|
 |
f6d6d9 |
From 45617b727e280cac384a28ae3d96145e066e6197 Mon Sep 17 00:00:00 2001
|
|
|
f6d6d9 |
From: Reid Wahl <nrwahl@protonmail.com>
|
|
|
f6d6d9 |
Date: Fri, 3 Feb 2023 12:08:57 -0800
|
|
|
f6d6d9 |
Subject: [PATCH] Fix: fencer: Prevent double g_source_remove of op_timer_one
|
|
|
f6d6d9 |
|
|
|
f6d6d9 |
QE observed a rarely reproducible core dump in the fencer during
|
|
|
f6d6d9 |
Pacemaker shutdown, in which we try to g_source_remove() an op timer
|
|
|
f6d6d9 |
that's already been removed.
|
|
|
f6d6d9 |
|
|
|
f6d6d9 |
free_stonith_remote_op_list()
|
|
|
f6d6d9 |
-> g_hash_table_destroy()
|
|
|
f6d6d9 |
-> g_hash_table_remove_all_nodes()
|
|
|
f6d6d9 |
-> clear_remote_op_timers()
|
|
|
f6d6d9 |
-> g_source_remove()
|
|
|
f6d6d9 |
-> crm_glib_handler()
|
|
|
f6d6d9 |
-> "Source ID 190 was not found when attempting to remove it"
|
|
|
f6d6d9 |
|
|
|
f6d6d9 |
The likely cause is that request_peer_fencing() doesn't set
|
|
|
f6d6d9 |
op->op_timer_one to 0 after calling g_source_remove() on it, so if that
|
|
|
f6d6d9 |
op is still in the stonith_remote_op_list at shutdown with the same
|
|
|
f6d6d9 |
timer, clear_remote_op_timers() tries to remove the source for
|
|
|
f6d6d9 |
op_timer_one again.
|
|
|
f6d6d9 |
|
|
|
f6d6d9 |
There are only five locations that call g_source_remove() on a
|
|
|
f6d6d9 |
remote_fencing_op_t timer.
|
|
|
f6d6d9 |
* Three of them are in clear_remote_op_timers(), which first 0-checks
|
|
|
f6d6d9 |
the timer and then sets it to 0 after g_source_remove().
|
|
|
f6d6d9 |
* One is in remote_op_query_timeout(), which does the same.
|
|
|
f6d6d9 |
* The last is the one we fix here in request_peer_fencing().
|
|
|
f6d6d9 |
|
|
|
f6d6d9 |
I don't know all the conditions of QE's test scenario at this point.
|
|
|
f6d6d9 |
What I do know:
|
|
|
f6d6d9 |
* have-watchdog=true
|
|
|
f6d6d9 |
* stonith-watchdog-timeout=10
|
|
|
f6d6d9 |
* no explicit topology
|
|
|
f6d6d9 |
* fence agent script is missing for the configured fence device
|
|
|
f6d6d9 |
* requested fencing of one node
|
|
|
f6d6d9 |
* cluster shutdown
|
|
|
f6d6d9 |
|
|
|
f6d6d9 |
Fixes RHBZ2166967
|
|
|
f6d6d9 |
|
|
|
f6d6d9 |
Signed-off-by: Reid Wahl <nrwahl@protonmail.com>
|
|
|
f6d6d9 |
---
|
|
|
f6d6d9 |
daemons/fenced/fenced_remote.c | 1 +
|
|
|
f6d6d9 |
1 file changed, 1 insertion(+)
|
|
|
f6d6d9 |
|
|
|
f6d6d9 |
diff --git a/daemons/fenced/fenced_remote.c b/daemons/fenced/fenced_remote.c
|
|
|
f6d6d9 |
index d61b5bd..b7426ff 100644
|
|
|
f6d6d9 |
--- a/daemons/fenced/fenced_remote.c
|
|
|
f6d6d9 |
+++ b/daemons/fenced/fenced_remote.c
|
|
|
f6d6d9 |
@@ -1825,6 +1825,7 @@ request_peer_fencing(remote_fencing_op_t *op, peer_device_info_t *peer)
|
|
|
f6d6d9 |
op->state = st_exec;
|
|
|
f6d6d9 |
if (op->op_timer_one) {
|
|
|
f6d6d9 |
g_source_remove(op->op_timer_one);
|
|
|
f6d6d9 |
+ op->op_timer_one = 0;
|
|
|
f6d6d9 |
}
|
|
|
f6d6d9 |
|
|
|
f6d6d9 |
if (!((stonith_watchdog_timeout_ms > 0)
|
|
|
f6d6d9 |
--
|
|
|
f6d6d9 |
2.31.1
|
|
|
f6d6d9 |
|