Skip to main content
These runbooks are for incident response and production operations of RedisCluster resources managed by this operator.

​
Runbook Index

ScenarioRunbook
Manual failover when operator is unavailableManual Failover
Reconciler loop appears stuckStuck Reconciler
All data pods lost, PVCs intactTotal Cluster Loss
Single corrupted replica PVCPVC Corruption
Suspected split-brain (two primaries)Split-Brain Recovery
Rotate authSecret without pod restartsSecret Rotation
Upgrade operator safelyOperator Upgrade

​
Shared Conventions

Set shell variables before executing commands:
  • NS=<rediscluster namespace>
  • CLUSTER=<rediscluster name>
  • Prefer touching annotations to force immediate reconciliation; the controller also reconciles periodically.
  • Validate the cluster has returned to status.phase=Healthy before closing the incident.

Build docs developers (and LLMs) love

Get started for freeTalk to us