zfs discuss - Apr 2007 - zpool import/export checksum error + panic + work-around

Hi List,

A few days we exported a zpool which caused our system to panic. Ever since we
are able to reimport the pool. We are quite sure all data is intact sinse the
pool is located on raid 10 and the other pools still work. We copied the entire
disk (iscsi) to a file and placed it back on a ide disk to make sure the problem
wasn''t system/iscsi related.

We''ve tried many solaris releases as well as build 61 with the
/etc/system entry. Nothing worked.

Then we tried zfs on linux with fuse. This also resulted in the same error.

We avoided the problem by modifying the checksum function from zio_checksum.c by
putting return 0; as the first line. So the bug is there somewhere. The zpool is
now imported and we can access our data.

This is our problem (we have crashdumps.... quite a lot ....)

log crash entry:

Apr  6 15:50:50 rbgz02 genunix: [ID 809409 kern.notice] ZFS: bad checksum (read
on <unknown> off 0: zio ffffffff9bc7fa00 [L0 packed nvlis
t] 4000L/400P DVA[0]=<0:1480013c00:400> DVA[1]=<0:b00053000:400>
fletcher4 lzjb LE contiguous birth=1791755 fill=1 cksum=6730853727:29b04
d00d6a8:b64ae1f6b3080:27fe414eb9930b6): error 50
Apr  6 15:50:51 rbgz02 unix: [ID 100000 kern.notice] 
Apr  6 15:50:51 rbgz02 genunix: [ID 655072 kern.notice] fffffe80010baac0
zfs:zfsctl_ops_root+2fe39f02 ()
Apr  6 15:50:51 rbgz02 genunix: [ID 655072 kern.notice] fffffe80010baad0
zfs:zio_next_stage+72 ()
Apr  6 15:50:51 rbgz02 genunix: [ID 655072 kern.notice] fffffe80010bab00
zfs:zio_wait_for_children+49 ()
Apr  6 15:50:51 rbgz02 genunix: [ID 655072 kern.notice] fffffe80010bab10
zfs:zio_wait_children_done+15 ()
Apr  6 15:50:51 rbgz02 genunix: [ID 655072 kern.notice] fffffe80010bab20
zfs:zio_next_stage+72 ()
Apr  6 15:50:51 rbgz02 genunix: [ID 655072 kern.notice] fffffe80010bab60
zfs:zio_vdev_io_assess+82 ()
Apr  6 15:50:51 rbgz02 genunix: [ID 655072 kern.notice] fffffe80010bab70
zfs:zio_next_stage+72 ()
Apr  6 15:50:51 rbgz02 genunix: [ID 655072 kern.notice] fffffe80010babd0
zfs:vdev_mirror_io_done+c1 ()
Apr  6 15:50:51 rbgz02 genunix: [ID 655072 kern.notice] fffffe80010babe0
zfs:zio_vdev_io_done+14 ()
Apr  6 15:50:51 rbgz02 genunix: [ID 655072 kern.notice] fffffe80010bac60
genunix:taskq_thread+bc ()
Apr  6 15:50:51 rbgz02 genunix: [ID 655072 kern.notice] fffffe80010bac70
unix:thread_start+8 ()


We can provide more details if required...
 
 
This message posted from opensolaris.org