20 November 2013: Wednesday


  • OSP notified me of a confused entry that had caused a second account to be created for a person.  We cleared up the problem in the SSN and I merged the two entries together.
  • A new former student showed up in the nightly update.  Searched but not found, so added.
  • An account added to a VPN group.


  • When replacing the failed drive in the new array yesterday, it was discovered that zfs did not like the new drive, even though it is identical to the failed drive.  Turns out that ZFS doesn’t like ANY of the drives in that pool.  Researching brings me to the conclusion that the M5120 firmware update in October is the root of this problem.  It appears that following that update, the M5120 is (correctly) reporting a 4K physical block size (instead of the 512 byte blocks it claimed previously) for these SEAGATE ST4000NM0023 disks.  Since ZFS pays attention to that, this means that it wants an ashift value of 12 now instead of the 9 it used when the pool was created.  Two ways out of this jam.  The first (and probably best) is to empty the pool and rebuild it.  The second is to find the parameter to make the M5120 go back to lying about the size of the disks.
    • I was told that there was a change about 512/4096 byte blocks in the firmware I put on.  I can’t find any mention of that in the change notes for that firmware (and those notes go all the way back to initial release).  I must find more details.


Leave a Reply

You must be logged in to post a comment.