Recovering an OpenShift Cluster After a Shutdown

From etcd to Console Access

Unexpected cluster shutdowns happen — maybe from power loss, infrastructure maintenance, or human error. When you power the cluster back on, it may not return to a healthy state automatically. The etcd datastore, control plane, pending CSRs, and operator-managed workloads (like console and ingress) can all require manual intervention.

This guide walks through a structured recovery approach to bring your cluster — and the OpenShift Web Console — back online.

1. Start with the Control Plane

The control plane must be healthy before anything else will work.

Verify Master Nodes

oc get nodes

• All masters should be Ready.
• If not, SSH into a failing master and check:

sudo systemctl status kubelet sudo journalctl -u kubelet -e | tail -n 50

Check etcd Health

oc get co etcd
oc -n openshift-etcd get pods -o wide
oc -n openshift-etcd logs -l app=etcd --tail=50

• Expect three etcd pods (one on each master).
• If only one pod is running or quorum is broken, the cluster will not function.
• If quorum is lost, the only supported solution is to restore etcd from backup using the cluster-restore.sh script.