SCSI: Read error SCSI Write Error

[Madanagopalan S]

We are using L1000/HA enclosure and FWSCSI:

We have been facing SCSI READ Error and SCSI
Write Error from the Harddisk in HA enclosure.

Every time HP replacing the Harddisk. Then I forced them to check the root cause. Then they replaced the HA enclosure back plane, scsi card and scsi cable. But still problem coming,
yesterday the same problem came and I restarted the server and the messages disappeared.

Anyone faced this type of problem and Pls help me to resolve this issue.

The error message in the syslog is:

SCSI: Read error -- dev: b 31 0x045000, errno: 126, resid: 32768,
blkno: 3862912, sectno: 7725824, offset: -339345408, bcount: 8192.
SCSI: Write error -- dev: b 31 0x045000, errno: 126, resid: 8192,
blkno: 4018520, sectno: 8037040, offset: -180002816, bcount: 8192.

[Stefan Farrelly]

We see these ocassionally too, and all it really means is at some point the disk(s) were asked for a read or write and did not complete the request within a certain amount of time.

If youve replaced everything and you still get these errors chances are these disks are simply working too hard to complete all requests withing the normal i/o timeout period (in LVM), so the thing to try next is increase the timeout value using pvchange -t . Default is 30 so bump up to say 60 and see if that fixes your errors.

[Sanjay_6]

Hi,

Try and see if you can replace the deivce c4t5d0. That might help.

Hope this helps.

regds

[Michael Steele_2]

c4t5d0 reads like a cdrom, I don't mean to state the obvious but was this replaced?

Monitor HW addresses via STM:

STM > TOOLS > UTILITY > RUN > LOGTOOL > FILE > VIEW > RAW SUMMARY.

Note the first and last dates of transactions and calculate the difference. If the difference is short, like 4 hours, then this is important to note. Now read down the report of hardware addresses and observe the integer numbers in parenthesis. Anything over 150 in this 4 hour period should be called into HP for replacement.

LOGTOOL is what HP CE's depend upon for evaluating HW. If you refer to this then they will listen.

Keep on them. Sometimes its takes longer then expected.

[Madanagopalan S]

Sanjay, that was HP recommendation, but this is 4th time we are replacing this harddisk. They could not find out the problem.
some times I receive message like lot i/o erros received from i/o subsystem. This problem is intermittent. yesterday after I restart the system , i can able to mount the volume and no messages. Is it possible to simulate this problem. In what circumstances, the read or write error will occur. I tried to do dd on all the disks at the same tiem but I did n't receive any error. If its harddisk problem then it suppose to say read error while doing dd right?

[Navid HUSSAIN_2]

Hi

Did you check harddisk uncorrected entry via expert tool
(It is password protected, Call HP Engineer)

Run STM .. Logtool to check logx.raw.cur ..Format the log and viewlog in details.

Does it point to same c4t5d0 disk. Check Disk Firmware..(may be latest has released)

May be SCSI Cumulative patch issue. Send detail like UX Version . Swlist and syslog output to get more details

Work on above hints..

Thanks

NH

[Eugeny Brychkov]

You did not mention SCSI terminators. Each SCSI bus should be terminated at the end. You should use HP A1658-62024 (p/n C2905A) FWD terminator. If both A and B HASS channels are daisy chained then there should be one terminator at the remaining HASS SCSI connector (at the bus end). If they're not daisy chained and connected to different cards, then there should be terminator at the each spare SCSI connector (each bus,A and B, should be terminated)
Eugeny