[OmniOS-discuss] A problem and puzzle with disappearing ZFS snapshots

[Dan McDonald]

> On Jan 6, 2017, at 6:09 PM, Chris Siebenmann <cks at cs.toronto.edu> wrote:
> 
> The @Fri-15 snapshot has since vanished from the pool. The full
> 'zpool history' output since the first /h/105 at Fri-16 snapshot is:
> 
> 	2017-01-06.16:10:01 zfs snapshot fs0-admin-02/h/105 at Fri-16
> 	2017-01-06.16:10:01 zfs snapshot fs0-admin-02/h/193 at Fri-16
> 	2017-01-06.16:10:06 zfs snapshot fs0-admin-02/h/194 at Fri-16

Look here...
> 	2017-01-06.16:43:03 zpool scrub fs0-admin-02
....

> 	2017-01-06.16:45:55 zfs snapshot fs0-admin-02/h/105 at Fri-16
> 	2017-01-06.17:10:11 zfs destroy fs0-admin-02/h/69 at Fri-17
> 	2017-01-06.17:10:22 zfs destroy fs0-admin-02/h/70 at Fri-17
> 	2017-01-06.17:10:32 zfs destroy fs0-admin-02/w/72 at Fri-17
> 	2017-01-06.17:10:43 zfs destroy fs0-admin-02/h/193 at Fri-17
> 	2017-01-06.17:10:54 zfs destroy fs0-admin-02/h/194 at Fri-17
> 	2017-01-06.17:11:04 zfs snapshot fs0-admin-02/h/69 at Fri-17
> 	2017-01-06.17:11:15 zfs snapshot fs0-admin-02/h/70 at Fri-17
> 	2017-01-06.17:11:26 zfs snapshot fs0-admin-02/w/72 at Fri-17
> 	2017-01-06.17:11:36 zfs snapshot fs0-admin-02/h/105 at Fri-17
> 	2017-01-06.17:11:47 zfs snapshot fs0-admin-02/h/193 at Fri-17
> 	2017-01-06.17:11:58 zfs snapshot fs0-admin-02/h/194 at Fri-17
> 
> And 'zfs list' says:
> 	NAME                        USED  AVAIL  REFER  MOUNTPOINT
> 	fs0-admin-02/h/105          344G   107G   343G  /h/105
> 	fs0-admin-02/h/105 at Fri-16   712M      -   343G  -
> 	fs0-admin-02/h/105 at Fri-17  9.42M      -   343G  -
> 
> This is completely mysterious to me. Any clues or things to look
> at are very welcome.

I wonder if the scrub has something to do with it?  Are you in the middle of the scrub when noticing missing snapshots?  Do they reappear after the scrub is done?  The scrub seems like it's relevant.  I take it you're doing a cron-driven scrub?  If so, with what frequency?

I'd normally push this along to the ZFS list, but given it's r151014 AND a non-updated r151014 at that, a bit more work here is probably a good idea.

Dan