Date: Mon, 3 May 2004 08:19:43 PDT From: John Uhlig <juhlig@parc.com> To: Stefan Bethke <stb@lassitu.de> Cc: Kris Kennaway <kris@obsecurity.org> Subject: Re: kmem_malloc crashes running FreeBSD 5.2.1-RELEASE-p5 Message-ID: <4096630F.9080500@parc.com> In-Reply-To: <B6D33856-9BBE-11D8-9E08-000A95C893E4@lassitu.de> References: <409024D6.7090800@parc.com> <20040429000158.GA61693@xor.obsecurity.org> <B6D33856-9BBE-11D8-9E08-000A95C893E4@lassitu.de>
next in thread | previous in thread | raw e-mail | index | archive | help
Well, Don Bowman was very helpful in suggesting some kernel parameters. Unfortunately, they did not work for us. If we had the time to experiment with more settings, perhaps we would have been able to find the right combination. Since we do not have that time due to our schedule I was forced to "fix" the problem by setting "hw.physmem=2G" in /boot/loader.conf (thanks again to Don). It ran for over 8 hours while I ran multiple scripts copying data from multiple systems to it and while continuously running the /etc/periodic/security scripts on it. All previous tests crashed the server (usually within 20 minutes). I am hoping that this issue will be resolved by the developers in the future as we would like to be able to use the full complement of available memory (4GB). thanks to all who have responded to my inquiry. John Uhlig. Stefan Bethke wrote: > Am 29.04.2004 um 02:01 schrieb Kris Kennaway: > >> On Wed, Apr 28, 2004 at 02:40:38PM -0700, John Uhlig wrote: >> >>> On the web, freebsd mailing lists and bug lists, I have seen >>> existing thread about "kmem_malloc and kmem_map too small" problems >>> - but no evidence of a fix or patch. We experienced the same >>> problem running 4.9 but were able to fix it by setting the MAXMEM >>> option in our kernel conf file to a value 1GB less than actual >>> physical memory size. This does not help with 5.2.1-RELEASE-p5. >>> >> See PR 53416. You just need to tune your kernel resource allocation >> to deal with the large amount of memory your system has. > > > I recently enquired about advice for this exact situation on -current > (see http://lists.freebsd.org/pipermail/freebsd-current/2004-April/ > 026065.html), but didn't really get any. > > The PR does not give advice on which parameters in which way, or list > a solution. My own experimentation has been inconclusive; the > machine still panics when running the daily scripts. I'm running a > cron job every minute to record various vm statistics now; hopefully > this will give me further pointers. > > How does PAE affect this? My box has 6 GB, but I have disabled PAE > for the moment. > > And just for clarification: the panic is with basically no load, just > some file system pressure (two jails running the daily scripts); > otherwise, the machine is unused. > > > Stefan >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4096630F.9080500>