Bug 1247303

Summary:	rabbitmq-cluster agent needs to forget stopped cluster nodes
Product:	Red Hat Enterprise Linux 7	Reporter:	David Vossel <dvossel>
Component:	resource-agents	Assignee:	Oyvind Albrigtsen <oalbrigt>
Status:	CLOSED ERRATA	QA Contact:	Leonid Natapov <lnatapov>
Severity:	urgent	Docs Contact:
Priority:	urgent
Version:	7.2	CC:	agk, cfeist, cluster-maint, cpaquin, djansa, dmaley, fdinitto, felipe.alfaro, jeckersb, jraju, jruemker, oalbrigt, ochalups, phagara, plemenko, royoung, snagar
Target Milestone:	rc	Keywords:	ZStream
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:	resource-agents-3.9.5-60.el7	Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:
Clones:	1311180 (view as bug list)		Environment:
Last Closed:	2016-11-03 23:57:47 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:
Bug Depends On:	1311025
Bug Blocks:	1311180

Description David Vossel 2015-07-27 18:09:58 UTC

Description of problem:

After fencing, a rabbitmq-cluster resource instance may not be able to rejoin the rabbitmq cluster.

Version-Release number of selected component (if applicable):


How reproducible:
We have seen this in customer deployments. This is not something that has a clear set of steps to reproduce.

Steps to Reproduce:
1. fence a osp cluster node.

Actual results:
rabbitmq-cluster instance on fenced node can not join rabbitmq cluster once node comes back online.

Expected results:
rabbitmq-cluster instance on fenced node should be able to start successfully after the node comes back online.

Comment 1 David Vossel 2015-07-27 18:11:26 UTC

upstream patch.

https://github.com/davidvossel/resource-agents/commit/f82b1f794df8af69371572dcd83e76792953ada9

Comment 5 Fabio Massimo Di Nitto 2015-09-01 03:20:37 UTC

John,

did you have a chance to test this patch again? is it still required?

Comment 6 John Eckersberg 2015-09-01 12:47:40 UTC

I still need to go back and test this.  The libvirt fence agent wasn't working for me, but David realized it was because the hostname and the libvirt domain name didn't match.  So I need to try that and see if I can get it to fence properly.

Leaving NEEDINFO for now to remind me to actually try it :)

Comment 7 Fabio Massimo Di Nitto 2015-09-03 03:35:31 UTC

(In reply to John Eckersberg from comment #6)
> I still need to go back and test this.  The libvirt fence agent wasn't
> working for me, but David realized it was because the hostname and the
> libvirt domain name didn't match.  So I need to try that and see if I can
> get it to fence properly.
> 
> Leaving NEEDINFO for now to remind me to actually try it :)

That´s fine, but we don´t have heaps of time to get this into 7.2. If you need help to configure fencing or other bits, please just contact me or #cluster.

Comment 9 Oyvind Albrigtsen 2015-11-20 12:37:36 UTC

Works as expected.

Tested with:
# service corosync stop &
# killall -9 corosync

on the node that should be fenced.

First notify:
rabbitmq-cluster(rmq)[15072]:   2015/11/20_13:30:10 NOTICE: Forgetting stopped node rabbit@rhel7-1
rabbitmq-cluster(rmq)[15072]:   2015/11/20_13:30:11 WARNING: Unable to forget offline node rabbit@rhel7-1.

Second notify:
rabbitmq-cluster(rmq)[15284]:   2015/11/20_13:30:26 NOTICE: Forgetting stopped node rabbit@rhel7-1

Comment 13 Fabio Massimo Di Nitto 2016-02-23 14:08:09 UTC

*** Bug 1299923 has been marked as a duplicate of this bug. ***

Comment 21 Leonid Natapov 2016-03-09 09:30:57 UTC

resource-agents-3.9.5-67.el7

Looks good:

After fencing controller-0 I see the following on controller-1 and controller-2:

overcloud-controller-1 
-------------------------
rabbitmq-cluster(rabbitmq)[32276]:	2016/03/09_09:19:35 NOTICE: Forgetting stopped node rabbit@overcloud-controller-0
rabbitmq-cluster(rabbitmq)[32276]:	2016/03/09_09:19:35 WARNING: Unable to forget offline node rabbit@overcloud-controller-0.

overcloud-controller-2
-------------------------
rabbitmq-cluster(rabbitmq)[16925]:	2016/03/09_09:19:35 NOTICE: Forgetting stopped node rabbit@overcloud-controller-0


Works as expected.

Comment 22 Peter Lemenkov 2016-06-24 13:33:09 UTC

(In reply to Leonid Natapov from comment #21)
> resource-agents-3.9.5-67.el7
> 
> Looks good:
> 
> After fencing controller-0 I see the following on controller-1 and
> controller-2:
> 
> overcloud-controller-1 
> -------------------------
> rabbitmq-cluster(rabbitmq)[32276]:	2016/03/09_09:19:35 NOTICE: Forgetting
> stopped node rabbit@overcloud-controller-0
> rabbitmq-cluster(rabbitmq)[32276]:	2016/03/09_09:19:35 WARNING: Unable to
> forget offline node rabbit@overcloud-controller-0.
> 
> overcloud-controller-2
> -------------------------
> rabbitmq-cluster(rabbitmq)[16925]:	2016/03/09_09:19:35 NOTICE: Forgetting
> stopped node rabbit@overcloud-controller-0
> 
> 
> Works as expected.

If someone still sees this issue, then please test this package:

resource-agents-3.9.5-76.el7

Comment 24 errata-xmlrpc 2016-11-03 23:57:47 UTC

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://rhn.redhat.com/errata/RHBA-2016-2174.html