Date: Tue, 16 Mar 2004 11:00:30 -0500 From: Sven Willenberger <sven@dmv.com> To: David Malone <dwmalone@maths.tcd.ie> Cc: freebsd-current@freebsd.org Subject: Re: kmem_map too small, revisited Message-ID: <1079452815.23554.69.camel@lanshark.dmv.com> In-Reply-To: <20040316152013.GA82071@walton.maths.tcd.ie> References: <1079446098.23554.49.camel@lanshark.dmv.com> <20040316152013.GA82071@walton.maths.tcd.ie>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, 2004-03-16 at 10:20, David Malone wrote: > On Tue, Mar 16, 2004 at 09:08:18AM -0500, Sven Willenberger wrote: > > Where should I begin to better diagnose what exactly is going on here? > > I can suggest two ways. The first is to record the output of "vmstat > -m" every 5 minutes from when the machine boots until it dies. If > something is consuming all your kernel memory, this may tell us > what subsystem it is. I will get that started on one of the boxes ASAP. > Another possibility is to add the DDB and DDB_UNATTENDED options > to your kernel and record the backtraces when the machine reboots. > Do I need to define dumpdev in rc.conf as well? Also, after the machine reboots after the crash, I can access the kernel dump with DDB by dropping to debug mode? (via the sysctl or keyboard per the handbook and assuming the DDB_UNATTENDED preserves the information upon reboot into multi-user) Sven
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1079452815.23554.69.camel>