Hi all,

It would appear that I have a zpool corruption issue to deal with... pool is exported, but upon trying to import it, server panics.  Are there any tools available on a zpool that is in an exported state?  I've got a separate test bed in which I'm trying to recreate, but I keep getting messages to the effect of need to import the pool first. 

Suggestions?

thanks
Jay


Hardware:

Two Sun-Fire T2000’s running Sol10 8/07 s10s_u4wos_12b (SPARC) under control of Veritas Cluster.

Sun StorageTek 6140 storage array.

 

Highlevel Configuration:

6140 is setup in RAID5.  Several volumes were created on it, each mapping to a different zpool.

 

ZFS Configuration:

Single storage volume for each zpool, no mirrors, no raidz groups.  

 

Events leading up to failure:

Both nodes of the cluster experienced reboots as a result of recovery efforts due to a power outage at the site.  After the cluster was online and functioning, in order to balance the load between the two T2000’s, the “home” zpool was exported from one T2000 and then imported to the other T2000.  Upon trying to import on the second T2000, there was a bad checksum error and then the system did a dump and rebooted.  Any subsequent attempts to import the “home” zpool have ended in the same.

 

A zfs support person from Sun had us try adding a few entries to /etc/system to see if we could then import, but we still had the same results.

 

Just to be 100% clear, there was no hardware failure in any of the storage subsystems.  The 6140 is in perfect health.  This is in my belief a ZFS issue, possibly caused by a failure in the Veritas Cluster in which two nodes tried to write to the same zpool simultaneously.  If not that, perhaps some random driver error or bug… I don’t know.  All theories aside, here is the output of the error we see every time the `zpool import home` command fails:

 

panic[cpu8]/thread=2a100b69cc0: ZFS: bad checksum (read on <unknown> off 0: zio 60015f26a40 [L0 packed nvlist] 4000L/400P DVA[0]=<0:4000b7000:400> DVA[1]=<0:4600063800:400> fletcher4 lzjb BE contiguous birth=3865837 fill=1 cksum=6e858ee992:2cad0296498c:cae1d40ad3d38:2e1daaef734ef08): error 50

 

000002a100b69880 zfs:zio_done+29c (0, 10, 60015f26a98, 60012e062c0, 60015f26a40, 0)

  %l0-3: 0000060015d04040 00000600160ffac0 00000000707de1c0 0000000000000032

  %l4-7: 0000000000000001 0000000000000002 0000000000000032 0000060015c05118

000002a100b69940 zfs:zio_vdev_io_assess+178 (60015f26a40, 8000, 10, 0, 0, 10)

  %l0-3: 0000000000010000 0000060015c05120 0000000000000000 0000000000000032

  %l4-7: 0000060015c02888 0000000000000002 0000000000000000 0000060015c05118

000002a100b69a00 genunix:taskq_thread+1a4 (60015c05150, 60015c050f8, 50001, 5751b2acf88, 2a100b69aca, 2a100b69ac8)

  %l0-3: 0000000000010000 0000060015c05120 0000060015c05128 0000060015c0512a

  %l4-7: 0000060015c02888 0000000000000002 0000000000000000 0000060015c05118

 

syncing file systems... 6 done

dumping to /dev/dsk/c0t0d0s1, offset 65536, content: kernel

100% done: 100532 pages dumped, compression ratio 6.53, dump succeeded

rebooting...

 

SC Alert: Host System has Reset

 

SC Alert: Host system has shut down.

 

What we’re looking for at this point is any method to recover data from this zpool.  Even a partial one would be better than nothing.

 


 



-- 
Thanks,
Jay
 
      ______
     /_____/\  Jay Hamilton
    /____ \\ \   Systems Support Engineer
   /_____\ \\ /   Sun Microsystems, Inc.
  /_____/ \/ / /   Sun Service
 /_____/ /   \//\   500 El Dorado Blvd ubrm05-420
 \_____\//\   / /    Broomfield, CO. 80021
  \_____/ / /\ /    Accessline: (720) 895-3026 (x43026)
   \_____/ \\ \    Email: [EMAIL PROTECTED]
    \_____\ \\    Pager: [EMAIL PROTECTED]
     \_____\/   
________________________________________________________________________

Submit, View and Update Service Tickets at 
http://www.sun.com/service/online

This email may contain confidential and privileged material for the sole
use of the intended recipient. Any review or distribution by others is 
strictly prohibited. If you are not the intended recipient please 
contact the sender and delete all copies.
_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss

Reply via email to