With leader-follower HA configuration, we witnessed that the follower pod waits for 5 minutes before it locks the file during the worker node with leader pod "poweroff" test.

We have ODF for storage, which is 3 node CEPH cluster running on INFRA nodes . We don't touch storage nodes during the tests of Broker HA. We only "poweroff" the worker node where Broker Leader is running. PVC used as shared storage is this:

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: shared-volume
spec:
  storageClassName: ocs-storagecluster-cephfs
  accessModes:
    - ReadWriteMany
  resources:
    requests:
      storage: 10Gi

The lock release issue only happens with "powering off" worker and the leader pod test scenario. In another scenario, when testing with command "oc delete <leader_pod>" to delete the leader pod, the follower pod acquired the lock immediately so HA failover worked perfectly fine.