4

While preparing to move a zfs pool to another server chassis I did an 'zfs export' while in multi-user mode. Hindsight I should have have done this from a rescue disk.

After the export, which failed, I rebooted and have never been able to import the pool since. All of the disk and the pool are in an online state. It's more like the problem is a software issue with zfs - possible a meta data issue. Has anyone else seen this before or does anyone have any suggestions for recovering data from a possibly corrupted pool? Outside of the I/O error I've not been able to get any good feedback as to what's causing the I/O error. I've tried using truss with the import to see what's going on.

truss output: http://pastebin.com/DSDpuR1i

gpart list output: http://pastebin.com/Wxgr2PMx

I set this up using FreeBSD9 and I believe this is zfs v28.

As a side note I know I should have had backups. The reason I did not was more about money than anything else. The plan was to move this pool to a new norco chassis and add in an equal number of disks for another pool to mirror to.

root@nas01:~ # zpool import
   pool: rpool
     id: 15664112288097167104
  state: ONLINE
 status: The pool was last accessed by another system.
 action: The pool can be imported using its name or numeric identifier and
        the '-f' flag.
   see: http://illumos.org/msg/ZFS-8000-EY
 config:

        rpool                  ONLINE
          raidz1-0             ONLINE
            diskid/DISK-%20p3  ONLINE
            da1p3              ONLINE
            da2p3              ONLINE
            da3p3              ONLINE
            da5p3              ONLINE
            da4p3              ONLINE
root@nas01:~ # zpool import -f -o altroot=/mnt rpool
cannot import 'rpool': I/O error
        Destroy and re-create the pool from
        a backup source.
root@nas01:~ #

edit - when I try with -nfF or -fF

root@nas01:~ # zpool import -nfF -o altroot=/mnt rpool
root@nas01:~ # echo $?
1
root@nas01:~ # zpool import -fF rpool
cannot import 'rpool': I/O error
        Destroy and re-create the pool from
        a backup source.
root@nas01:~ #
Improve this question
3
  • Is there a way to have truss output what is actually being opened? openat(0x6,0x802c390d0,0x0,0x0,0x7fffffffb830,0x802c00c78) = 7 (0x7) isn't too helpful.
    – Mark Wagner
    Sep 24, 2014 at 0:40
  • Not that I know off. I'm more of a linux guy so the truss output has been more of a mystery to me than if this had been strace output. Basically I ran truss while trying to import the pool to try and get a better idea what the I/O error actually means.
    – Glen
    Sep 24, 2014 at 8:49
  • Sounds lilke the issue I had in OpenZFS 0.75 serverfault.com/a/1004188/52734
    – Louis Waweru
    Feb 22, 2020 at 8:33

1 Answer 1

Reset to default
2

try with:

 zpool import -nfF rpool

if you don't see any critical error, try to use the previous command without n option

 zpool import -fF rpool
Improve this answer
4
  • Thanks. Oddly the -nfF doesn't show anything but i still get the I/O error despite disk health.
    – Glen
    Sep 23, 2014 at 22:44
  • what happened when you used "zpool import -fF rpool" ?
    – c4f4t0r
    Sep 24, 2014 at 8:07
  • I get the "cannot import 'rpool': I/O error.
    – Glen
    Sep 24, 2014 at 8:45
  • try to use "zpool clear rpool"
    – c4f4t0r
    Sep 24, 2014 at 8:51

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .