topic Re: SCSI error in Operating System - HP-UX

SCSI error

Robert Milne — Tue, 21 Jan 2003 06:25:06 GMT

Hi there,

I noticed the following message in our syslog.log file:

Jan 21 14:31:45 vmunix: scsi_probe: when probing target 6, lun 1 (dev=cb04
6100): couldn't probe lun 0 (no longer responding?)

I am not aware of any other problems with the system. How can I investigate this error further to find out what the problem is what may be failing, or about to fail ? No other error messages at all have appeared in the log. What other logs may reveal more information as to the problem ?

We have HP-UX 11i on a 9000 L1000 server, with SCSI mirrored disks and LVM.

Any assistance appreciated.

All the best.

Rob.

Re: SCSI error

Eugeny Brychkov — Tue, 21 Jan 2003 07:27:47 GMT

Rob,
attach 'ioscan -fn', 'diskinfo -v /dev/rdsk/c4t6d0' and 'diskinfo -v /dev/rdsk/c4t6d1' to your next reply
Eugeny

Re: SCSI error

Armin Feller — Tue, 21 Jan 2003 08:34:57 GMT

Hi Rob,

please check also if the newest SCSI patch is installed.

PHKL_22759 SCSI IO Subsystem Cumulative Patch

If not please install it.

Regards ...
Armin

Re: SCSI error

Robert Milne — Wed, 22 Jan 2003 04:05:34 GMT

Hi Eugeny and Armin,

Thanks for your prompt response !

Please see the ioscan attached as ioscan.txt.
I have no device files of the name c4t6d0 or c4t6d1 for disks. But do have device filenames of c4t6d0BESTxx for a DLT8000 tape device.

I was interested in how you determined the relationship of the error reported where dev=cb046100. Does this number translate to a device file minor number like 0x046100 in which case the SCSI card instance number is 4, the SCSI target is 6 and the device number or LUN is 1, therefore producing a device file name of the type c4t6d1 or c4t6d0 for LUN 0 ?

If so then the only device on the SCSI card with instance 4 is the tape device. As this is a single tape drive, not a library, I assume there would be no LUN 1 only 0. Possibly there may be a problem with the tape drive, although it was not in use at the time of the error message. I am not quite sure why such an error would appear, any suggestions ?

On the patch you mentioned I am not quite sure how to check this, but don't believe we have installed any patches for about 9 months. I did an swlist -l patch | grep SCSI and found the following entries which may be relevant:

PHKL_22857 SCSI (stape) cumulative
PHKL_23666 SCSI IO Subsystem cumulative
PHKL_25896 SCSI IO Cumulative patch

How can I find out if the new patch you have recommended may solve my problem, or indeed what problems it addresses ? I am not adverse to putting in patches, but with the downtime etc. required (this is a main production database/application server), will usually only put in patches to address specific issues. I don't usually just install all of the latest greatest if the system is stable. But then if it's not I will !

All the best,

Rob.

PS so far error has not re-appeared.

Re: SCSI error

Michael Tully — Wed, 22 Jan 2003 04:40:06 GMT

Hi,

Yes indeed this is how these devices are worked out
cb4= controller 4
target 6 = target 6
LUN 0 = LUN0
/dev/???/c4t6d0

It looks to me as if there was some sort of SCSI reset. What you could do is disable EMS from attempting to do any sort of scsi probe on the device, as well having your kernel parameter 'st_ats_enabled' set to '0'.
Attached is a document that explains how to turn off various pieces of EMS monitoring.

The latest patches are:
PHKL_28096 SCSI IO cumulative
PHKL_28217 SCSI Tape
PHKL_23666 SCSI Subsystem

You can get to the patch area, from a link below this message in the blue box called 'download hp-ux patches in the Patch Database.

I'm not suggesting you load them seeing everything is going okay at the moment, but it might be something to look at later. You'll most likey find that the issue is more hardware related in any case. Tape drives cannot be relied upon forever, no matter how many patches are installed.

HTH
Michael

Re: SCSI error

Robert Milne — Thu, 23 Jan 2003 06:05:00 GMT

Hi Michael,

Thanks for your info.
Yes, as it does appear to be the tape drive, from the device number, and so far it has not "re-offended". So I'll just keep my eye on it. You're right about tape drives not lasting forever, this is my third one in one year ! Previous ones had the tape cartridge jam in them due to some sort of mechanism failure. And it wasn't the tapes with the leaders breaking etc.

I still have an old Digital 20GB DLT (a TZ87) from the pre Quantum days, which is still going strong. They just don't make them like they used to, and HP were charging twice the price for the equivalent drive in the pre Compaq merger drives for the luxury of this drive. Thank god for maintenance contracts.

Thanks for your help, no I award the points !

Regards,

Rob.