Can following scenario cause data loss on Repanda cluster using local disks even with replication?

[pdeva]

Assume that I have a 3 node cluster RP running on K8s. RP is using the local disk of node for storage.
I have a topic T with 3 ISR replicas on nodes A, B, C.

An admin decides to update the nodegroup running the RP cluster and updates nodes 1 at a time.

1. Node A is deleted and recreated. The local data on Node A is gone when RP starts on it again. Node A is trying to replicate topic T from B & C, but its taking a while due to size of data.
   Cluster State: Node A - out of sync, Node B, C: In Sync

2. Node B is now deleted and recreated by k8s due to the update operation above. Node B comes back but again its local data is gone. It only has Node C with the fully copy of topic T. Cluster State: Node A, B - out of sync, Node C: In Sync

3. Node C is now deleted and recreated. Node C comes back without any local data. Cluster State: Node A, B, C - out of sync. Node C now has nowhere left to get its copy of topic T's data from.

[jcsp]

Redpanda's health API (as used in rpk cluster health) will report an unhealthy status when any partitions are under replicated. In the example above, partitions will be under-replicated after step 1, until all data has re-replicated to the replacement redpanda node.

The health API and associated metrics are used to ensure that redpanda node replacements are done safely:

• When managing redpanda nodes automatically with Kubernetes, Redpanda's Kubernetes Operator and Helm chart configure a Pod Disruption Budget that requires the cluster to report OK health. This means that procedures like rolling upgrades will wait for the cluster to be healthy before stopping the next kubernetes node.
• If an administrator is manually eliminating redpanda nodes and their storage (whether it's kubernetes nodes, or removing bare metal servers), then the onus is on the user to ensure there are no under-replicated partitions before they proceed to the next node, as in the case of a rolling upgrade (https://docs.redpanda.com/docs/manage/cluster-maintenance/rolling-upgrade/#check-metrics)