Bug 600395

Summary:	allow_kill does not prevent eviction via disk
Product:	Red Hat Enterprise Linux 5	Reporter:	Lon Hohberger <lhh>
Component:	cman	Assignee:	Lon Hohberger <lhh>
Status:	CLOSED DUPLICATE	QA Contact:	Cluster QE <mspqa-list>
Severity:	medium	Docs Contact:
Priority:	low
Version:	5.5	CC:	clasohm, cluster-maint, edamato, twilkins
Target Milestone:	rc
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2011-01-25 15:11:50 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Lon Hohberger 2010-06-04 15:54:13 UTC

Description of problem:

When you set allow_kill to 0, the qdiskd master no longer asks CMAN to evict/fence the presumed-dead host.  However, it will still write an eviction notice to disk.  If the presumed-dead host reads in the eviction notice, it will reboot itself.

If storage is flaky (for example, exhibiting long delays for I/O), users may wish to disable all qdiskd failure-recovery actions in order to sabilize the cluster - this includes both CMAN *and* disk eviction.

Today, the former is possible, but the latter is not.

We need the ability to disable disk-based eviction.

Comment 1 Lon Hohberger 2010-06-04 15:56:07 UTC

There are two possibilities:

1) Simply do not write eviction notices, or

2) Do not act on eviction notices by reboot -- instead, restart qdiskd with a new incarnation #.

Comment 4 Lon Hohberger 2011-01-25 15:11:50 UTC


*** This bug has been marked as a duplicate of bug 602731 ***