A node is fenced when an LVM-activate resource fails during OS shutdown in a Pacemaker cluster
Issue
-
An
ocf:heartbeat:LVM-activate
resource fails to stop during an OS reboot or shutdown, causing the node to be fenced.Nov 24 17:47:25 fastvm-rhel-8-0-24 systemd[1]: Stopped Logout off all iSCSI sessions on shutdown. Nov 24 17:47:25 fastvm-rhel-8-0-24 LVM-activate(lvm)[1989]: ERROR: Volume group "test_vg1" not found. Cannot process volume group test_vg1 Nov 24 17:47:25 fastvm-rhel-8-0-24 LVM-activate(lvm)[1989]: ERROR: test_vg1: failed to deactivate. Nov 24 17:47:25 fastvm-rhel-8-0-24 pacemaker-execd[1875]: notice: lvm_stop_0[1989] error output [ ocf-exit-reason:test_vg1: failed to deactivate. ] Nov 24 17:47:25 fastvm-rhel-8-0-24 pacemaker-controld[1878]: notice: Result of stop operation for lvm on node2: error
-
An
ocf:heartbeat:LVM
resource stops successfully during an OS reboot or shutdown, but it logs a "volume group not found" error.Nov 28 00:30:34 fastvm-rhel-7-6-22 systemd: Stopped Logout off all iSCSI sessions on shutdown. Nov 28 00:30:39 fastvm-rhel-7-6-22 LVM(lvm)[3976]: INFO: Volume group test_vg1 not found Nov 28 00:30:39 fastvm-rhel-7-6-22 lrmd[2069]: notice: lvm_stop_0:3976:stderr [ Volume group "test_vg1" not found ] Nov 28 00:30:39 fastvm-rhel-7-6-22 lrmd[2069]: notice: lvm_stop_0:3976:stderr [ Cannot process volume group test_vg1 ] Nov 28 00:30:39 fastvm-rhel-7-6-22 crmd[2072]: notice: Result of stop operation for lvm on node2: 0 (ok)
-
The
shutdown-lock
cluster property works fine when Pacemaker is stopped (e.g., withpcs cluster stop
), but it fails when a node is rebooted or shut down.
Environment
- Red Hat Enterprise Linux 7 (with the High Availability Add-on)
- Red Hat Enterprise Linux 8 (with the High Availability Add-on)
- May be more likely with LVM physical volumes presented via iSCSI
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.