Saved by ZFS – a disk is about to die

I have setup an NFS share under Solaris 10. It utilizes ZFS, which in turn uses  an IBM DS400 for backend storage. On top of that I have nagios running to monitor it. I got an alarm about the zfs pool being in a degraded state. I logged into the system and found this in the dmesg:

May 17 03:20:20 files DESC: The number of checksum errors associated with a ZFS device
May 17 03:20:20 files exceeded acceptable levels.  Refer to for more information.

To see more information I ran

-bash-3.00# zpool status
 pool: rz2pool
 state: DEGRADED
status: One or more devices has experienced an unrecoverable error.  An
 attempt was made to correct the error.  Applications are unaffected.
action: Determine if the device needs to be replaced, and clear the errors
 using 'zpool clear' or replace the device with 'zpool replace'.
 scrub: scrub completed after 5h21m with 0 errors on Wed May 19 08:41:49 2010

 NAME                        STATE     READ WRITE CKSUM
 rz2pool                     DEGRADED     0     0     0
   raidz2                    ONLINE       0     0     0
     c3t21000000D12643DEd0   ONLINE       0     0     0
     c3t21000000D12643DEd1   ONLINE       0     0     0
     c3t21000000D12643DEd2   ONLINE       0     0     0
     c3t21000000D12643DEd3   ONLINE       0     0     0
     c3t21000000D12643DEd4   ONLINE       0     0     0
     c3t21000000D12643DEd5   ONLINE       0     0     0
     c3t21000000D12643DEd6   ONLINE       0     0     0
   raidz2                    DEGRADED     0     0     0
     c3t21000000D12643DEd7   ONLINE       0     0     0
     c3t21000000D12643DEd8   ONLINE       0     0     0
     c3t21000000D12643DEd9   ONLINE       0     0     0
     c3t21000000D12643DEd10  ONLINE       0     0     0
     c3t21000000D12643DEd11  ONLINE       0     0     0
     c3t21000000D12643DEd12  DEGRADED     0     0   234  too many errors
     c3t21000000D12643DEd13  ONLINE       0     0     0

errors: No known data errors

And this is where ZFS is awsome. It may not be the fastest volume manager on the planet, or the smartest. But I trust the integrity of it (having read whitepapers on it).

What is really cool here:

  1. It has detected that the underlying LUN is misbehaving.
  2. It has marked the LUN as degraded
  3. It has saved my data from silent corruption.

There are not many volume managers out there, which does that. I have not lost data, the dataintegrity is still in place and I know what disk is about to fail. Kudos and thanks to the ZFS dev team!

Leave a Reply

You must be logged in to post a comment.