Restart a fleet

Request a restart with a new token:

kubectl --context YOUR_CONTEXT -n fleets patch celldfleet my-fleet   --type merge -p '{"spec":{"maintenance":{"restartToken":"restart-2026-09-20-1"}}}'
kubectl --context YOUR_CONTEXT -n fleets get celldfleet my-fleet -o yaml

The operator writes the token to the Pod template annotation celld.eric.dev/restart-token, and members are replaced one at a time. Each member drains on SIGTERM within lifecycle.shutdownSeconds. No downtime permission is needed; allowCoordinatedDowntime is ignored. The workload controller performs the rolling update: a Bucket fleet’s Deployment or StatefulSet, or a PersistentFleet’s StatefulSet.

PersistentFleet

The StatefulSet uses RollingUpdate. Kubernetes restarts one member at a time, highest ordinal first, and waits for each to be Ready before the next. Readiness is celld’s health endpoint, which returns 200 only after the member has recovered its previous session. The replacement Pod reattaches the same disk. The operator deletes no Pods to restart them. While the rollout runs, the fleet reports Provisioning with Rolling out one member at a time; waiting for updated, ready replicas. See the PersistentFleet lifecycle.

The current token is not replayed; use a new token for another restart. maintenance.paused: true stops the operator from writing workload changes (MaintenancePaused); a rollout the workload controller has already started continues. See lifecycle troubleshooting if a rollout keeps waiting.