When preemptively changing primary, the charm does not update Pod labels on the same handler, which will disrupt client connections for more time that's actually necessary, since the k8s services used to map the primary and replicas will be outdated until some other hook updates it.
Todo:
on preemptive primary switchover, call k8s update_pod_labels method
[optional] test k8s service endpoint for primary before exiting the hook.
When preemptively changing primary, the charm does not update Pod labels on the same handler, which will disrupt client connections for more time that's actually necessary, since the k8s services used to map the primary and replicas will be outdated until some other hook updates it.
Todo:
update_pod_labels
method