Post #2382400
2026-05-06 20:54 UTC
@zwol@masto.hackers.town @djfiander@code4lib.social @mwl@io.mwl.io @b0rk@social.jvns.ca Add one to the count of people who’ve lost data due to a ZFS error, but I’m not sure that it was the lack of fsck to blame. As I understand it, ZFS lacks the metadata redundancy it would need to figure out how to fsck the problems that can happen.
OTOH, it as been a very long time since ZFS lost data on me. 18 years at least, since it was on an early version of Solaris 10. Sun originally claimed that the only way for ZFS to lose data was simultaneous disk failures.
Replies (2)
-
@mwl@io.mwl.io 2026-05-06 21:06
@grumpybozo@toad.social @zwol@masto.hackers.town @djfiander@code4lib.social @b0rk@social.jvns.ca generally, scrub does everything that fsck does and more. You can offline your pool if you want, but it works fine live. Most often, folks lose data because they don't bother to scrub.
-
@zwol@masto.hackers.town 2026-05-06 21:26
@grumpybozo@toad.social @djfiander@code4lib.social @mwl@io.mwl.io @b0rk@social.jvns.ca that's what I'm talking about when I say unforgivable hubris