Document link: https://docs.openshift.com/container-platform/4.16/backup_and_restore/control_plane_backup_and_restore/replacing-unhealthy-etcd-member.html

Some steps suggest to rsh to the etcd pod like this:

$ oc rsh -n openshift-etcd etcd-ip-10-0-154-204.ec2.internal

Although in theory this should work, in practice there are a number of scenarios where it may not work even with 2 etcd peers up (e.g. if there is a quorum issue).

So it is safer to suggest to ssh to the node and use crictl to exec inside the etcdctl container. Something like this:

$ ssh core@ip-10-0-154-204.ec2.internal
[core@ip-10-0-154-204]$ sudo -i
[root@ip-10-0-154-204]$ crictl exec -t -i $(crictl ps --name etcdctl -q) /bin/bash 

So the above should either replace the oc rsh approach or be offered as an alternative.