Problem/var/lib/nfs/etab is not cleared on system panic causing incorrect operation of VCS Share Agent
VCS Share agent not working correctly due to presence of /var/lib/nfs/etab file after system panic.
When a Linux system panics, /var/lib/nfs/etab file is not cleared at reboot. This results in VCS incorrectly seeing share resources as online, and may cause service group to not failover to other node (service group goes into partial state).
From Veritas Cluster Server Release Notes:
To ensure proper monitoring by the Share agent, verify that the file /var/lib/nfs/etab is clear for Red Hat upon system reboot.
Sharetab entries remains when node crashes and subsequent reboot of node causes Share resource to go ONLINE in its first probe even if dependent resources are OFFLINE. To fix this issue; a new attribute “NFSRes” is added for Share resource. This attribute points to the NFS Resource configured on the system. During the first probe of Share resource if NFS resource is OFFLINE and sharetab entries are present for the resource, then those entries are removed (as these are invalid entries) and status of Share resource is returned OFFLINE.
Modify rc.sysinit file to include 'rm /var/lib/nfs/etab' command so that the file is removed on reboot.
Symantec Engineering is aware of this limitation and will address this issue (e1789008) via an hotfix patch on top of VCS 5.0MP3RP2.