Bug 545682

Summary: fence_ilo fails to reboot, possibly timing problem with ilo2 1.70
Product: [Retired] Red Hat Cluster Suite Reporter: Michael Kearey <mkearey>
Component: fenceAssignee: Marek Grac <mgrac>
Status: CLOSED ERRATA QA Contact: Cluster QE <mspqa-list>
Severity: urgent Docs Contact:
Priority: urgent    
Version: 4CC: cfeist, cluster-maint, djansa, edamato, jason.nelson, jwest, tao
Target Milestone: rcKeywords: ZStream
Target Release: ---   
Hardware: All   
OS: Linux   
Whiteboard:
Fixed In Version: fence-1.32.68-5.el4 Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2011-02-16 16:15:33 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:
Bug Depends On:    
Bug Blocks: 562261, 562263    
Attachments:
Description Flags
Proposed patch none

Description Michael Kearey 2009-12-09 05:01:45 UTC 
Description of problem:

When a node in the cluster is fenced, the fencing agent turns the power off but then fails to turn the power back on. The fencing action is still regarded as "successful".

This appears to be same issue as descibed in Bug 507514 at:
https://bugzilla.redhat.com/show_bug.cgi?id=507514

The Ilo2 versions this happened were 1.79 and 1.80.

How reproducible:

Seems to happen every time.

Steps to Reproduce:

Disconnect one node from cluster by executing e.g. "ifdown bond0" and let the cluster fence the node.

Actual results:

The server is powered off. BUT will not be powered on again as is required with a reboot action from fence agent.

Expected results:

The server is powered on after being powered off. ie it is rebooted.

Additional info:

I have tested adding "time.sleep(5)" statement to fencing.py as described in bugzille. With sleep(5) the fencing works again as expected, and the node is powered back on as it should be.

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


Additional Information:

Changes have been made to releive the problems as per :

https://bugzilla.redhat.com/show_bug.cgi?id=507514#c21

We would like to see if the changes are possible for RHEL4
Comment 1 Marek Grac 2009-12-09 15:23:10 UTC 
It can be fixed in next update, unfortunately general solution (as in RHEL5) is not possible, so I will just add a special option for hp ilo. If you need timing feature also for some other agents, please let me know.
Comment 3 Marek Grac 2010-01-20 18:07:45 UTC 
Created attachment 385741 [details]
Proposed patch

Option -G (power_wait on stdin/in cluster.conf) was added for fence_ilo. It defines how many seconds will fence agent wait after issuing power OFF command in 'reboot' action (only one used by fence daemon). Feel free to test this patch. It does not change default behaviour but '-G 5' should be equal to time.sleep(5). But this won't slow down other fence agents or older iLO that are fast enough.
Comment 15 Marek Grac 2010-07-14 12:54:33 UTC 
*** Bug 608500 has been marked as a duplicate of this bug. ***
Comment 17 errata-xmlrpc 2011-02-16 16:15:33 UTC 
An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on therefore solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.

http://rhn.redhat.com/errata/RHSA-2011-0266.html