Network & Infrastructure / Storage Networks & Distributed Data
Erasure-coded placement and correlated shard loss
An erasure-coded pool has the expected shard count but still cannot tolerate the intended rack failure. Several required shards may share the same physical risk domain.
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
Review coding profile, CRUSH rule, actual shard placement and the number of distinct eligible domains. Model correlated failure rather than considering only independent disk failures.
Explain the result
Coding tolerance depends on how many shards are lost and whether enough remain available. A nominal parity count does not guarantee resilience to a rack that holds multiple relevant shards.
Plan the fix
Align coding and placement with the actual fault-domain objective and supported platform design. Plan capacity and data movement before changing profiles or rules on an active pool.
Check the result and rollback
Verify placement and recovery behavior with representative objects in a safe test environment. Revert if the new constraints are unsatisfiable or degraded-mode performance cannot meet the application's needs.