What kernel are you running ? There was a problem with the MTRR code in
the 2.4.21-120 kernel you get if you just load from the CDs or DVD that
caused driver problems if you have more than 4 GB. I have several S2885s
with 8 GB of memory that have been up for months now. They are running
We did see some problems on Opterons with the 193 kernel in our Singapore
office using an Adaptec 29160 SCSI card. It seemed to be tied to a newer
version of the Adaptec driver that SuSE was using compared to an older
driver on some P4 systems running RedHat. The new driver had some new SCSI
functionality (I think what was causing us problems was Domain Validation
and the attached RAID box did not support Domain Validation). This
functionality was not in older versions of the Adaptec driver, so the
problem did not show up.
In /var/log/messages, you would see the following info when the kernel
loaded the driver for the card...
Mar 5 15:29:16 ppc003 kernel: PCI: Enabling device 02:01.0 (0015 -> 0017)
Mar 5 15:29:16 ppc003 kernel: scsi0 : Adaptec AIC7XXX EISA/VLB/PCI SCSI
HBA DRIVER, Rev 6.2.36
Mar 5 15:29:16 ppc003 kernel:
Mar 5 15:29:16 ppc003 kernel: aic7892: Ultra160 Wide Channel A,
SCSI Id=7, 32/253 SCBs
Mar 5 15:29:16 ppc003 kernel:
Mar 5 15:29:31 ppc003 kernel: scsi0:A:0:0: DV failed to configure device.
Please file a bug report against this driver.
Mar 5 15:29:35 ppc003 kernel: (scsi0:A:0): 160.000MB/s transfers (80.000
MHz DT, offset 31, 16bit)
Mar 5 15:29:35 ppc003 kernel: Vendor: BROWNIE Model: 1600U3P Rev:
0001
Mar 5 15:29:35 ppc003 kernel: Type: Direct-Access ANSI SCSI revision:
03
Mar 5 15:29:35 ppc003 kernel: scsi: host 0 ch 0 id 0 lun
0x42004f574e494520 has a LUN larger than currently supported.
There were also problems with partitions larger than 1 TB. We had to have
partitions smaller than 1 TB, and ended up dropping back to a P$ running
RedHat 8, as it did not have the problems with the Adaptec controller....
We do have quite a few clusters running the Arima motherboards, but we use
Qlogic and LSI FC cards...
Kevin
Kevin Gassiot
Advanced Systems Group
Visualization Systems Support
Veritas DGC
10300 Town Park Dr.
Houston, Texas 77072
832-351-8978
kevin_gassiot@veritasdgc.com
Bryan Stillwell
To
Kevin_Gassiot@veritasdgc.com
05/03/2004 03:32 cc
PM suse-amd64@suse.com
Subject
Re: [suse-amd64] Dual-Opteron w/
8GB RAM boot problem
I was running 2.4.21-149-smp, but now I'm using 2.4.21-211-smp after
running YOU (I'm somewhat new to SuSE, so I didn't know about it). The
problem is still there though... :( I'm beginning to think there's
something tied in with the SCSI subsystem (it died in scsi_do_req_Rsmp
once). Are you using any SCSI cards or just IDE?
The weird thing is it worked in the Rioworks board using the same SCSI
card...
Bryan
--
Aspen Systems, Inc. | http://www.aspsys.com/
Production Engineer | Phone: (303)431-4606
bryans@aspsys.com | Fax: (303)431-7196
On Mon, May 03, 2004 at 02:22:55PM -0500, Kevin_Gassiot@veritasdgc.com
wrote:
the
2.4.21-193-smp kernel and Nvidia 5332 graphics driver. I had a beta BIOS
at first, but the last ones that I built are running the v102 BIOS that
fixes the MTRR layout that lets the 193 and later kernels and Nvidia
drivers work correctly together... I do have one running the v102 BIOS
and
2.4.21-201-smp kernel that has been stable, but it gets booted between
Windows, RedHat 9, and SuSE 9.0 so much that I can't tell anything about
long-term stability....
To get things going when loading from the CD, I had to remove memory to
get
to 4 GB or less, flash the BIOS, install from CDs, upgrade the system via
YOU, then put the memory back in. From there, I could install the Nvidia
drivers, and go....