Cluster node is fenced during service relocation
Issue
When clustered service is about to relocate (either manual relocation or due to failure) the filesystem resource is not unmounted because it is used by application. Service ends up in failed state instead of relocating because it is unable to unmount fs (it is used by app) so LVM is unable to deactivate VG.
In older versions of resource-agents there was a fuser directive in /usr/share/cluster/utils/lib-fs.sh which works but the directive was removed in official release.
If the self_fence is enabled the node is fenced and service relocates but this happens also during manual relocation and customer is not very happy about this behaviour.
Environment
- Red Hat Enterprise Linux 6
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.