Bug 1147252

Summary:	Pacemaker OCF volume Resource Agent fails when bricks are in different domain to the system hostname.
Product:	[Community] GlusterFS	Reporter:	Niels de Vos <ndevos>
Component:	scripts	Assignee:	Steven Hale <bugzilla>
Status:	CLOSED DEFERRED	QA Contact:
Severity:	medium	Docs Contact:
Priority:	unspecified
Version:	mainline	CC:	atumball, bugs, sarumuga
Target Milestone:	---	Keywords:	EasyFix, Patch, Triaged
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:	1130763	Environment:
Last Closed:	2019-05-11 00:32:57 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:
Bug Blocks:	1130763

Description Niels de Vos 2014-09-28 12:45:24 UTC

+++ This bug was initially created as a clone of Bug #1130763 +++
+++                                                           +++
+++ This bug is used to make the change in the master branch. +++
+++                                                           +++

Description of problem:

I'm setting up a four-node cluster using PCS/Pacemaker/Corosync on CentOS7.  I have installed glusterfs-resource-agents.noarch 0:3.5.2-1.el7 from the main GlusterFS yum repository.

As is common when setting up a cluster, I am using a storage network separate from the back channel network.  I have added entries to my local nameserver to reflect these networks.  The hostnames are as follows:

node?.bcn
node?.sn

When trying to start a volume using the "/usr/lib/ocf/resource.d/glusterfs/volume" resource, the volume fails to start because the script does not recognise that the bricks are in a different domain to the system hostname.

I have fixed this bug and will attach a patch file to this report.  The fix simply involves editing a regular expression and a file path.

I also had to update the volume path to reflect that volume information is now stored in /var/lib/ rather than /etc.

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

glusterfs-resource-agents.noarch 0:3.5.2-1.el7

How reproducible:

Always

Steps to Reproduce:
1. Create a GlusterFS volume where the bricks are in a different domain to the central system hostname.
2. Try to start the volume using the OCF volume resource agent for Pacemaker.

Actual results:

The volume does not start.  The script always thinks the volume is already running.

Expected results:

The volume should start normally.

Additional info:

Patchfile attached to bug 1130763 as attachment #927478 [details].

Comment 1 Niels de Vos 2014-09-28 12:57:34 UTC

Steven, do you want to send your patch through Gerrit?

1. create an account on https://review.gluster.org
2. checkout the sources
   $ ssh://steven.org/glusterfs
3. create a branch for this change
   $ git checkout -t -b bug-1147252 origin/master
4. apply the patch
   $ cd extras/ocf
   $ curl 'https://bugzilla.redhat.com/attachment.cgi?id=927478' | git apply
   $ cd -
5. commit the change
   $ git commit -seam 'A suitable one line subject'
6. check the log and patch
   $ git show
7. (optional) make corrections to the commit message
   $ git commit --amend
8. post the patch for review
   $ ./rfc.sh

After that has been done, this bugreport should automatically get a URL for the change.

Comment 2 Anand Avati 2014-10-10 11:11:03 UTC

REVIEW: http://review.gluster.org/8922 (Fix for Pacemaker clusters using bricks on separate network interfaces.) posted (#1) for review on master by Steven Hale (stevenhale)

Comment 3 Amar Tumballi 2019-05-11 00:32:57 UTC

We haven't seen the review comments addressed at all in above patch. As none of the current developers can work on it, marking it DEFERRED, happy to get help and close it. If no one bothers, we will revisit it after sometime.