On Sun, Jun 27, 2004 at 05:56:10PM +0100, Tom Shaw wrote:
Jun 24 00:03:46 digit kernel: I/O error: dev 08:30, sector 47436960
That's coming from the SCSI code (__scsi_end_request).
8:30 could be /dev/sdb14 which sounds a bit unusual, maybe it aliases
to something else too?
Jun 24 00:03:46 digit kernel: I/O error in filesystem ("lvm(58,0)")
meta-data dev lvm(58,0) block 0x3ca5d320 ("xfs_trans_read_buf")
error 5 buf count 4096
XFS, error message because it the error above handed to it. At this
point XFS can't really cope with things and you'll loose access to the
filesystem to prevent further problems.
I have done a badblocks read-write test on all of the drives and
nothing shows up, so the drives are not at fault as far as I can
tell.
Do the errors occur at the same place or different places? If the
latter and badblocks shows clean then it might be sporadic corruption
(bad cables, power supply, the 500 tons of yellow cakes stored in your
basement, etc).
One small note here...what is 8:30?? I have looked in
/proc/partitions but it is not listed, nor can I find it in my
linux-2.4.26/Documentation/devices.txt.
Check the device files:
cw@pain:~$ ls -l /dev/sd*|grep "^b.*8, .*30"
brw-rw---- 1 root disk 8, 30 Mar 14 2002 /dev/sdb14
Does that sound reasonable? Do you even have such a device/parition?
--cw