|
 |
d42a06 |
From 47d75f8de9dc912da035805f141c674885ce432f Mon Sep 17 00:00:00 2001
|
|
|
d42a06 |
From: John Eckersberg <jeckersb@redhat.com>
|
|
|
d42a06 |
Date: Thu, 16 Jan 2020 10:20:59 -0500
|
|
|
d42a06 |
Subject: [PATCH] rabbitmq-cluster: ensure we delete nodename if stop action
|
|
|
d42a06 |
fails
|
|
|
d42a06 |
|
|
|
d42a06 |
If the stop action fails, we want to remove the nodename from the crm
|
|
|
d42a06 |
attribute. Currently it is possible for the stop action to fail but
|
|
|
d42a06 |
the rabbitmq server does actually stop. This leaves the attribute
|
|
|
d42a06 |
still present. This means if the entire rabbitmq cluster is stopped,
|
|
|
d42a06 |
it is not possible to start the cluster again because the first node
|
|
|
d42a06 |
to start will think there is at least one other node running. Then
|
|
|
d42a06 |
the node tries to join an existing cluster instead of rebootstrapping
|
|
|
d42a06 |
the cluster from a single node.
|
|
|
d42a06 |
---
|
|
|
d42a06 |
heartbeat/rabbitmq-cluster | 2 ++
|
|
|
d42a06 |
1 file changed, 2 insertions(+)
|
|
|
d42a06 |
|
|
|
d42a06 |
diff --git a/heartbeat/rabbitmq-cluster b/heartbeat/rabbitmq-cluster
|
|
|
d42a06 |
index 7837e9e3c..a9ebd37ad 100755
|
|
|
d42a06 |
--- a/heartbeat/rabbitmq-cluster
|
|
|
d42a06 |
+++ b/heartbeat/rabbitmq-cluster
|
|
|
d42a06 |
@@ -552,6 +552,7 @@ rmq_stop() {
|
|
|
d42a06 |
|
|
|
d42a06 |
if [ $rc -ne 0 ]; then
|
|
|
d42a06 |
ocf_log err "rabbitmq-server stop command failed: $RMQ_CTL stop, $rc"
|
|
|
d42a06 |
+ rmq_delete_nodename
|
|
|
d42a06 |
return $rc
|
|
|
d42a06 |
fi
|
|
|
d42a06 |
|
|
|
d42a06 |
@@ -565,6 +566,7 @@ rmq_stop() {
|
|
|
d42a06 |
break
|
|
|
d42a06 |
elif [ "$rc" -ne $OCF_SUCCESS ]; then
|
|
|
d42a06 |
ocf_log info "rabbitmq-server stop failed: $rc"
|
|
|
d42a06 |
+ rmq_delete_nodename
|
|
|
d42a06 |
exit $OCF_ERR_GENERIC
|
|
|
d42a06 |
fi
|
|
|
d42a06 |
sleep 1
|