Network & Infrastructure / Storage Networks & Distributed Data
Storage maintenance admission based on degraded-state capacity
A planned host reboot begins while the cluster is already recovering from another failure. The combined loss can exceed replica, quorum or performance tolerance.
Environment and inputs: Ceph and storage transport dependencies; match health identifiers and recovery behavior to the installed release. Diagnostic/design example; not tested on production equipment.
Collect the evidence
Check current health, ongoing recovery, placement, failure-domain overlap and surviving capacity before admitting the maintenance action. Review whether the requested host contains several critical roles.
Explain the result
Nominal redundancy assumes specific available members and placements. Existing degradation consumes part of that budget, so a routine action can become unsafe in the current state.
Plan the fix
Define maintenance prerequisites and an explicit defer condition based on the actual cluster architecture. Use the release's supported maintenance workflow and avoid broad flags whose effects are not understood.
Check the result and rollback
Proceed only when the expected remaining protection and service objectives are demonstrated. Abort and restore the host if recovery stalls, placement becomes unavailable or the agreed degraded-mode limits are exceeded.