Bug 996576

Summary:	Unfence at cluster startup with fence_scsi
Product:	Red Hat Enterprise Linux 7	Reporter:	Jaroslav Kortus <jkortus>
Component:	pacemaker	Assignee:	Andrew Beekhof <abeekhof>
Status:	CLOSED CURRENTRELEASE	QA Contact:	Cluster QE <mspqa-list>
Severity:	high	Docs Contact:
Priority:	high
Version:	7.0	CC:	cluster-maint, dvossel
Target Milestone:	rc
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:	pacemaker-1.1.10-16.el7	Doc Type:	Bug Fix
Doc Text:	Cause: Fencing is configured in Pacemaker not cman Consequence: The call to fence_node -U in the cman init script is a no-op Fix: Support the concept of automated unfencing in pacemaker Result: Messages such as these should be visible in the log files during startup: stonith-ng[7472]: notice: cib_device_update: Unfencing ourselves with fence_scsi (Fabric) ... stonith-ng[7472]: notice: log_operation: Operation 'on' [7485] (call -1 from stonith-ng) for host 'pcmk-5' with device 'Fabric' returned: 0	Story Points:	---
Clone Of:
Clones:	996850 (view as bug list)		Environment:
Last Closed:	2014-06-13 12:02:56 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:	1009067, 1009069
Bug Blocks:

Description Jaroslav Kortus 2013-08-13 13:08:01 UTC

Description of problem:
Pure pacemaker + fence_scsi does not provide complete fencing solution. The node needs to register with the device on which it's supposed to write prior to any write action.

Until now this was done with cman + <unfence> mechanics which triggered fence_node -U at startup. Currently with pure pacemaker+corosync combo there is no trick to fire the same operation and get the node registered.

Please provide a way to do such an action and sucessfully register with the device.

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


How reproducible:
always

Steps to Reproduce:
1. setup a cluster with fence_scsi and compliant shared disk
2. startup the cluster
3. verify keys on the devices (sg_persist)

Actual results:
no keys present

Expected results:
keys present as in cman 

Additional info:

Comment 2 Andrew Beekhof 2013-08-14 00:45:35 UTC

I should have a functioning patch later today.
We'll also need to clone this for RHEL6.5 since 'fence_node -U' is a no-op when cluster.conf only references pcmk_fence.

In any case, stonithd wouldn't be running to allow it to succeed.
The same patch will address the issue in both versions.

Comment 3 Andrew Beekhof 2013-08-14 07:01:02 UTC

A related patch has been committed upstream:
  https://github.com/beekhof/pacemaker/commit/8af0d0c

with subject:

   Feature: Fencing: Support agents that need the host to be unfenced at startup

Further details (if any):

Comment 8 Ludek Smid 2014-06-13 12:02:56 UTC

This request was resolved in Red Hat Enterprise Linux 7.0.

Contact your manager or support representative in case you have further questions about the request.