Business-critical storage failure?1-800-228-8800
ADRADR Data RecoveryRAID · Server · NAS · Virtualization Start Emergency Triage

Why Intervention Can Make a RAID 6 Failure Worse

RAID 6 can tolerate two unavailable members only while the remaining data, both parity relationships, and the selected member history are sufficiently consistent. Once a third member becomes unreadable—or an incorrect rebuild changes the set—the original dual-parity protection may no longer describe one recoverable state.

Preserve the current state.

Do not use force-online, initialization, member substitution, or another rebuild as a diagnostic test. Preserve every original member and record the order of events.

What this condition means

The number of drives labelled failed does not tell the whole story. Members can drop at different times, contain different metadata generations, or remain mostly readable despite a controller failure label.

A rebuild writes a new member state from the sources selected by the controller. If one source is stale, unstable, or incorrectly included, those writes can reduce the number of defensible reconstruction choices.

What to record

What not to do

How ADR evaluates it

ADR compares member readability, metadata generations, failure history, and parity consistency to determine which sources can represent the same array state.

Unstable members are protected before reconstruction. Candidate layouts are tested away from the hardware controller, and the resulting filesystem and priority data are checked before recovery proceeds.

What happens next

Intervention does not automatically make a RAID 6 case unrecoverable. The practical question is what each action wrote, which members remain readable, and whether a consistent array state can still be reconstructed.