ZFS: drive replacement performance

[Mahlon E. Smith]

On Tue, Jul 07, 2009, Freddie Cash wrote:
> 
> I think (never tried) you can use "zpool scrub -s store" to stop the
> resilver.  If not, you should be able to re-do the replace command.

Hmm.  I think I may be stuck.

% zpool scrub -s store

% zpool status | grep scrub
   scrub: resilver in progress for 0h0m, 0.00% done, 745h41m to go

% zpool replace store 2025342973333799752 da8
  invalid vdev specification
  use '-f' to override the following errors:
  /dev/da8 is part of active pool 'store'
  
% zpool replace -f store 2025342973333799752 da8
  invalid vdev specification
  the following errors must be manually repaired:
  /dev/da8 is part of active pool 'store'

% zpool detach store da8
  cannot detach da8: no valid replicas

% zpool detach store 2025342973333799752
  cannot detach 2025342973333799752: no valid replicas



I also tried another export/import cycle, in the random hope that would
stop the active replace -- no dice.  *However*, on the import, now I see
this flooding my console (wasn't there previously, strangely):

Jul  7 16:50:15 disobedience root: ZFS: vdev I/O failure, zpool=store path=/dev/da2 offset=262144 size=8192 error=6
Jul  7 16:50:15 disobedience root: ZFS: vdev I/O failure, zpool=store path=/dev/da2 offset=499988824064 size=8192 error=6

I now have to wonder if that's really the active da2 it is complaining
about (the one claiming to be online with 0 errors) or the one I'm
trying to replace with da8.  The current da2 doesn't seem to be
having any additional problems, like the checksum mismatches or other
associated console errors I've come to expect, but of course the old one
is no longer attached to the machine.  In any event, I'd wager that
isn't something I normally want to see, and I may have something else
going on here.  (Bad controller, etc?)

Serves me right for naming a machine 'disobedience', I guess.  Next one
is getting named 'subservience.'

Going to halt and pull da8 under the assumption that will at least stop
the resilver, and try the detach again.  I'll holler back if I get stuff
going again, but this is looking more like a hardware problem.

Thanks again for the insight!

-Mahlon

--
Mahlon E. Smith  
http://www.martini.nu/contact.html
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 155 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20090708/c5c33ed4/attachment.pgp