interpreting error messages

[Arnold_14]

In /var/adm/syslog/syslog.log, I see a bunch of these:

Apr 17 00:52:40 skyplan1 vmunix: SCSI: Async write error -- dev: b 31 0x0e0000, errno: 126, resid: 65536,
Apr 17 00:52:40 skyplan1 vmunix: blkno: 76238336, sectno: 152476672, offset: 758644736, bcount: 65536.

Apr 17 00:52:40 skyplan1 vmunix: SCSI: Async write error -- dev: b 31 0x0e0000,
errno: 126, resid: 32768,

I was wondering if I could get some help interpreting those messages. I understand it's some kind of SCSI error, but maybe you can help me with more details.

By the way, this server has been experiencing a slow Disk I/O performance, specially when it tries to write big files to the SAN storage.

Thanks
Arnold.

[Victor Fridyev]

Hi,
0x00e000 means disk c0t14d0,
run:
ls -ls /dev/dsk/*|grep -i 0x00e

I'm not sure, but error 126 may mean "powerfail". Usually such an error precede the disk physical failure.
Could you replace the disk ? In any case back up the data located on it.

Good luck

[Michael Tully]

What sort of SAN storage is this and how is it set up? We could be able to help further knowing this.

[Arnold_14]

The SAN's type is EMC that's connected through 2 Tachyon cards. I've been doing some testing by writing 100Mb file to the internal disks and to the SAN. The write time to the SAN is considerably much bigger than the time to write to the internal disks. I'm sorry I could not give you more detail about the SAN. I'm not really a SAN admin.

I've been doing some searches on Google, and they all point to bad disks or maybe bad cables. Someone also tried to change the SCSI controller and it fixed the problem. How to test if my fibre cards are good/bad?

[Victor Fridyev]

If you have an alternate path to the disk, check with dd copy via both paths whether you obtain errors. According to such a test you can clarify which of channels has problems.