Re: [suse-amd64] SLES 8 and Oracle
I have finally figured out the problems I was having, and witnessed insanity in the process. The patient is a MSI K8D Master F/FT board with dual Opteron 244 on which I attempted to install an Adaptec 29160 SCSI HBA and SLES 8. It turns out that the release that comes on the CDs (32 or 64 bits) simply will not boot up on that board with the BIOS that comes with it if I installed it with the 29160 also installed. The installtion procdure recognizes the 29160, and configures it into the initrd file. If I installed SLES 8 without the card installed, there is still a chance that the OS will hang on bootup. Odd that the install procedure seems to work with it just fine, and the "safe" mode boots up. I had to get a special BIOS from MSI just to get SLES 8 to boot up more often than it hangs, so there is some incompatibility with this OS and the K8D Master board. The added confusion comes in where the Adaptec HBA would not work properly with the new BIOS; it would just hang up at its BIOS start up. This happens with two boards that I tried; Adaptec 29160 and 3ware Escalade 7506. So here is the sequnce of events that worked for me: First install the OS with the HBA installed using the old BIOS. This puts the configuration for the HBA into initrd so that when the system is working, it will automatically configure the HBA into the system. Now the system will hang on bootup, so I need to pull out the HBA and install the new BIOS so the OS would boot. Then use YOU to update the OS so it gets the updated drivers to handle the board properly (the ISAPnP configurator patch seems to be key here). Then re-install the old BIOS before plugging the HBA back into the system, and the next time it boots, everything comes up as expected. Thanks to Joe Bedard for pointing out the need for GCC 2.95 with the 32 bit version of Oracle 9i. We must have had it configured properly at one time, as we did install one data base successfully. However, we were informed by Oracle that there was some bug with an early release of SP3 for SLES8/x86 that did something to the GCC 2.95 libraries, and we suspect this is why we were no longer able to install any more data bases after our upgrade. The upgrade did fix the hang on boot issue, as well as the addressing of disks greater than 1 TB. Joe also mentioned that the 64 bit version of Oracle 9i has been certified on SLES8/x86_64, so we started toward that direction, which was what I wanted to do when I started this a couple weeks ago, and will abandon the 32 bit SLES8. I have just finished the basic installation of the 64 bit SLES 8 off the CDs, and will try to do the online update as soon as the OS lets me connect to www.suse.com. For some reason, when I first start up the system, it just does not want to connect to any web servers outside of my immediate subnet. It's not a routing issue, as I can ping the web server just fine, and I get a rejection message when I try to telnet to it, but Mozilla just hangs when I try to make a connection. So of course YOU does not work during this time. Then some random time later, the web access just magically starts to work all by itself. eyc PS: I wonder if SuSE will give me a refund for the SLES8 X86 after I've held on to it for so long?
On Mon, May 17, 2004 at 09:51:17PM -0700, Eugene Chu wrote:
The patient is a MSI K8D Master F/FT board with dual Opteron 244 on which I attempted to install an Adaptec 29160 SCSI HBA and SLES 8. It turns out that the release that comes on the CDs (32 or 64 bits) simply will not boot up on that board with the BIOS that comes with it if I installed it with the 29160 also installed. The installtion procdure recognizes the 29160, and configures it into the initrd file. If I installed SLES 8 without the card installed, there is still a chance that the OS will hang on bootup. Odd that the install procedure seems to work with it just fine, and the "safe" mode boots up.
That sounds more like a hardware problem. AFAIK people have used SLES8-SP3 on K8D successfully. I would first check out if your system runs memtest86 for at least 24h to rule out any memory issues. -Andi
participants (2)
-
Andi Kleen
-
chu@tes-mail.jpl.nasa.gov