Quorum disk does not log qdisk owner when master_wins is set on Red Hat Enterprise Linux 5
Issue
-
I just did an "ifdown eth0" on the active node and watched /var/log/messages on the inactive node and it appears the resource manager crashed and restarted. I was able to do this twice in a row.
-
I do not see any log messages indicating which node is the qdisk master or fencing when I ifdown the cluster heartbeat network interface.
Environment
- Red Hat Enterprise Linux 5 (RHEL5) Advanced Platform
- Observed on RHEL5 update 4, kernel-2.6.18-164.11.1.el5
- Red Hat Cluster Suite configured in a 2 node cluster
- Quorum disk is configured in "master_wins" mode. An example /etc/cluster/cluster.conf segment:
<clusternodes> <clusternode name="rhel5ha1.redhat.net" nodeid="1"> <fence> <method name="fence_node1"> <device domain="rhel5ha1" name="fence_xvm"/> </method> </fence> </clusternode> <clusternode name="rhel5ha2.redhat.net" nodeid="2"> <fence> <method name="fence_node2"> <device domain="rhel5ha2" name="fence_xvm"/> </method> </fence> </clusternode> </clusternodes> <quorumd interval="3" label="thedisk" tko="3" votes="1" master_wins="1"/> <cman expected_votes="3"/>
- Services are managed by rgmanager
- Quorum disk is configured in "master_wins" mode. An example /etc/cluster/cluster.conf segment:
- When network communication is broken between nodes using "ifdown", the node is successfully rebooted but log messages showing the quorum disk mastership are not displayed.
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.