The disk storage is necessary to maintain state (under the hood, raft protocol), which is similar to what clusters use to maintain their state and elect leaders, such as zookeeper, etcd, and consul. Thus if the data is lost, which means the state of the cluster is lost, and you in a way start from scratch.
One thing to look at is if the same claim (PVC) is still used to the correct volume (PV), or are new claims made. The reclaim policy for the storage should not be set to delete, but rather retain, so that it can scope up existing storage after an outage event. Otherwise, if the pods are cycled out, they may not be able to grab onto the same volume.