2026-09-21
The pool is two mirrored 8 TB drives. For a long time I ran scrubs by hand whenever I remembered, which meant roughly never. Ubuntu already ships a timer for this, so the fix was mostly reading the docs.
systemctl list-timers | grep scrub zpool status -v tank
The first scrub took a little under nine hours and repaired nothing, which is the boring result you want. The second drive is a different batch from the first, so a bad lot should not take both out at once.
Next step: get the SMART reports emailed to me instead of checking them when something already feels wrong.