Behaviour when entire vdevs of disks drop all at once and then come back online

I’ve been looking for clarity for a bit and could find anything definite so figured I would ask

Say I have a pool of 8 vdevs (disks on two separate controllers) then a controller dies and 3 of the vdevs go entirely offline.

obviously the pool stops/can no longer function but what happens when the controller is replaced and the disks all come back online at once?

does the pool recover? or is it just toast

I’ve had this happen a few times and the pool recovered every time. Some errors which after a scrub went away and the pool continued to function perfectly for many years.

1 Like

In my experience the pool’s I/O gets suspended, and after shutting down, correcting the issue and restarting, the pool came back, and a scrub was good.

I presume the SLOG did its job too as sync=always

1 Like

That’s fantastic to hear, thanks both for sharing your experiences