On Jul 05 13:51:06, Otto Moerbeek wrote:
> On Wed, Jul 04, 2012 at 09:00:27PM +0200, Jan Stary wrote:
>
> > On Jul 04 10:57:45, Otto Moerbeek wrote:
> > > On Tue, Jul 03, 2012 at 07:06:37PM +0200, Jan Stary wrote:
> > >
> > > > > this is another problem. configure
> > > > > swap to be double the RAM size, or if you insist on loading the
> > > > > machine about 4 times the RAM size. you obviously forgot to or didn't
> > > > > want to configure swap.
> > > >
> > > > is the "twice the RAM" mantra still valid today?
> > >
> > > No, that only makes sense for relatively small systems. The disklabel
> > > auto allocation policy uses a max size if 2x physmem for physmem <
> > > 256M. Otherwise it allocates a bit more than physical mem size.
> >
> > Well, this machine has 128M - does that mean that having 256M of swap
> > would make my system generally better?
>
> Depends how much memory is actually used by the processes on it. If
> you never will hit swap having it or not should not make a difference.
>
I unloaded some of the work from the machine a few days ago,
leaving just pf, named, dhcpd and postfix. I am pretty sure
I am not hitting the (nonexistent) swap now. The segfaults
have not appeared since.
On Jul 05 14:50:02, Philip Guenther wrote:
> On Tue, Jul 3, 2012 at 8:16 AM, Jan Stary <h...@stare.cz> wrote:
> > This is 5.1-beta/i386 on an ALIX about five years old,
> > running as my home server.
> >
> > Recently, processes started to die for reasons unknown, as in
> >
> > pid 20260 (postgres): user write of 118784@0x28052000 at 159088 failed: 14
> > pid 1872 (cron): user write of 118784@0x2b1e3000 at 30224 failed: 14
> >
> > 14 is EFAULT as per sys/errno,h
> > - what can be causing it?
>
> Having briefly looked at this, I suspect there are (at least) two
> cases where that can occur:
> 1) a race in the code that prevents writing to a file that is mapped
> for execution, and
> 2) system runs out of memory when trying to do copy-on-write; the
> process is killed because of
> that and then the coredump logic hits the missing page when trying
> to write out the process's
> memory and logs that failure.
>
> I think (1) is what sthen@ described; I think you're hitting (2).
Yes, I believe that was the case.
Thanks to all for the insight.
Jan
