RHEL 6 High Availability cluster node gets fenced after its corosync process is sitting in function audit_log_start during the period where it failed to send its token
Issue
- A node gets fenced frequently in our cluster
- We keep seeing nodes fenced, and the
ha-resourcemon
'sps
output showscorosync
sitting in functionaudit_log_start
during the window where it should be sending tokens but is apparently unresponsive - Why is
corosync
getting stuck behindaudit
causing a node to get fenced?
Environment
- Red Hat Enterprise Linux (RHEL) 6 with the High Availability Add-On
audit
- Some sort of
audit
watch or rule that may trigger on corosync's operations and activities
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.