Description of problem: Currently it doesn't choose either of them and just fails to initialize. Feb 25 11:32:29 taft-04 qdiskd[23382]: <crit> Initialization failed Version-Release number of selected component (if applicable): 2.6.18-128.el5 cman-2.0.98-1.el5
It should also give a warning about 2 being found.
http://git.fedorahosted.org/git/?p=cluster.git;a=commit;h=992658141fa495d902715fdba44f74dcbca8c8c8
Fix verified in cman-2.0.99-1.el5. Mar 17 11:18:19 taft-01 qdiskd[7705]: <warning> 2 matches found for label 'TAFT'; please use 'device=' instead! Mar 17 11:18:19 taft-01 qdiskd[7705]: <info> Quorum Partition: /dev/disk/by-id/scsi-3600805f3000a05b0000000005022000e-part1 Label: TAFT Mar 17 11:18:19 taft-01 qdiskd[7706]: <info> Quorum Daemon Initializing Mar 17 11:18:22 taft-01 qdiskd[7706]: <info> Heuristic: 'ping -c3 -t5 sts.lab.msp.redhat.com' UP Mar 17 11:18:29 taft-01 qdiskd[7706]: <info> Initial score 1/1 Mar 17 11:18:29 taft-01 qdiskd[7706]: <info> Initialization complete Mar 17 11:18:29 taft-01 openais[7633]: [CMAN ] quorum device registered Mar 17 11:18:29 taft-01 qdiskd[7706]: <notice> Score sufficient for master operation (1/1; required=1); upgrading Mar 17 11:18:35 taft-01 qdiskd[7706]: <info> Assuming master role
Cause: Having two or more qdisk partitions with the same label available to a node Consequence: Qdiskd selects one at random, with no warning. This may be the wrong quorum disk for a given cluster, causing split-brain behavior. Fix: Add a warning when two quorum disk partitions with the same label are found. Result: Warnings are now placed in the logs if qdiskd detects more than one device with the same label. Administrators should only ever have one quorum disk with a given label available to any given cluster node.
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-2009-1341.html