The "lvmlockd" pacemaker resource enters a "FAILED" state when the lvmlockd service is started outside the cluster.

Issue

When lvmlockd is running within a cluster, it goes into a FAILED state when attempts are made to startoutside of the cluster. This additionally leads to fence of the the node experiencing a failure:

$ pcs status
------------------------------->8------------------------------
Node List:
  * Node rhel8-node2: UNCLEAN (online)
  * Online: [ rhel8-node1 ]

Full List of Resources:
  * xvmfence    (stonith:fence_xvm):     Started rhel8-node1
  * Clone Set: locking-clone [locking]:
    * Resource Group: locking:0:
      * dlm     (ocf::pacemaker:controld):   Started rhel8-node2
      * lvmlockd        (ocf::heartbeat:lvmlockd):   FAILED rhel8-node2 <-----
    * Started: [ rhel8-node1 ]

Failed Resource Actions:
  * lvmlockd_monitor_30000 on rhel8-node2 'not running' (7): call=42, status='complete', last-rc-change='Fri Jun  2 17:03:46 2023', queued=0ms, exec=0ms

Pending Fencing Actions:
  * reboot of rhel8-node2 pending: client=pacemaker-controld.2368, origin=rhel8-node1 <---

Ahead of the fence, the lvmlockd service may additionally print the below messages to logs:

rhel8-node2 $ cat /var/log/messages
------------------------------->8------------------------------
Jun  2 13:51:55 rhel8-node2 systemd[1]: Starting LVM lock daemon...
Jun  2 13:51:55 rhel8-node2 lvmlockd[5078]: Cannot lock lockfile [/run/lvmlockd.pid], error was [Resource temporarily unavailable]
Jun  2 13:51:55 rhel8-node2 lvmlockd[5078]: Failed to acquire lock on /run/lvmlockd.pid. Already running?