Date: Wed, 25 Dec 1996 19:10:08 -0600 From: dkelly@hiwaay.net To: Brandon Gillespie <brandon@glacier.cold.org> Cc: freebsd-hardware@freebsd.org Subject: Re: AHA 2940UW? Linux works.. Message-ID: <199612260110.TAA02014@nexgen.HiWAAY.net> In-Reply-To: Message from Brandon Gillespie <brandon@glacier.cold.org> of "Tue, 24 Dec 1996 10:29:00 MST." <Pine.NEB.3.95.961224102248.27912A-100000@glacier.cold.org>
next in thread | previous in thread | raw e-mail | index | archive | help
> I REALLY need help with this, our server is continually rebooting since we > installed the AHA 2940UW controller and new UltraSCSI drive. It is not a > heavilly loaded system, but it does run a database system that NEEDS to > syncronize its database before shutting down. Every time this system > reboots we loose critical data to the last backup. This is flat out > unacceptable. I really like FreeBSD much more than Linux, but at this > point others in the group are suggesting we move to Linux because they > know this hardware works fine in Linux. In my experience I've never known > a system to simply reboot when it was a hardware problem (generally if its > a hardware problem it simply freezes up or does not work at all). This is > working for two to three days when suddenly *wham* reboot. I recently had "system crash and reboot with hardware SCSI problems." I read the followups and didn't see any mention of what MB is being used. Recently I couldn't pass up an opportunity to buy a PCI MB w/ AMD 5x86/133 for $99. Paid $17 more for 256k extra cache. This was to upgrade my 486DX33 FreeBSD system at work. Took this machine (the one I'm on now) down to 48M from 80M and pulled my 2nd 2940 out, and put it on the new MB. (Hint: you are supposed to be laughing, I am, $200 of memory and a $150 controller, another $250 of SCSI drives, on a $116 MB/CPU). Anyhow, FreeBSD 2.1.5 crashed under heavy SCSI usage. Upgraded to 2.2-current, then 3.0-current, upgraded Adaptec BIOS from 1.10 to 1.16. No change. Compared GENERIC against my config file and tried all the AHC options. Crash usually said something like, "Panic: Page Missing". Finally figured out the cache on the MB was the single cause of my crashes. Later found a reference that knew everything there was to know about this cheapie MB: <http://users.aol.com/sdnd/vip.htm>. Thought I'd tried both WB and WT cache modes, but after reading the above page I got brave (or bored), put the cache back in, tried WT cahce mode, and its been running a month or so w/o any SCSI problems. Interesting thing about this problem was iozone couldn't make the system fail, "make world" with the source/obj on a SCSI drive was one of the best ways to crash (boot disk and running system are IDE). Another way to crash was to put a swap partition on a SCSI disk, but this was too good of a way to crash the system. My problem was hardware. -- David Kelly N4HHE, dkelly@hiwaay.net ===================================================================== The human mind ordinarily operates at only ten percent of its capacity -- the rest is overhead for the operating system.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199612260110.TAA02014>