I replaced a faulted drive and the pool still sounded nervous

You replace the faulted Exos, you watch the resilver climb, you think “nice, now it’s fine.” And then you remember: ZFS only protected you from the drive that already died. The one next to it is the same age, same hours, same thermal history.

This is the part nobody puts on the sticker: a pool that just ate a drive is the most likely moment in its life to eat another one. RAIDZ2 exists precisely because that second drive is a when, not an if.

What the resilver actually means

The scrub-and-copy isn’t “healing” — it’s re-deriving every block of the failed drive from its mirrors on the survivors. During that window the remaining drives are working harder than they have in years. That’s when the aged one gives up, statistically.

The lesson, short

1. Keep a spare of the same model/era on the shelf. Not a “compatible” one — the same one.
2. Watch the resilver to completion, and keep watching for a few days after. An early symptomless re-fault is the classic RAIDZ1-flavored tragedy.
3. If you’re on a farm of matched drives and got warned to buy spares, buy them before the first one dies, not after.

Because the pool will keep running. But it’ll be doing it with the quiet confidence of someone who’s already seen their twin get eaten.