RAID 10 Mirror Mismatch: Determine Which Copy Is Current
A mirror mismatch means two members expected to contain the same data no longer agree. One may be stale, partly rebuilt, unreadable in places, or associated with a different write history. Synchronizing first lets the controller choose which history survives.
Do not synchronize, rebuild, or force either member online. Keep both sides of the mirror and record their current status.
What this condition means
Mirroring protects against a member failure, but it cannot decide automatically which copy is correct after interrupted writes, controller changes, partial rebuilds, or metadata disagreement.
A healthy label does not establish recency. File structures, RAID metadata, readable coverage, and the sequence of events must be compared before selecting a source.
What to record
- Affected mirror pair and each member position
- Serial number, size, status, and error history of both members
- Last normal operation and first mismatch alert
- Rebuild, synchronization, power, controller, or enclosure events
- Any member that was removed, replaced, cloned, or reinserted
What not to do
- Synchronize the mirror automatically
- Force one member online by guesswork
- Discard the member labeled stale or failed
- Start a rebuild onto either original
- Repair the filesystem before the current copy is identified
How ADR evaluates it
ADR compares the two members for identity, generation, readable coverage, metadata, filesystem structures, and priority data.
The most defensible member state is selected for each required stripe while original sources remain unchanged.
What happens next
Keeping both copies preserves options. ADR may use one member as the primary source and recover specific regions from the other without synchronizing the originals.