Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 01 May 2011 19:57:38 +0000
From:      "Michael D. Norwick" <mnorwick@centurytel.net>
To:        freebsd-questions@freebsd.org
Subject:   freebsd zfs question
Message-ID:  <4DBDBB32.7010906@centurytel.net>

Next in thread | Raw E-Mail | Index | Archive | Help
Good Day;

A week or so ago I experienced an error trying to compile openoffice.org 
from ports.  The build failed from an error I was since able to resolve.
This machine is at #>uname -r rainey 8.2-RELEASE-p1 FreeBSD 
8.2-RELEASE-p1 #2: Wed Apr 27 04:37:38 UTC 2011     
michael@rainey:/usr/obj/usr/src/sys/KERNEL_042511  amd64
During that episode the workstation locked up and the only way to 
recover was to do a hard reset (power off).  I then noticed during the 
next weekly scrub that I had increasing errors listed for the root pool 
during that particular scrub;

# zpool status
   pool: tank
  state: ONLINE
status: One or more devices has experienced an error resulting in data
     corruption.  Applications may be affected.
action: Restore the file in question if possible.  Otherwise restore the
     entire pool from backup.
    see: http://www.sun.com/msg/ZFS-8000-8A
  scrub: none requested
config:

     NAME        STATE     READ WRITE CKSUM
     tank        ONLINE       0     0     0
       ad2p3     ONLINE       0     0     0
       ad3p1     ONLINE       0     0     0

errors: 604 data errors, use '-v' for a list

   pool: tank1
  state: ONLINE
  scrub: none requested
config:

     NAME        STATE     READ WRITE CKSUM
     tank1       ONLINE       0     0     0
       ad12p2    ONLINE       0     0     0

errors: No known data errors

Tank consists of 2 300G PATA drives in a mirror.  Tank1 is a 500G SATA 
drive I just added recently to use for data archiving.  The stuff I wish 
to protect is backed up to a network file server manually via NFS.  I 
have scrubbed the pool showing errors several times now with no 
increases or decreases in error counts.  I have issued #>zpool clear 
tank a number of times with no change in the error count.  The document 
listed (www.sun.com/msg/ZFS-8000-8A) was of no apparent help for my 
condition.  I have drives I can export to and import from but I am 
unclear as to whether I will be just moving the bad blocks around.

Sample of the output of #>zpool status -v tank;

<
         tank/root:<0x1097e7>
         tank/root:<0x1096e8>
         tank/root:<0x1097e8>
         tank/root:<0x1096e9>
         tank/root:<0x1097e9>
         tank/root:<0x1095ea>
         tank/root:<0x1097ea>
         tank/root:<0x1096eb>
         tank/root:<0x1097eb>
         tank/root:<0x1096ec>
         tank/root:<0x1097ec>
         tank/root:<0x1095ed>
         tank/root:<0x1096ed>
         tank/root:<0x1097ed>
         tank/root:<0x1094ee>
         tank/root:<0x1096ee>
         tank/root:<0x1095ef>
 >

Not sure what to do with these.  Why doesn't #>zpool clear tank delete 
these?  The directory /usr/ports/editors/openoffice.org-3/work was not 
able to be deleted after the failed build, so I moved it to /oldwork to 
get the port to build.  /oldwork still cannot be deleted.

rainey# rm -Rf ./oldwork
rm: ./oldwork/OOO330_m20/dictionaries: Directory not empty
rm: ./oldwork/OOO330_m20/lucene/unxfbsdx.pro/bin: Directory not empty
rm: ./oldwork/OOO330_m20/lucene/unxfbsdx.pro/misc/build: Directory not empty
rm: ./oldwork/OOO330_m20/lucene/unxfbsdx.pro/misc: Directory not empty
rm: ./oldwork/OOO330_m20/lucene/unxfbsdx.pro: Directory not empty
rm: ./oldwork/OOO330_m20/lucene: Directory not empty
rm: ./oldwork/OOO330_m20/jfreereport: Directory not empty
rm: ./oldwork/OOO330_m20/libxslt: Directory not empty
rm: ./oldwork/OOO330_m20/sal: Directory not empty
rm: ./oldwork/OOO330_m20: Directory not empty
rm: ./oldwork: Directory not empty

Attempts to delete the above directories fail.
I've read articles about 'bit rot' and such in ZFS metadata but 
memtest86 completes without error on this machine's 3G of ram.  I see no 
applicable information in dmesg or /var/log/messages.  The drives have 
been running 24/7 since the initial incident with no increase in the 
error count.

Thank You,
Michael



Want to link to this message? Use this URL: <http://docs.FreeBSD.org/cgi/mid.cgi?4DBDBB32.7010906>