SCSI Errors in Syslog

[Rob Smith]

Hello All,

The other day these errors appeared in the syslog on our k box;

Oct 29 03:03:59 hpqad vmunix:
Oct 29 03:03:59 hpqad vmunix: SCSI: Read error -- dev: b 31 0x11000, errno: 126,
Oct 29 03:03:59 hpqad vmunix: blkno: 335256, sectno: 670512, offset: 34330214.
Oct 29 03:05:59 hpqad vmunix: LVM: PV 0 has been returned to vg[2].
Oct 29 03:05:59 hpqad vmunix: LVM: PV 1 has been returned to vg[2].
Oct 29 03:15:59 hpqad vmunix:
Oct 29 03:15:59 hpqad vmunix: SCSI: Read error -- dev: b 31 0x9000, errno: 126,,
Oct 29 03:15:59 hpqad vmunix: blkno: 333224, sectno: 666448, offset: 34122137.
Oct 29 03:15:59 hpqad vmunix: LVM: vg[2]: pvnum=0 (dev_t=0x1f009000) is POWERFAD

These errors appeared during a backup job, all disks are present and in sync. There was no power failure at the time. Any ideas?

Thanks in Advance,
Robert

[Paula J Frazer-Campbell]

Hi
Check out

http://forums.itrc.hp.com/cm/QuestionAnswer/1,1150,0x4cf4f841489fd4118fef0090279cd0f9,00.html
It may point You in the right direction.

Paula

[Patrick Wallek]

There was a thread on an error message very similar to this with the last few days. Check out this link:

http://my1.itrc.hp.com/cm/QuestionAnswer/1,1150,0x4cf4f841489fd4118fef0090279cd0f9,00.html

[Michael Lampi]

You have two disk drives that are exhibiting read errors. The first one, on bus 0x01, SCSI ID 1, seems to currently have a recoverable block.

The second one on bus 0x00, SCSI ID 9, seems to have an unrecoverable error, or at least bad enough that the system chose to powerfail the drive. This also assumes that there were no subsequent messages indicating that the drive was returned to the volume group.

Some of these errors can be reduced, masked and/or eliminated by changing the I/O timeouts from the default to 180 seconds or more. In general, however, the default value is quite adequate to handle normal drive seek errors, so increasing this value serves mainly to mask the fact that the disk drives have one or more bad spots that are not in the drive defect tables.

I would recommend backing up the data on these drives and running mediainit to reformat and verify them. If they pass mediainit's scrutiny, then you can restore your data and continue to use the drives. Otherwise, replace them at your next opportunity.

[Michael Lampi]

You have two disk drives that are exhibiting read errors. The first one, on bus 0x01, SCSI ID 1, seems to currently have a recoverable block.

The second one on bus 0x00, SCSI ID 9, seems to have an unrecoverable error, or at least bad enough that the system chose to powerfail the drive. This also assumes that there were no subsequent messages indicating that the drive was returned to the volume group.

Some of these errors can be reduced, masked and/or eliminated by changing the I/O timeouts from the default to 180 seconds or more. In general, however, the default value is quite adequate to handle normal drive seek errors, so increasing this value serves mainly to mask the fact that the disk drives have one or more bad spots that are not in the drive defect tables.

I would recommend backing up the data on these drives and running mediainit to reformat and verify them. If they pass mediainit's scrutiny, then you can restore your data and continue to use the drives. Otherwise, replace them at your next opportunity.

[Victor Mendham]

Just wondering if you know if these were real errors yet? I need
to monitor SCSI errors in IT/O and I haven't seen any. So I'm wondering
if I can follow this syntax for my testing? Pleas elet me know if these were real errors.

[Lawrence Mahan]

I have seen these type of error in the pass when I had a bad or missing terminater. Try replacing the SCSI Terminator and see this the error go away.