We are using cluster/leave[1] in post script of the dbcluster pods which would not work with pods non-graceful stop or worker node crash. We would need to handle this or atleast add a way to recover from such situations. It shouldn't be an issue with single pod db cluster(replicas: 1) but could be when there are more replicase(like 3+) and more than half(like 2+) of them crash.

 [1] https://github.com/openstack-k8s-operators/ovn-operator/blob/main/templates/ovndbcluster/bin/cleanup.sh#L24