Red Hat Bugzilla – Bug 489809
Broken device detection for DRAC3 ERA/O in fence_drac
Last modified: 2016-04-26 09:45:11 EDT
Created attachment 334870 [details]
Patch to fix operation of fence_drac on Dell embedded DRAC3 ERA/O cards
Description of problem:
Fencing agent for Dell Remote Access Controller (DRAC) ERA/O (DRAC3 variant) has broken device detection. The fencing agent looks for the following regular expression:
/Dell Embedded Remote Access Controller \(ERA\)\nFirmware Version/
The actual device string (with latest firmware), is:
Dell Embedded Remote Access Controller (ERA/O)
Firmware Version 3.37 (Build 08.13)
Thus, the regular expression match should be:
/Dell Embedded Remote Access Controller \(ERA\/O\)\nFirmware Version/
Version-Release number of selected component (if applicable):
All versions of cman up to and including 2.0.98 in RHEL 5.3
Steps to Reproduce:
1. Set up a cluster of nodes, one of which has a DRAC3 ERA/O in it (e.g. Dell PowerEdge 1650)
2. Pull the plug on the DRAC3 node.
3. Cluster will hang. The surviving nodes will try to fence using DRAC but won't be able to identify the DRAC card, and will keep failing.
Cluster hangs indefinitely waiting for the node to get fenced.
Node gets fenced and cluster resumes operation.
Patch to fix this is attached.
Thanks for patch.
But I would like to ask you if you can help us with writing new fence agent (fence_drac5.py) to support also your device. I will try to write it (using old agent) but I don't have device to test it. I believe that we can do that in 2-3 iterations (I will need just verbose output).
Sure, I'll be happy to test it for you and forward any output back to you. Please email me the instructions.
I see this patch hasn't made it into RHEL5.4 (cman-2.0.115-1.el5_4.3). Is it likely to get pushed out any time soon? The current fence_drac agent completely fails to work on the DRAC 3 ERA/O management modules without the provided patch.
Patch changed so it should not break backward compatibility.
if (/Dell Embedded Remote Access Controller \(ERA(\/O)?\)\nFirmware Version/m)
If it is possible, please try test build: cman-2.0.115-18.el5
Gladly, where can I get the new package?
@Gordan: Can you please test a new package and send results?
Any feedback on this yet?
Sorry, forgot to get back to you about this. The updated version linked above has been working absolutely fine for me.
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.