What I do when a ZFS scrub finds errors

2026-07-12

Monthly scrub on the first Sunday. For a year the mail said 0 errors and I stopped reading it. Then it didn't.

  pool: tank
 state: ONLINE
status: One or more devices has experienced an unrecoverable error.
errors: No known data errors

The important line is the last one. No known data errors means the pool healed everything from redundancy. The device counters are the thing to watch, not to panic about.

My checklist

• Note which disk, and the counts (READ / WRITE / CKSUM). One checksum error after a year is noise. Errors that grow between scrubs are a disk on its way out.
• smartctl -a on that disk. Reallocated sectors, pending sectors, and the SATA CRC counter. CRC errors are almost always the cable, not the disk. I have replaced two cables and zero disks because of this line.
• zpool clear and scrub again right away. If the second scrub is clean, note the date and move on.
• If it is not clean, order a disk. Do not wait for the third scrub.

The mirror is two disks from different batches, which is the only reason I sleep. The replacement is a resilver of about five hours on this hardware.