Etcd pod in CrashLoopBackOff after rebuilding master node in RHOCP 4

Issue

• After rebuilding or replacing a bare-metal control-plane node, the newly added node shows a Ready status in oc get nodes, but the corresponding etcd static pod remains in CrashLoopBackOff or fails to join the cluster.

• The BareMetalHost (BMH) object exists but has no CONSUMER assigned, and there is no corresponding Machine object for the new node:

  NAME                 STATUS       STATE       CONSUMER   ONLINE   ERROR   AGE
  <rebuilt_node_name>  discovered   unmanaged               true            19h
  

• The openshift-etcd-operator pod logs continuously output the following warning:

  Ignoring node (<rebuilt_node_name>) for scale-up: no Machine found referencing this node's internal IP (<node_ip>)