hesela.devRAID consistency check
A RAID consistency check tests whether redundant representations agree, such as mirrored blocks or parity and their associated data.
A controller consistency check can verify and correct mirror or parity data on fault-tolerant virtual disks. This is a redundancy-level operation, distinct from patrol read and from filesystem or application validation. [1] [4]
The mdadm 4.2 manual distinguishes readable mismatches from read errors. MD check records the former without reconciling them, but normal read-error recovery can still write reconstructed data. MD repair can rewrite parity or mirror copies; the command name check is not a write-free guarantee. [2]
Agreement alone does not identify the intended payload. Two matching copies can contain the same wrong application output. A single-error model shows that unvalidated maintenance reads can propagate bad data into parity. This does not establish that every controller behaves that way. [3] This is an integrity-boundary inference, not a measured fault rate. Retain coverage, completion, mismatches and repair actions separately. [4]
Relations
- related
- Patrol read, Data scrubbing, MD mismatch count, Parity inconsistency, Parity pollution, Checksum, RAID rebuild