SCSI read write errors - disk failed

[Mark van Hassel]

Hi,

During the last week we've had several disk failures on a system. It is always the same (SEAGATE) disk. We get the following messages in the syslog.log every second:
vmunix: blkno: 16459512, sectno: 32919024, offset: -325328896, bcount: 8192.
vmunix: blkno: 16876632, sectno: 33753264, offset: 101801984, bcount: 8192.
vmunix: SCSI: Write error -- dev: b 31 0x04d000, errno: 126, resid: 8192,
vmunix: SCSI: Async write error -- dev: b 31 0x04d000, errno: 126, resid: 8192,
vmunix: blkno: 16459504, sectno: 32919008, offset: -325337088, bcount: 8192.
vmunix: blkno: 16772576, sectno: 33545152, offset: -4751360, bcount: 8192.
vmunix: SCSI: Async write error -- dev: b 31 0x04d000, errno: 126, resid: 8192,
vmunix: blkno: 15393840, sectno: 30787680, offset: -1416577024, bcount: 8192.
vmunix: SCSI: Write error -- dev: b 31 0x04d000, errno: 126, resid: 8192,
vmunix: blkno: 16489200, sectno: 32978400, offset: -294928384, bcount: 8192.
vmunix: SCSI: Read error -- dev: b 31 0x04d000, errno: 126, resid: 2048,
vmunix: SCSI: Async write error -- dev: b 31 0x04d000, errno: 126, resid: 8192,
vmunix: blkno: 8, sectno: 16, offset: 8192, bcount: 2048.
vmunix: SCSI: Async write error -- dev: b 31 0x04d000, errno: 126, resid: 8192,
vmunix: blkno: 16478648, sectno: 32957296, offset: -305733632, bcount: 8192.
vmunix: SCSI: Read error -- dev: b 31 0x04d000, errno: 126, resid: 2048,
vmunix: blkno: 8, sectno: 16, offset: 8192, bcount: 2048.
vmunix: SCSI: Async write error -- dev: b 31 0x04d000, errno: 126, resid: 8192,
vmunix: blkno: 16907184, sectno: 33814368, offset: 133087232, bcount: 8192.


Ioscan reports NO-HW and diskinfo fails.
HP has replaced the disk several times. The last disk they replaced they tested in their labs and it worked fine !

Has enayone encountered something like this before ?

Thanks in advance,

Mark

[Vincent Farrugia]

Hello,

If it is not a disk problem, then it is a communication problem. How is this disk connected? What is the server or disk array where it lies?

Vince

[Paula J Frazer-Campbell]

Mark

These errors on block number indicate that it is the disk at fault again.

THe errors are being reported from the disk itself which indicates that the system is "Seeing" all the way to the disk.

To err on the safe side re-seat all cables in the chain and then get a replacement disk.

HTH

Paula

[Mark van Hassel]

Thanks for the quick responses.
The disk is one of 8 in a HASS disk cabinet. The backplane of the HASS has been replaced as well as the ( FW differential) SCSI cables and the SCSI controller.
When I pull out the disk and replace it, the disk is finem again, however after a few hours the messages appear again and the disk is not accessible.

Mark

[Paula J Frazer-Campbell]

Mark

Check the firmware level on the disk.

Also has mediainit been used on this disk? as this could couse a problem.

Paula

[Mark van Hassel]

No mediainit has been used. Firmware lvl is HP05, the same as the other disks.

[Paula J Frazer-Campbell]

Mark

Can you move this disk to another slot in the HASS ?

My feeling is still a bad disk -

An Idea !!

Can you zap it and do a newfs.

Paula

[Mark van Hassel]

All slots are occupied. The disk has been replaced 6 times over a period of three weeks, so the chance that it is a bad disk is very slim in my opinion.
I can't do a newfs, because it is part of a fs that is spread over multiple disks.

[Ashwani Kashyap]

Hi ,

I remember I had encountered something like that . Everythinh showed towards a disk failure . Finally I got the SCSI controller replaced and that took care of the problems .
Try the same thing and see if it helps .

Thx