@ mem_put (void *ptr)
void
mem_pool_destroy (struct mem_pool *pool)
{
+ /* remove this pool from the owner (glusterfs_ctx_t) */
+ LOCK (&pool->ctx->lock);
+ {
+ list_del (&pool->owner);
+ }
+ UNLOCK (&pool->ctx->lock);
+
+ /* free this pool, but keep the mem_pool_shared */
GF_FREE (pool);
/*
diff --git a/libglusterfs/src/mem-pool.h b/libglusterfs/src/mem-pool.h
index 057d957..0ebb63b 100644
--- a/libglusterfs/src/mem-pool.h
+++ b/libglusterfs/src/mem-pool.h
@@ -16,6 +16,7 @@
#include "atomic.h"
#include "logging.h"
#include "mem-types.h"
+#include "glusterfs.h" /* for glusterfs_ctx_t */
#include <stdlib.h>
#include <inttypes.h>
#include <string.h>
@@ -207,11 +208,15 @@ out:
/* kind of 'header' for the actual mem_pool_shared structure, this might make
* it possible to dump some more details in a statedump */
struct mem_pool {
+ /* object size, without pooled_obj_hdr_t */
unsigned long sizeof_type;
unsigned long count;
char *name;
- struct mem_pool_shared *pool;
+ struct list_head owner; /* glusterfs_ctx_t->mempool_list */
+ glusterfs_ctx_t *ctx; /* take ctx->lock when updating owner */
+
+ struct mem_pool_shared *pool; /* the initial pool that was returned */
};
typedef struct pooled_obj_hdr {
@@ -276,9 +281,14 @@ void mem_pools_init_late (void); /* start the pool_sweeper thread */
void mem_pools_fini (void); /* cleanup memory pools */
struct mem_pool *
-mem_pool_new_fn (unsigned long sizeof_type, unsigned long count, char *name);
+mem_pool_new_fn (glusterfs_ctx_t *ctx, unsigned long sizeof_type, unsigned long
+ count, char *name);
+
+#define mem_pool_new(type, count) (mem_pool_new_fn (THIS->ctx, \
+ sizeof(type), count, #type))
-#define mem_pool_new(type,count) mem_pool_new_fn (sizeof(type), count, #type)
+#define mem_pool_new_ctx(ctx, type, count) (mem_pool_new_fn (ctx, \
+ sizeof(type), count, #type))
void mem_put (void *ptr);
void *mem_get (struct mem_pool *pool);
diff --git a/libglusterfs/src/rbthash.c b/libglusterfs/src/rbthash.c
index 52d8a15..06fc7ee 100644
--- a/libglusterfs/src/rbthash.c
+++ b/libglusterfs/src/rbthash.c
@@ -83,7 +83,7 @@ err:
*/
rbthash_table_t *
-rbthash_table_init (int buckets, rbt_hasher_t hfunc,
+rbthash_table_init (glusterfs_ctx_t *ctx, int buckets, rbt_hasher_t hfunc,
rbt_data_destroyer_t dfunc,
unsigned long expected_entries,
struct mem_pool *entrypool)
@@ -123,7 +123,8 @@ rbthash_table_init (int buckets, rbt_hasher_t hfunc,
if (expected_entries) {
newtab->entrypool =
- mem_pool_new (rbthash_entry_t, expected_entries);
+ mem_pool_new_ctx (ctx, rbthash_entry_t,
+ expected_entries);
if (!newtab->entrypool) {
goto free_buckets;
}
diff --git a/libglusterfs/src/rbthash.h b/libglusterfs/src/rbthash.h
index b093ce9..949b88a 100644
--- a/libglusterfs/src/rbthash.h
+++ b/libglusterfs/src/rbthash.h
@@ -52,7 +52,7 @@ typedef struct rbthash_table {
} rbthash_table_t;
extern rbthash_table_t *
-rbthash_table_init (int buckets, rbt_hasher_t hfunc,
+rbthash_table_init (glusterfs_ctx_t *ctx, int buckets, rbt_hasher_t hfunc,
rbt_data_destroyer_t dfunc, unsigned long expected_entries,
struct mem_pool *entrypool);
diff --git a/libglusterfs/src/statedump.c b/libglusterfs/src/statedump.c
index a4635f3..4aad014 100644
--- a/libglusterfs/src/statedump.c
+++ b/libglusterfs/src/statedump.c
@@ -377,11 +377,11 @@ gf_proc_dump_mem_info_to_dict (dict_t *dict)
void
gf_proc_dump_mempool_info (glusterfs_ctx_t *ctx)
{
-#if defined(OLD_MEM_POOLS)
struct mem_pool *pool = NULL;
gf_proc_dump_add_section ("mempool");
+#if defined(OLD_MEM_POOLS)
list_for_each_entry (pool, &ctx->mempool_list, global_list) {
gf_proc_dump_write ("-----", "-----");
gf_proc_dump_write ("pool-name", "%s", pool->name);
@@ -396,6 +396,20 @@ gf_proc_dump_mempool_info (glusterfs_ctx_t *ctx)
gf_proc_dump_write ("cur-stdalloc", "%d", pool->curr_stdalloc);
gf_proc_dump_write ("max-stdalloc", "%d", pool->max_stdalloc);
}
+#else
+ LOCK (&ctx->lock);
+ {
+ list_for_each_entry (pool, &ctx->mempool_list, owner) {
+ gf_proc_dump_write ("-----", "-----");
+ gf_proc_dump_write ("pool-name", "%s", pool->name);
+ gf_proc_dump_write ("sizeof-type", "%lu", pool->sizeof_type);
+ gf_proc_dump_write ("padded-sizeof", "%d", 1 << pool->pool->power_of_two);
+ gf_proc_dump_write ("shared-pool", "%p", pool->pool);
+ }
+ }
+ UNLOCK (&ctx->lock);
+
+ /* TODO: details of (struct mem_pool_shared) pool->pool */
#endif
}
diff --git a/xlators/cluster/ec/src/ec-method.c b/xlators/cluster/ec/src/ec-method.c
index e0dd8e7..a2dd2bd 100644
--- a/xlators/cluster/ec/src/ec-method.c
+++ b/xlators/cluster/ec/src/ec-method.c
@@ -310,7 +310,7 @@ ec_method_init(xlator_t *xl, ec_matrix_list_t *list, uint32_t columns,
INIT_LIST_HEAD(&list->lru);
int32_t err;
- list->pool = mem_pool_new_fn(sizeof(ec_matrix_t) +
+ list->pool = mem_pool_new_fn(xl->ctx, sizeof(ec_matrix_t) +
sizeof(ec_matrix_row_t) * columns +
sizeof(uint32_t) * columns * columns,
128, "ec_matrix_t");
diff --git a/xlators/nfs/server/src/nfs3.c b/xlators/nfs/server/src/nfs3.c
index 040d316..b053eb3 100644
--- a/xlators/nfs/server/src/nfs3.c
+++ b/xlators/nfs/server/src/nfs3.c
@@ -5786,7 +5786,8 @@ nfs3_init_state (xlator_t *nfsx)
localpool = nfs->memfactor * GF_NFS_CONCURRENT_OPS_MULT;
gf_msg_trace (GF_NFS3, 0, "local pool: %d", localpool);
- nfs3->localpool = mem_pool_new (nfs3_call_state_t, localpool);
+ nfs3->localpool = mem_pool_new_ctx (nfsx->ctx, nfs3_call_state_t,
+ localpool);
if (!nfs3->localpool) {
gf_msg (GF_NFS3, GF_LOG_ERROR, ENOMEM, NFS_MSG_NO_MEMORY,
"local mempool creation failed");
diff --git a/xlators/performance/io-cache/src/io-cache.c b/xlators/performance/io-cache/src/io-cache.c
index de44ad2..d7b3b37 100644
--- a/xlators/performance/io-cache/src/io-cache.c
+++ b/xlators/performance/io-cache/src/io-cache.c
@@ -1146,7 +1146,7 @@ ioc_readv (call_frame_t *frame, xlator_t *this, fd_t *fd,
if (!ioc_inode->cache.page_table) {
ioc_inode->cache.page_table
= rbthash_table_init
- (IOC_PAGE_TABLE_BUCKET_COUNT,
+ (this->ctx, IOC_PAGE_TABLE_BUCKET_COUNT,
ioc_hashfn, NULL, 0,
table->mem_pool);
--
1.8.3.1

96
0493-mem-pool-count-allocations-done-per-user-pool.patch Normal file
View File

@ -0,0 +1,96 @@
From bb668d69815209bd7f7f4669142191d4b48bcde8 Mon Sep 17 00:00:00 2001
From: Mohit Agrawal <moagrawa@redhat.com>
Date: Tue, 18 Dec 2018 19:41:42 +0530
Subject: [PATCH 493/493] mem-pool: count allocations done per user-pool
Count the active allocations per 'struct mem_pool'. These are the
objects that the calling component allocated and free'd in the memory
pool for this specific type. Having this count in the statedump will
make it easy to find memory leaks.
> Updates: #307
> Change-Id: I797fabab86f104e49338c00e449a7d0b0d270004
> Signed-off-by: Niels de Vos <ndevos@redhat.com>
> Reviewed-on: https://review.gluster.org/18074
> Smoke: Gluster Build System <jenkins@build.gluster.org>
> CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
> Reviewed-by: Jeff Darcy <jeff@pl.atyp.us>
Change-Id: I2e4375fd59f11288f41a36cad131d794bff19fbb
BUG: 1648893
Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/158961
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
libglusterfs/src/mem-pool.c | 4 ++++
libglusterfs/src/mem-pool.h | 3 ++-
libglusterfs/src/statedump.c | 4 ++++
3 files changed, 10 insertions(+), 1 deletion(-)
diff --git a/libglusterfs/src/mem-pool.c b/libglusterfs/src/mem-pool.c
index 999a83f..d82a371 100644
--- a/libglusterfs/src/mem-pool.c
+++ b/libglusterfs/src/mem-pool.c
@@ -711,6 +711,7 @@ mem_pool_new_fn (glusterfs_ctx_t *ctx, unsigned long sizeof_type,
new->count = count;
new->name = name;
new->pool = pool;
+ GF_ATOMIC_INIT (new->active, 0);
INIT_LIST_HEAD (&new->owner);
LOCK (&ctx->lock);
@@ -864,6 +865,8 @@ mem_get (struct mem_pool *mem_pool)
retval->pool_list = pool_list;
retval->power_of_two = mem_pool->pool->power_of_two;
+ GF_ATOMIC_INC (mem_pool->active);
+
return retval + 1;
#endif /* GF_DISABLE_MEMPOOL */
}
@@ -894,6 +897,7 @@ mem_put (void *ptr)
pt_pool = &pool_list->pools[hdr->power_of_two-POOL_SMALLEST];
hdr->magic = GF_MEM_INVALID_MAGIC;
+ GF_ATOMIC_DEC (hdr->pool->active);
(void) pthread_spin_lock (&pool_list->lock);
if (!pool_list->poison) {
diff --git a/libglusterfs/src/mem-pool.h b/libglusterfs/src/mem-pool.h
index 0ebb63b..aa1d045 100644
--- a/libglusterfs/src/mem-pool.h
+++ b/libglusterfs/src/mem-pool.h
@@ -210,8 +210,9 @@ out:
struct mem_pool {
/* object size, without pooled_obj_hdr_t */
unsigned long sizeof_type;
- unsigned long count;
+ unsigned long count; /* requested pool size (unused) */
char *name;
+ gf_atomic_t active; /* current allocations */
struct list_head owner; /* glusterfs_ctx_t->mempool_list */
glusterfs_ctx_t *ctx; /* take ctx->lock when updating owner */
diff --git a/libglusterfs/src/statedump.c b/libglusterfs/src/statedump.c
index 4aad014..a04c535 100644
--- a/libglusterfs/src/statedump.c
+++ b/libglusterfs/src/statedump.c
@@ -400,10 +400,14 @@ gf_proc_dump_mempool_info (glusterfs_ctx_t *ctx)
LOCK (&ctx->lock);
{
list_for_each_entry (pool, &ctx->mempool_list, owner) {
+ int64_t active = GF_ATOMIC_GET (pool->active);
+
gf_proc_dump_write ("-----", "-----");
gf_proc_dump_write ("pool-name", "%s", pool->name);
+ gf_proc_dump_write ("active-count", "%"GF_PRI_ATOMIC, active);
gf_proc_dump_write ("sizeof-type", "%lu", pool->sizeof_type);
gf_proc_dump_write ("padded-sizeof", "%d", 1 << pool->pool->power_of_two);
+ gf_proc_dump_write ("size", "%lu", (1 << pool->pool->power_of_two) * active);
gf_proc_dump_write ("shared-pool", "%p", pool->pool);
}
}
--
1.8.3.1

62
glusterfs.spec
View File

@ -192,7 +192,7 @@ Release: 0.1%{?prereltag:.%{prereltag}}%{?dist}
%else
Name: glusterfs
Version: 3.12.2
Release: 32%{?dist}
Release: 33%{?dist}
%endif
License: GPLv2 or LGPLv3+
Group: System Environment/Base
@ -720,6 +720,44 @@ Patch0452: 0452-glusterd-make-max-bricks-per-process-default-value-t.patch
Patch0453: 0453-server-Resolve-memory-leak-path-in-server_init.patch
Patch0454: 0454-glusterd-set-cluster.max-bricks-per-process-to-250.patch
Patch0455: 0455-glusterd-fix-get_mux_limit_per_process-to-read-defau.patch
Patch0456: 0456-Update-rfc.sh-to-rhgs-3.4.3.patch
Patch0457: 0457-cluster-dht-sync-brick-root-perms-on-add-brick.patch
Patch0458: 0458-glusterd-fix-crash.patch
Patch0459: 0459-glfsheal-add-a-nolog-flag.patch
Patch0460: 0460-cli-add-a-warning-confirmation-message-in-peer-detac.patch
Patch0461: 0461-mount-fuse-Add-support-for-multi-threaded-fuse-reade.patch
Patch0462: 0462-posix-Do-not-log-ENXIO-errors-for-seek-fop.patch
Patch0463: 0463-build-glusterfs.spec-.in-firewalld-file-doesn-t-use-.patch
Patch0464: 0464-build-exclude-packaging-crypt.so.patch
Patch0465: 0465-build-add-missing-explicit-package-dependencies.patch
Patch0466: 0466-extras-Add-group-distributed-virt-for-single-brick-o.patch
Patch0467: 0467-glusterd-glusterd-to-regenerate-volfiles-when-GD_OP_.patch
Patch0468: 0468-core-move-invalid-port-logs-to-DEBUG-log-level.patch
Patch0469: 0469-nfs-set-ctx-for-every-inode-looked-up-nfs3_fh_resolv.patch
Patch0470: 0470-dht-fix-use-after-free-in-dht_rmdir_readdirp_cbk.patch
Patch0471: 0471-glusterd-migrating-profile-commands-to-mgmt_v3-frame.patch
Patch0472: 0472-glusterd-introduce-a-new-op-version-for-rhgs-3.4.3.patch
Patch0473: 0473-rpc-bump-up-server.event-threads.patch
Patch0474: 0474-afr-open_ftruncate_cbk-should-read-fd-from-local-con.patch
Patch0475: 0475-glusterd-perform-store-operation-in-cleanup-lock.patch
Patch0476: 0476-afr-add-checks-for-allowing-lookups.patch
Patch0477: 0477-glusterd-perform-rcu_read_lock-unlock-under-cleanup_.patch
Patch0478: 0478-libglusterfs-fix-memory-corruption-caused-by-per-thr.patch
Patch0479: 0479-ganesha-ha-ensure-pacemaker-is-enabled-after-setup.patch
Patch0480: 0480-geo-rep-Make-slave-volume-read-only-by-default.patch
Patch0481: 0481-extras-hooks-Do-not-blindly-remove-volume-share-from.patch
Patch0482: 0482-extras-hooks-General-improvements-to-S30samba-start..patch
Patch0483: 0483-Do-not-blindly-add-volume-share-section-to-smb.conf.patch
Patch0484: 0484-extras-New-group-volume-set-command-for-Samba-integr.patch
Patch0485: 0485-cluster-ec-Prevent-volume-create-without-redundant-b.patch
Patch0486: 0486-performance-rda-Fixed-dict_t-memory-leak.patch
Patch0487: 0487-mem-pool-add-tracking-of-mem_pool-that-requested-the.patch
Patch0488: 0488-cluster-afr-Allow-lookup-on-root-if-it-is-from-ADD_R.patch
Patch0489: 0489-cluster-afr-Do-not-update-read_subvol-in-inode_ctx-a.patch
Patch0490: 0490-glusterd-migrating-rebalance-commands-to-mgmt_v3-fra.patch
Patch0491: 0491-glusterd-tag-rebalance-mgmt_v3-command-to-op-version.patch
Patch0492: 0492-mem-pool-track-glusterfs_ctx_t-in-struct-mem_pool.patch
Patch0493: 0493-mem-pool-count-allocations-done-per-user-pool.patch
%description
GlusterFS is a distributed file-system capable of scaling to several
@ -739,6 +777,7 @@ Summary: GlusterFS api library
Group: System Environment/Daemons
Requires: %{name}%{?_isa} = %{version}-%{release}
Requires: %{name}-client-xlators%{?_isa} = %{version}-%{release}
Requires: %{name}-libs%{?_isa} = %{version}-%{release}
%description api
GlusterFS is a distributed file-system capable of scaling to several
@ -757,6 +796,7 @@ Group: Development/Libraries
Requires: %{name}%{?_isa} = %{version}-%{release}
Requires: %{name}-devel%{?_isa} = %{version}-%{release}
Requires: libacl-devel
Requires: %{name}-api%{?_isa} = %{version}-%{release}
%description api-devel
GlusterFS is a distributed file-system capable of scaling to several
@ -793,6 +833,8 @@ Requires: %{name}%{?_isa} = %{version}-%{release}
%if ( 0%{!?_without_extra_xlators:1} )
Requires: %{name}-extra-xlators = %{version}-%{release}
%endif
Requires: %{name}-libs%{?_isa} = %{version}-%{release}
Requires: %{name}-server%{?_isa} = %{version}-%{release}
%description devel
GlusterFS is a distributed file-system capable of scaling to several
@ -842,6 +884,7 @@ Requires: %{name}-client-xlators%{?_isa} = %{version}-%{release}
Obsoletes: %{name}-client < %{version}-%{release}
Provides: %{name}-client = %{version}-%{release}
Requires: %{name}-libs%{?_isa} = %{version}-%{release}
%description fuse
GlusterFS is a distributed file-system capable of scaling to several
@ -911,6 +954,7 @@ BuildRequires: python-ctypes
Requires: python2-gluster = %{version}-%{release}
Requires: rsync
Requires: util-linux
Requires: %{name}-libs%{?_isa} = %{version}-%{release}
%description geo-replication
GlusterFS is a distributed file-system capable of scaling to several
@ -983,6 +1027,7 @@ BuildRequires: libibverbs-devel
BuildRequires: librdmacm-devel >= 1.0.15
%endif
Requires: %{name}%{?_isa} = %{version}-%{release}
Requires: %{name}-libs%{?_isa} = %{version}-%{release}
%description rdma
GlusterFS is a distributed file-system capable of scaling to several
@ -1113,6 +1158,7 @@ This package provides the glusterfs server daemon.
%package client-xlators
Summary: GlusterFS client-side translators
Group: Applications/File
Requires: %{name}-libs%{?_isa} = %{version}-%{release}
%description client-xlators
GlusterFS is a distributed file-system capable of scaling to several
@ -1675,7 +1721,7 @@ exit 0
%exclude %{_tmpfilesdir}/gluster.conf
%endif
%if ( 0%{?_with_firewalld:1} )
%exclude /usr/lib/firewalld/services/glusterfs.xml
%exclude %{_prefix}/lib/firewalld/services/glusterfs.xml
%endif
%endif
%doc ChangeLog COPYING-GPLV2 COPYING-LGPLV3 INSTALL README.md THANKS
@ -1705,8 +1751,8 @@ exit 0
%{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/debug/trace.so
%if ( ! ( 0%{?rhel} && 0%{?rhel} < 6 ) )
# RHEL-5 based distributions have a too old openssl
%dir %{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/encryption
%{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/encryption/crypt.so
%exclude %dir %{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/encryption
%exclude %{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/encryption/crypt.so
%endif
%dir %{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/features
%{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/features/access-control.so
@ -2021,7 +2067,9 @@ exit 0
%attr(0644,-,-) %{_sharedstatedir}/glusterd/groups/metadata-cache
%attr(0644,-,-) %{_sharedstatedir}/glusterd/groups/gluster-block
%attr(0644,-,-) %{_sharedstatedir}/glusterd/groups/db-workload
%attr(0644,-,-) %{_sharedstatedir}/glusterd/groups/distributed-virt
%attr(0644,-,-) %{_sharedstatedir}/glusterd/groups/nl-cache
%attr(0644,-,-) %{_sharedstatedir}/glusterd/groups/samba
%dir %attr(0755,-,-) %{_sharedstatedir}/glusterd/glusterfind
%dir %attr(0755,-,-) %{_sharedstatedir}/glusterd/glusterfind/.keys
%ghost %dir %attr(0755,-,-) %{_sharedstatedir}/glusterd/glustershd
@ -2668,6 +2716,12 @@ fi
%endif
%changelog
* Tue Dec 18 2018 Milind Changire <mchangir@redhat.com> - 3.12.2-33
- fixes bugs bz#1350745 bz#1362129 bz#1541568 bz#1597252 bz#1599220
bz#1633177 bz#1637564 bz#1639476 bz#1639568 bz#1643370 bz#1645480 bz#1648296
bz#1648893 bz#1651040 bz#1651460 bz#1652466 bz#1652537 bz#1653224 bz#1653613
bz#1654103 bz#1654161 bz#1655385 bz#1655578 bz#1656357 bz#1659439
* Fri Dec 07 2018 Milind Changire <mchangir@redhat.com> - 3.12.2-32
- fixes bugs bz#1656924