SCSI: request timeout

[Mark Ellzey]

I have a C360 w/ 2 internal drives. The root drive is fine, but the other drive keeps loading up the syslog file with the attatched SCSI error message. Has anyone else seen this, and if so, does it mean a hardware failure, or just a buggy SCSI interface?

Regards,
Mark

P.S. This threadhas beenmoevd from General to HP-UX > sysadmin.- Hp Forum moderator

[A. Clay Stephenson]

You might try using pvchange to increase the timeout value on the drive. It is possible that the two drives have different timeout values. The other usual SCSI suspects (terminators, cables seated, terminator power, terminated in exactly two places, ...) should also be examined. If the above does not fix it and the problems persist then it's time to start looking at the drive itself.

[PIYUSH D. PATEL]

Hi,

It could be a timeout problem also whoich could be changed with pvchange command.

Check out for loose connections and scsi termination.

If not then there seems to be a problem with the harddisk drive.

# dd if=/dev/rdsk/c1t2d0 of=/dev/null bs=1024k

to check the second harddisk drive.

If there are bad sectors then the output of the command will show.

Piyush

[Sanjay_6]

Hi,

You can change the timeout using the pvchange command,

pvchange -t 180 /dev/dsk/cxtydz

the timeout value can be upto 300 seconds. By default it is 30 seconds. If you are getting a lot of these errors you should replace the disk. I always prefer replacing the disk than changing the default timeout.

If this is an autoraid, you can use the logprint / arraydsp to check the array logs and decide which disk in the array might need replacement.

1f005000 --> c0t5d0

Hope this helps.

Regds

[Mark Ellzey]

Thanks for the suggestions, but this disk is HFS, not LVM. As a result, pvchange won't work.

[Paula J Frazer-Campbell]

Mark

Instead of looking at changing the timeout and therefore hiding the problem look at the fault. it is in the SCSI chain.

I suggest that you shutdown, power off and then reseat all devices in the path.

BACKUP

1. Reseat cards
2. Unplug and replug all cables and terminators several times (this will clear any oxidation on the contacts.


Bring the server back up and monitor.

If the problem persits then swap out the disk.


HTH

Paula

[Sanjay_6]

Hi Mark,

Check the termination. If that is fine, replace the disk.

Hope this helps.

Regds

[S.K. Chan]

To echo what the rest have said, it's probably more likely that you would have to replace the disk. In the past I've had a lot of SCSI reset error on and HFS disk and no matter what I do (plug, unplug, replace cable, etc) it does not help. But that's just what I experience.

[Michael Lampi · Accepted solution]

The log shows that the timeouts are occurring during a "2A" command, which is a SCSI write.

This would indicate that the drive is having troubles writing to media, and the retry mechanism within the drive is taking too long to respond before the HP SCSI driver times out.

You could try increasing the HP SCSI driver timeouts, or just replace the disk drive.

While it is possible that the SCSI bus termination is at fault (double check to make sure that the external terminator is in place), you might have a failing drive.