854ff6077b
- Fix overall timeout calculation if watchdog and another fencing device share a topology level - Resolves: rhbz2187424
143 lines
5.9 KiB
Diff
143 lines
5.9 KiB
Diff
From 17cc49e1564b0ae55cc8212d14c5c055f88040da Mon Sep 17 00:00:00 2001
|
|
From: Klaus Wenninger <klaus.wenninger@aon.at>
|
|
Date: Tue, 14 Feb 2023 15:35:37 +0100
|
|
Subject: [PATCH] Fix: watchdog-fencing: terminate dangling timer before
|
|
watchdog-waiting
|
|
|
|
---
|
|
daemons/fenced/fenced_remote.c | 6 +++++-
|
|
1 file changed, 5 insertions(+), 1 deletion(-)
|
|
|
|
diff --git a/daemons/fenced/fenced_remote.c b/daemons/fenced/fenced_remote.c
|
|
index 5c3fe25e3..aab185adb 100644
|
|
--- a/daemons/fenced/fenced_remote.c
|
|
+++ b/daemons/fenced/fenced_remote.c
|
|
@@ -1,5 +1,5 @@
|
|
/*
|
|
- * Copyright 2009-2022 the Pacemaker project contributors
|
|
+ * Copyright 2009-2023 the Pacemaker project contributors
|
|
*
|
|
* The version control history for this file may have further details.
|
|
*
|
|
@@ -1702,6 +1702,10 @@ check_watchdog_fencing_and_wait(remote_fencing_op_t * op)
|
|
"client %s " CRM_XS " id=%.8s",
|
|
(stonith_watchdog_timeout_ms / 1000),
|
|
op->target, op->action, op->client_name, op->id);
|
|
+
|
|
+ if (op->op_timer_one) {
|
|
+ g_source_remove(op->op_timer_one);
|
|
+ }
|
|
op->op_timer_one = g_timeout_add(stonith_watchdog_timeout_ms,
|
|
remote_op_watchdog_done, op);
|
|
return TRUE;
|
|
--
|
|
2.39.0
|
|
|
|
From f2cc2a4277124230903a18713e50604a8f1842cd Mon Sep 17 00:00:00 2001
|
|
From: Klaus Wenninger <klaus.wenninger@aon.at>
|
|
Date: Wed, 1 Mar 2023 15:00:15 +0100
|
|
Subject: [PATCH] Refactor: watchdog-fencing: convenience function
|
|
pcmk__is_fencing_action
|
|
|
|
for consistency and add comment making clear why this block exits
|
|
with new timer set in any case
|
|
---
|
|
daemons/fenced/fenced_remote.c | 5 ++++-
|
|
1 file changed, 4 insertions(+), 1 deletion(-)
|
|
|
|
diff --git a/daemons/fenced/fenced_remote.c b/daemons/fenced/fenced_remote.c
|
|
index aab185adb..e0f8de057 100644
|
|
--- a/daemons/fenced/fenced_remote.c
|
|
+++ b/daemons/fenced/fenced_remote.c
|
|
@@ -1834,7 +1834,7 @@ request_peer_fencing(remote_fencing_op_t *op, peer_device_info_t *peer)
|
|
if (!((stonith_watchdog_timeout_ms > 0)
|
|
&& (pcmk__str_eq(device, STONITH_WATCHDOG_ID, pcmk__str_none)
|
|
|| (pcmk__str_eq(peer->host, op->target, pcmk__str_casei)
|
|
- && !pcmk__str_eq(op->action, "on", pcmk__str_none)))
|
|
+ && pcmk__is_fencing_action(op->action)))
|
|
&& check_watchdog_fencing_and_wait(op))) {
|
|
|
|
/* Some thoughts about self-fencing cases reaching this point:
|
|
@@ -1854,6 +1854,9 @@ request_peer_fencing(remote_fencing_op_t *op, peer_device_info_t *peer)
|
|
Otherwise the selection of stonith-watchdog-timeout at
|
|
least is questionable.
|
|
*/
|
|
+
|
|
+ /* coming here we're not waiting for watchdog timeout -
|
|
+ thus engage timer with timout evaluated before */
|
|
op->op_timer_one = g_timeout_add((1000 * timeout_one), remote_op_timeout_one, op);
|
|
}
|
|
|
|
--
|
|
2.39.0
|
|
|
|
From c4eb45a986f8865fc5e69350fd5b9f4b056d9d69 Mon Sep 17 00:00:00 2001
|
|
From: Klaus Wenninger <klaus.wenninger@aon.at>
|
|
Date: Tue, 14 Feb 2023 11:57:17 +0100
|
|
Subject: [PATCH] Fix: watchdog-fencing: correctly derive timeout with topology
|
|
|
|
up to now the timeout for watchdog-fencing was just added to
|
|
the overall timeout if the node to be fenced was visible and
|
|
reported back to the query.
|
|
---
|
|
daemons/fenced/fenced_remote.c | 28 +++++++++++++++++++++++++---
|
|
1 file changed, 25 insertions(+), 3 deletions(-)
|
|
|
|
diff --git a/daemons/fenced/fenced_remote.c b/daemons/fenced/fenced_remote.c
|
|
index e0f8de057..3b7ab05e9 100644
|
|
--- a/daemons/fenced/fenced_remote.c
|
|
+++ b/daemons/fenced/fenced_remote.c
|
|
@@ -969,8 +969,9 @@ advance_topology_level(remote_fencing_op_t *op, bool empty_ok)
|
|
return pcmk_rc_ok;
|
|
}
|
|
|
|
- crm_info("All fencing options targeting %s for client %s@%s failed "
|
|
+ crm_info("All %sfencing options targeting %s for client %s@%s failed "
|
|
CRM_XS " id=%.8s",
|
|
+ (stonith_watchdog_timeout_ms > 0)?"non-watchdog ":"",
|
|
op->target, op->client_name, op->originator, op->id);
|
|
return ENODEV;
|
|
}
|
|
@@ -1434,8 +1435,17 @@ stonith_choose_peer(remote_fencing_op_t * op)
|
|
&& pcmk_is_set(op->call_options, st_opt_topology)
|
|
&& (advance_topology_level(op, false) == pcmk_rc_ok));
|
|
|
|
- crm_notice("Couldn't find anyone to fence (%s) %s using %s",
|
|
- op->action, op->target, (device? device : "any device"));
|
|
+ if ((stonith_watchdog_timeout_ms > 0)
|
|
+ && pcmk__is_fencing_action(op->action)
|
|
+ && pcmk__str_eq(device, STONITH_WATCHDOG_ID, pcmk__str_none)
|
|
+ && node_does_watchdog_fencing(op->target)) {
|
|
+ crm_info("Couldn't contact watchdog-fencing target-node (%s)",
|
|
+ op->target);
|
|
+ /* check_watchdog_fencing_and_wait will log additional info */
|
|
+ } else {
|
|
+ crm_notice("Couldn't find anyone to fence (%s) %s using %s",
|
|
+ op->action, op->target, (device? device : "any device"));
|
|
+ }
|
|
return NULL;
|
|
}
|
|
|
|
@@ -1531,6 +1541,18 @@ get_op_total_timeout(const remote_fencing_op_t *op,
|
|
continue;
|
|
}
|
|
for (device_list = tp->levels[i]; device_list; device_list = device_list->next) {
|
|
+ /* in case of watchdog-device we add the timeout to the budget
|
|
+ regardless of if we got a reply or not
|
|
+ */
|
|
+ if ((stonith_watchdog_timeout_ms > 0)
|
|
+ && pcmk__is_fencing_action(op->action)
|
|
+ && pcmk__str_eq(device_list->data, STONITH_WATCHDOG_ID,
|
|
+ pcmk__str_none)
|
|
+ && node_does_watchdog_fencing(op->target)) {
|
|
+ total_timeout += stonith_watchdog_timeout_ms / 1000;
|
|
+ continue;
|
|
+ }
|
|
+
|
|
for (iter = op->query_results; iter != NULL; iter = iter->next) {
|
|
const peer_device_info_t *peer = iter->data;
|
|
|
|
--
|
|
2.39.0
|
|
|