Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.

For bugs related to Red Hat Enterprise Linux 5 product line. The current stable release is 5.10. For Red Hat Enterprise Linux 6 and above, please visit Red Hat JIRA https://issues.redhat.com/secure/CreateIssue!default.jspa?pid=12332745 to report new issues.

Bug 461065

Summary:	RFE: GFS should accept "onerror=panic" and "onerror=withdraw" or similar mount option
Product:	Red Hat Enterprise Linux 5	Reporter:	Vinny Valdez <vvaldez>
Component:	gfs-utils	Assignee:	Robert Peterson <rpeterso>
Status:	CLOSED DUPLICATE	QA Contact:	Cluster QE <mspqa-list>
Severity:	medium	Docs Contact:
Priority:	medium
Version:	5.2	CC:	edamato, swhiteho
Target Milestone:	rc	Keywords:	FutureFeature
Target Release:	5.5
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Enhancement
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2009-08-05 13:56:30 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 Vinny Valdez 2008-09-03 22:19:13 UTC

Description of problem:
RFE: Currently the debug options seems to be the only way to cause the
system to panic on a GFS error.  The scenario I am testing is when a
cluster node loses all access to storage.  Currently it withraws, but to
ensure the system is fenced the debug option can cause it to panic.

Version-Release number of selected component (if applicable):
5.2
gfs-utils-0.1.17-1.el5


How reproducible:
Everytime

Steps to Reproduce:
1. Setup GFS
2. Remove all storage access from a node
3. System withdraws GFS, but stays operational in the cluster
  
Actual results:
If the node is running services, and cluster communications are working fine, and the services do not fail over.

Adding "debug" to the mount option allows the system to panic instead.
This causes it to fence, and everything else falls into place.

Expected results:
If a system loses access to the storage and thus GFS, it should not remain operational in the cluster.

An option to panic the node on any GFS errors as documented in the mount man page for other filesystems would be nice.  This ensures the system is fenced and services fail over.

Comment 1 Steve Whitehouse 2008-12-10 15:15:09 UTC

We should probably consider doing this in GFS2 first, and the maybe back port to GFS.

Comment 3 Robert Peterson 2009-08-05 13:56:30 UTC

Now that we've decided how to handle bug #488499, and have a
working prototype patch for it, it clearly implements the RFE
in this bugzilla.  So I'm closing this bug as a duplicate of
that one.

*** This bug has been marked as a duplicate of bug 488499 ***