[TriLUG] SCSI Timeout: kernel? hardware?

[Scott Chilcote]

Hi Janyne,

This may not pertain to your problem.  I just fixed a SCSI timeout 
problem on a system that has a AIC-7899 U160 controller on the mainboard.

This is a dual-bus controller.  On one side I had U160 drives, on the 
other side I had a SCSI CD-ROM and an external scanner.

The problem was that for this controller, slow devices require special 
(non-default) BIOS configuration and single-ended termination.  I had 
previously been using them on a 7895 with both ends terminated and 
didn't have any problems, but I'm new to U160 setup.

Basically I had to turn off U160 support and reduce transfer rates for 
the 2nd bus on the controller.

Good luck,

                                 Scott

Janyne Kizer wrote:
> I have two servers that are consistently reporting what appears to be a
> SCSI timeout error.  I have two other servers with the same model
> hardware (one of which is literally a copy of the one below) that are
> not reporting this problem.  The error that I am looking at (AAC:SCSI
> Channel[0]: Timeout Detected On 1 Command(s)) always occurrs after the
> "su(pam_unix)[12432]: session closed for user nobody" but as you can see
> there is no time relationship.  I'm not sure what is going on here and
> anything to point me in the right direction would be appreciated.  The
> hardware is Dell PowerEdge 2400.  The OS is RH 7.2.  The BIOS and PERC2
> are up-to-date.  I guess one of my main questions is:  Is this a kernel
> issue or a hardware issue?