Choose the operational conclusion supported by actual control-plane and application state. A successful API call, scheduled Pod, or elected leader covers only one boundary.
Review these lessons
- Admission webhook outage: choose a deliberate failure path
- etcd quorum: preserve a voting majority during maintenance
- Pod address capacity: diagnose network allocation before adding nodes
- CSI attach limits: verify storage placement as well as CPU placement
- CRD storage migration: retire an API version only after stored objects move
- Controller watches: recover from expired resource versions without a relist storm
- Pod priority: reserve recovery capacity without evicting the wrong work
- Leader leases: fence side effects after ownership changes
Other checks
Common Mistakes
- Do not treat Pod scheduling as a mounted volume.
- Do not treat a newly served CRD version as completed storage migration.