vmunix: SCSI: Request Timeout error in syslog

[Magnus Andersen]

Hi Everyone,

I've has SCSI errors in my syslog for the last two days. I've looked around, but I couldn't find any solutions. Attached is a tail on my syslog with the errors.

Any help is much appreciated.

Thanks,
Magnus

[Cheryl Griffin]

You might have a disk dropping offline or a conflicting SCSI ID.

[Chris Wilshaw]

The first thing to try is to increase the timeout on the disk - it is quite often the case that a disk doesn't respond to an I/O request in time.

pvchange -t 180 /dev/dsk/cXtYdZ

This will set the timeout to 180 seconds, and may prevent your errors. If the errors still show up after doing this, it is more likely to be a hardware error.

[Elena Leontieva]

Magnus,

According to some old document these errors can be due to:

1. The IO timeout setting for the physical disk being set too low. Adjust this parameter as suggested above and monitor the /var/adm/syslog/sylog.log file to see if it corrects the errors.
2. A problem with the SCSI connector for that device. This will require a hardware ticket be opened with the Hardware Support Team.
3. Not having the latest SCSI/LVM patches installed.

HTH,
Elena.

[RAC_1]

SCSI serrors could be due to following.

1. Not up to date on SCSI/LVM patches.
2. connections/connectors problem SCSI devices. (loose connections etc)
3. Timeout settings for PVs.

For
1. Always be up to date SCSI/LVM pacthes.
2. As far as possible do not connect new devices on line. Also whenever you shutdown systems, make sure your connections/connectors are Ok
3. The default timeout setting is 60 sec.
make it 120--> pvchange -t 120 /dev/dsk/cxtxdx

[Steven E. Protter]

You have a bad disk or scsi cable or something is not terminated correctly.


If the message does not go away on booting the machine, you will most likely be replacing a disk soon.

Make sure you are all set on backups, make_tape_recoery and data backups.

SEP

[Bryan D. Quinn]

Hello,

I have had this issue on two different servers in the past and I had two different solutions. I know both of these have been mentioned already. I am just adding my experiences here for any benefit that it may provide. So, please, NO POINTS.

1. On one server I had to update SCSI patches and this took care of the errors.

2. On the other server I had to up the PV timeout on my EMC disks that were still at the default setting.

So, from my experience with this type of error, I would say make sure SCSI patches are up to date and check timeout on pv's with pvdisplay.

Oh, I think someone else mentioned the LVM patches also, wanna make sure those are up to date.

-Bryan