Date: Sat, 30 Oct 2004 20:32:31 -0400 From: Louis LeBlanc <FreeBSD@keyslapper.org> To: FreeBSD_Questions FreeBSD_Questions <freebsd-questions@freebsd.org> Subject: Re: More Intel ICH5 SATA150 hangs Message-ID: <20041031003231.GD9405@keyslapper.org> In-Reply-To: <69B28466-2A27-11D9-B312-000393BB56F2@HiWAAY.net> References: <20041029223412.GD4114@keyslapper.org> <clo5o0tnm80sddh2sovffd0t7avaf49lcc@4ax.com> <69B28466-2A27-11D9-B312-000393BB56F2@HiWAAY.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On 10/29/04 10:54 PM, David Kelly sat at the `puter and typed: > > On Oct 29, 2004, at 7:48 PM, Mike Tancsa wrote: > > >> Same result the other poor schmuck got when he tried it. There was > >> also a suggestion that HT be turned off. Kinda defeats the whole > > > > HT works against you unless the scheduler is specifically aware of it, > > and right now the ULE scheduler is broken. For sure turn HT off in > > your BIOS and make sure you use SCHED_4BSD and NOT ULE > > I turned off HT on my Dell PE400SC (uses Intel ICH5 chipset) almost > instantly after installing FreeBSD. Partitioning and labeling my SATA > drives went just fine. System is on a PATA drive. Even striping the two > SATA drives with Vinum followed by newfs was fine. System locked hard > the instant I actually tried to use the SATA drives. Reformatted > without vinum and had similar lock/panic. This was possibly 5.2.1-p10. > Have darn near the latest 5.3 on it now and its running nicely w/o HT. > > Pretty sure ULE was the scheduler being used when I had problems as I > remember when SCHED_4BSD was the new default. IMO, when it worked, ULE > ran smoother. Now it seems that one process can hog the filesystem > resources and block out another doing the exact same thing but to > another file. > > That instance with HT was the first and only time FreeBSD messed up a > filesystem and I've had at least one FreeBSD machine running 24/7 since > 2.0.0-RELEASE. As a result I'm not much in the mood to re-enable HT now > that I have real data on the drives. This is a bit disconcerting. That implies that 5.2 isn't SMP stable. Lou -- Louis LeBlanc FreeBSD@keyslapper.org Fully Funded Hobbyist, KeySlapper Extrordinaire :) http://www.keyslapper.org ԿԬ Weinberg's Principle: An expert is a person who avoids the small errors while sweeping on to the grand fallacy.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20041031003231.GD9405>