Date: Tue, 5 Oct 1999 09:30:15 -0400 (EDT) From: David Gilbert <dgilbert@velocet.ca> To: Cy Schubert - ITSD Open Systems Group <Cy.Schubert@uumail.gov.bc.ca> Cc: David Gilbert <dgilbert@velocet.ca>, freebsd-stable@FreeBSD.ORG Subject: Re: More on the crashes already mentioned. Message-ID: <14329.64871.579761.476450@trooper.velocet.net> In-Reply-To: <199910051323.GAA17221@cwsys.cwsent.com> References: <14329.4421.847994.918399@trooper.velocet.net> <199910051323.GAA17221@cwsys.cwsent.com>
next in thread | previous in thread | raw e-mail | index | archive | help
>>>>> "Cy" == Cy Schubert <- ITSD Open Systems Group <Cy.Schubert@uumail.gov.bc.ca>> writes: Cy> It sure sounds like memory. I can we assume you tried completely Cy> different memory chips? Even then that's no guarantee. Also try Cy> increasing the memory refresh rate. I had a panic with FreeBSD Cy> 2.0.5 was solved by increasing the refresh rate. If that doesn't Cy> help try reducing the memory. Some systems cannot refresh memory Cy> fast enough to be stable under certain memory access patterns. Cy> The quality of memory for PC's has always been scatological, Cy> however now that most memory is non-parity, you don't even get Cy> notified of single bit errors. It may even be worth it to Cy> purchase some ECC memory. IIRC ECC uses the Reed-Solomon Cy> algorithm (same as used by NASA for deep space probes), to detect Cy> and correct memory errors, so the rate of detection and correction Cy> of an error burst of 16 bits should should approach 99.99999%. Well... the same machine had an 80 day uptime before the upgrade (running 3.0-RELEASE). We have also tried alternate motherboards. Now... all this hardware is K6-2/400's --- it appears that substantial amounts of new code playing with processor bits has been introduced between 3.2 and 3.3 I also tried cutting the memory in half. The system had been running as configured for 80ish days w/o reboot. It has 2x 128M in it, which I reduced to 1x 128M... if anything, it crashed faster (although I can't say this for sure). The machine is running 100+ apaches that are ~88M virtual size. It's happening every 5 minutes now, it seems. sigh. Dave. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?14329.64871.579761.476450>