File 0029-Medium-Raid1-Ignore-transient-devices-after-stopping.patch of Package resource-agents.8843
From 3b432453fcfda69e6fdfea6935e3bdfee281639a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Michal=20Koutn=C3=BD?= <mkoutny@suse.com>
Date: Tue, 30 Jan 2018 12:01:53 +0100
Subject: [PATCH 29/29] Medium: Raid1: Ignore transient devices after stopping
a device (bsc#1077416)
There may be a transient invalid device after we stop the MD due to
other programs processing uevent, the original device is stopped though
(`mdadm --stop` only returns when MD is stopped).
The idea suggested by Neil Brown <nfbrown@suse.com>.
---
heartbeat/Raid1 | 12 ++++++++++--
1 file changed, 10 insertions(+), 2 deletions(-)
diff --git a/heartbeat/Raid1 b/heartbeat/Raid1
index f7cb1eaf..a3c99402 100755
--- a/heartbeat/Raid1
+++ b/heartbeat/Raid1
@@ -384,8 +384,16 @@ raid1_monitor_one() {
2) ocf_exit_reason "$mddev has failed."
return $OCF_ERR_GENERIC
;;
- 4) ocf_exit_reason "mdadm failed on $mddev."
- return $OCF_ERR_GENERIC
+ 4)
+ if [ "$__OCF_ACTION" = "stop" ] ; then
+ # There may be a transient invalid device after
+ # we stop MD due to uevent processing, the
+ # original device is stopped though.
+ return $OCF_NOT_RUNNING
+ else
+ ocf_exit_reason "mdadm failed on $mddev."
+ return $OCF_ERR_GENERIC
+ fi
;;
*) ocf_exit_reason "mdadm returned an unknown result ($rc)."
return $OCF_ERR_GENERIC
--
2.16.2
