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

import kronosnet-1.22-4.el9_0

Browse Source
This commit is contained in:
CentOS Sources 2022-08-09 10:53:56 -04:00 committed by Stepan Oksanichenko
parent 4a857aa4e2
commit c02379b03e
3 changed files with 136 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

82
SOURCES/bz2112923-icmp-faster-link-down-detection.patch Normal file
View File

@ -0,0 +1,82 @@
commit 28ddb87a2f4562c5d1752a778744cc56136f81c1
Author: Fabio M. Di Nitto <fdinitto@redhat.com>
Date: Sun Nov 7 17:02:05 2021 +0100
[udp] use ICMP error messages to trigger faster link down detection
this solves a possible race condition when:
- node1 is running
- node2 very fast
- node1 does NOT have enough time to detect that node2 has gone
and reset the local seq numbers / buffers
- node1 will start rejecting valid packets from node2
There is still a potential minor race condition where app
can restart so fast that kernel / network don't have time
to generate an ICMP error. This will be addressed using
instance id in onwire v2 protocol, as suggested by Jan F.
Signed-off-by: Fabio M. Di Nitto <fdinitto@redhat.com>
diff --git a/libknet/transport_udp.c b/libknet/transport_udp.c
index 963340d..32dd032 100644
--- a/libknet/transport_udp.c
+++ b/libknet/transport_udp.c
@@ -364,6 +364,46 @@ static int read_errs_from_sock(knet_handle_t knet_h, int sockfd)
log_debug(knet_h, KNET_SUB_TRANSP_UDP, "Received ICMP error from %s: %s destination unknown", addr_str, strerror(sock_err->ee_errno));
} else {
log_debug(knet_h, KNET_SUB_TRANSP_UDP, "Received ICMP error from %s: %s %s", addr_str, strerror(sock_err->ee_errno), addr_remote_str);
+ if ((sock_err->ee_errno == ECONNREFUSED) || /* knet is not running on the other node */
+ (sock_err->ee_errno == ECONNABORTED) || /* local kernel closed the socket */
+ (sock_err->ee_errno == ENONET) || /* network does not exist */
+ (sock_err->ee_errno == ENETUNREACH) || /* network unreachable */
+ (sock_err->ee_errno == EHOSTUNREACH) || /* host unreachable */
+ (sock_err->ee_errno == EHOSTDOWN) || /* host down (from kernel/net/ipv4/icmp.c */
+ (sock_err->ee_errno == ENETDOWN)) { /* network down */
+ struct knet_host *host = NULL;
+ struct knet_link *kn_link = NULL;
+ int link_idx, found = 0;
+
+ for (host = knet_h->host_head; host != NULL; host = host->next) {
+ for (link_idx = 0; link_idx < KNET_MAX_LINK; link_idx++) {
+ kn_link = &host->link[link_idx];
+ if (kn_link->outsock == sockfd) {
+ if (!cmpaddr(&remote, &kn_link->dst_addr)) {
+ found = 1;
+ break;
+ }
+ }
+ }
+ if (found) {
+ break;
+ }
+ }
+
+ if ((host) && (kn_link) &&
+ (kn_link->status.connected)) {
+ log_debug(knet_h, KNET_SUB_TRANSP_UDP, "Setting down host %u link %i", host->host_id, kn_link->link_id);
+ /*
+ * setting transport_connected = 0 will trigger
+ * thread_heartbeat link_down process.
+ *
+ * the process terminates calling into transport_link_down
+ * below that will set transport_connected = 1
+ */
+ kn_link->transport_connected = 0;
+ }
+
+ }
}
}
break;
@@ -436,5 +476,9 @@ int udp_transport_link_dyn_connect(knet_handle_t knet_h, int sockfd, struct knet
int udp_transport_link_is_down(knet_handle_t knet_h, struct knet_link *kn_link)
{
+ /*
+ * see comments about handling ICMP error messages
+ */
+ kn_link->transport_connected = 1;
return 0;
}

44
SOURCES/bz212923-fix-dst_seq_num_init-race.patch Normal file
View File

@ -0,0 +1,44 @@
commit 62271c5c0b08041b24930310c04e3933720917c6
Author: Fabio M. Di Nitto <fdinitto@redhat.com>
Date: Mon Nov 8 09:14:22 2021 +0100
[host] fix dst_seq_num initialization race condition
There is a potential race condition where the sender
is overloaded, sending data packets before pings
can kick in and set the correct dst_seq_num.
if this node is starting up (dst_seq_num = 0),
it can start rejecing valid packets and get stuck.
Set the dst_seq_num to the first seen packet and
use that as reference instead.
Signed-off-by: Fabio M. Di Nitto <fdinitto@redhat.com>
diff --git a/libknet/host.c b/libknet/host.c
index f02ef02..54061fd 100644
--- a/libknet/host.c
+++ b/libknet/host.c
@@ -573,6 +573,21 @@ int _seq_num_lookup(knet_handle_t knet_h, struct knet_host *host, seq_num_t seq_
char *dst_cbuf_defrag = host->circular_buffer_defrag;
seq_num_t *dst_seq_num = &host->rx_seq_num;
+ /*
+ * There is a potential race condition where the sender
+ * is overloaded, sending data packets before pings
+ * can kick in and set the correct dst_seq_num.
+ *
+ * if this node is starting up (dst_seq_num = 0),
+ * it can start rejecing valid packets and get stuck.
+ *
+ * Set the dst_seq_num to the first seen packet and
+ * use that as reference instead.
+ */
+ if (!*dst_seq_num) {
+ *dst_seq_num = seq_num;
+ }
+
if (clear_buf) {
_clear_cbuffers(host, seq_num);
}

12
SPECS/kronosnet.spec
View File

@ -37,12 +37,13 @@
Name: kronosnet
Summary: Multipoint-to-Multipoint VPN daemon
Version: 1.22
Release: 3%{?dist}
Release: 4%{?dist}
License: GPLv2+ and LGPLv2+
URL: https://kronosnet.org
Source0: https://kronosnet.org/releases/%{name}-%{version}.tar.xz
#Patch0: bz1954551-enable-stack-clash-protection.patch
Patch0: bz2112923-icmp-faster-link-down-detection.patch
Patch1: bz212923-fix-dst_seq_num_init-race.patch
# Build dependencies
BuildRequires: make
@ -87,6 +88,9 @@ BuildRequires: autoconf automake libtool
%prep
%setup -q -n %{name}-%{version}
%patch0 -p1 -b bz2112923-icmp-faster-link-down-detection
%patch1 -p1 -b bz212923-fix-dst_seq_num_init-race
#%patch0 -p1 -b .bz1954551-enable-stack-clash-protection
%build
@ -439,6 +443,10 @@ Requires: libknet1%{_isa} = %{version}-%{release}
%endif
%changelog
* Wed Aug 03 2022 Christine Caulfield <ccaulfie@redhat.com> - 1.22-4
- Fix "node randomly not rejoining cluster after a pcs cluster stop/start"
Resolves: rhbz#2112923
* Wed Oct 06 2021 Christine Caulfield <ccaulfie@redhat.com> - 1.22-3
- Add back the test suite that got lost in the pull from Fedora
Resolves: rhbz#1999980