kernel: scsi : aborting command due to timeout

[Edwin Ruiz_2]

Hi:

My server Red Hat A.S. 2.1 was blocked, i could not login with any user, now i can login and i saw that messages on /var/log/messages. Whats wrong. Pls Help me!

Jun 7 14:31:33 salato-ap11 kernel: scsi : aborting command due to timeout : pid 0, scsi0, channel 2, id 0, lun 0 Read (10) 00 0a 49 88 de 00 00 08 00
Jun 7 14:31:33 salato-ap11 kernel: scsi : aborting command due to timeout : pid 0, scsi0, channel 2, id 0, lun 0 Write (10) 00 0a 15 ae 9f 00 00 08 00
Jun 7 14:31:33 salato-ap11 kernel: scsi : aborting command due to timeout : pid 0, scsi0, channel 2, id 0, lun 0 Write (10) 00 09 ef e5 af 00 00 08 00
Jun 7 14:31:33 salato-ap11 kernel: scsi : aborting command due to timeout : pid 0, scsi0, channel 2, id 0, lun 0 Write (10) 00 09 ef e5 af 00 00 08 00
Jun 7 14:31:55 salato-ap11 sshd(pam_unix)[24811]: session opened for user root by (uid=0)

[Stuart Browne · Accepted solution]

Hrm.. Timeout's reading or writing to your SCSI bus. Not the happiest of things to read.

This could mean a few things unfortunately. It could mean something as simple as an incorrectly seated cable.

It could also mean bad termination on the SCSI chain.

It could also mean a damaged cable.

It could also unfortunately mean a bad disk.

By the same token, it could just be a SCSI driver bug.

Start by making sure you're on the latest kernel for AS2.1.

Then continue by powering down, checking cabling and termination.

If it still occurs after doing all of this, try un-populating the SCSI chain (removing CD-ROM's, Tape drives, external devices from the same chain).

If after all of this, they still occur, then it would sound as if your disk is on the way out.

Does it make strange sounds when these errors occur, or does it just have normal spinning-sounds?

[Edwin Ruiz_2]

Thanks Stuart!

I have one DLT tape conected in same SCSI channel that my disks are. Server is one HP LH4. but if i connect in other SCSI Channel Redhat Linux A.S. 2.1 cant detect the device. what is ur recomendation?

Thanks for ur help!

[Steven E. Protter]

For good througput the makers of most tape/dlt/ultrium drives recommends that the tape device get its own scsi chain.

To do otherwise adversely impacts performance.

SEP

[Thomas Bianco]

stuart has is right.

a few more unlikly things that could cause this:

the cable takes a corner too tight. check cable routing for very sharp bends. very unlikly on modern LVD scsi busses, but the old 30v ones could have some interesting problems here.

noise on the cable. we had a DEC server with intermittent scsi problems. the cable was too close to the AC, and when the dehumidifier kicked on it drowned out the scsi signals. proper gounding on the external cable fixed that.

[Stuart Browne]

An LH4 comes standard with 2 on-board SCSI controllers, and 3 bus'.

One of the on-board controllers is a MegaRAID controller.

The BIOS can control whether the two main bus' are dedicated to the MegaRAID controller, or whether one is for RAID, the other is for the 2nd controller (I beleive it's a symbios 53c8xxx badged thing).

The 3rd bus is for an external interface, and is not wired out-of-the-box (it's at 90 degree's to the other two edge-based ports).

The core two and labeled A and B.

If you're using the onboard MegaRAID, and you've got both A and B as benig part of the MegaRAID, and you have two how-swap cages, then that's the end of the ball game.

If you've only got the one cage (standard), and are only using one of the A/B ports for disks currently, then you can go into the BIOS, set port B to be just plain 'on-board SCSI', and put your tape drive on that chain instead.

[Edwin Ruiz_2]

Ok! thanks.. i will check that configuration..