Bug 1126362 - stonigh resources should stop only after other resources
Summary: stonigh resources should stop only after other resources
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: Red Hat Enterprise Linux 7
Classification: Red Hat
Component: pacemaker
Version: 7.0
Hardware: Unspecified
OS: Unspecified
unspecified
high
Target Milestone: rc
: ---
Assignee: Andrew Beekhof
QA Contact: cluster-qe@redhat.com
URL:
Whiteboard:
Depends On:
Blocks: 1126880 1126884
TreeView+ depends on / blocked
 
Reported: 2014-08-04 09:32 UTC by Fabio Massimo Di Nitto
Modified: 2014-08-05 14:02 UTC (History)
2 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
: 1126880 (view as bug list)
Environment:
Last Closed: 2014-08-04 14:19:06 UTC


Attachments (Terms of Use)
crm_report (61.44 KB, application/x-bzip)
2014-08-04 09:32 UTC, Fabio Massimo Di Nitto
no flags Details

Description Fabio Massimo Di Nitto 2014-08-04 09:32:47 UTC
Created attachment 923818 [details]
crm_report

I was randomly monitoring crm_mon -r while issuing a pcs cluster stop --all on another node and I noticed that all stonith resources were stopped before other resources.

this could potentially be an issue if a resource fail to stop and we need to complete recovery of the node to shutdown everything else properly.

I attached the crm_report just in case it was a visual artifact ;)

Comment 1 David Vossel 2014-08-04 13:19:31 UTC
(In reply to Fabio Massimo Di Nitto from comment #0)
> Created attachment 923818 [details]
> crm_report
> 
> I was randomly monitoring crm_mon -r while issuing a pcs cluster stop --all
> on another node and I noticed that all stonith resources were stopped before
> other resources.

This should be fine.

As long as we haven't gone in and explicitly disabled a fencing device or setup constraints in a way that prevent a fencing device from executing on a node, it doesn't matter if crm_mon shows that the device is down or not.  The device will still be available to stonith.

Think of all stonith resources as a special type of clone resource.  We only show one instance of the stonith device running, and that's the location we expect to execute the stonith operation for that device. but... if the stonith device is not placed anywhere or the device's location is the target of the fencing operation, we'll attempt to use the same device on another node.

-- Vossel

> 
> this could potentially be an issue if a resource fail to stop and we need to
> complete recovery of the node to shutdown everything else properly.
> 
> I attached the crm_report just in case it was a visual artifact ;)


Note You need to log in before you can comment on or make changes to this bug.