You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
During a reboot, ZFS must unmount all file systems. This is frequently very slow. The problem seems to be that zfs_umount calls zfs_teardown which calls txg_wait_synced. And syncing a transaction is very slow if a scrub or resilver is ongoing. In such cases, I've frequently seen the reboot take tens of minutes, and sometimes hours (assuming I don't hit the power button first). What can we do to speed things up? Manually pausing scrubs before rebooting seems to help, but I can't manually pause a resilver. And I shouldn't have to do it manually. Can we add some kind of pre-shutdown hook that suspends all scrubs and resilvers? Or somehow batch multiple unmounts together and only wait on a single TXG sync?
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
During a reboot, ZFS must unmount all file systems. This is frequently very slow. The problem seems to be that
zfs_umount
callszfs_teardown
which callstxg_wait_synced
. And syncing a transaction is very slow if a scrub or resilver is ongoing. In such cases, I've frequently seen the reboot take tens of minutes, and sometimes hours (assuming I don't hit the power button first). What can we do to speed things up? Manually pausing scrubs before rebooting seems to help, but I can't manually pause a resilver. And I shouldn't have to do it manually. Can we add some kind of pre-shutdown hook that suspends all scrubs and resilvers? Or somehow batch multiple unmounts together and only wait on a single TXG sync?FreeBSD 14, amd64, if it matters.
Beta Was this translation helpful? Give feedback.
All reactions