Cluster services lock up and no fencing occurs in a two node cluster after a token loss in RHEL 5 or processor failure in RHEL 6
Issue
- After a token loss (RHEL 5), processor failure (RHEL 6), or other type of node failure (either version), the remaining node does not fence the missing node. Cluster services such as
clvmd
,GFS
orGFS2
, andrgmanager
are blocked or unresponsive. - After fencing, resources or resource group, or fence agents do not start on second node.
Environment
- Red Hat Enterprise Linux (RHEL) 5 or 6 with the High Availability Add On with RGManager
- Red Hat Enterprise Linux (RHEL) 6 with the High Availability Add On with Pacemaker
- two-node cluster without a QDisk
- two_node not enabled in
/etc/cluster/cluster.conf
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.