The cluster has 3 pods 3.7.5 RabbitMQ replica's in a StatefulSet. Using rabbit_peer_discovery_k8s, cluster_formation.randomized_startup_delay_range and cluster_partition_handling = autoheal

Sometimes when a pod restarts it can't connect to the existing cluster due to an {badrpc,nodedown} error. And then starts as a blank standalone node.

How can I prevent the standalone node?