[ previous ] [ next ] [ threads ]
 
 From:  Chris Buechler <cbuechler at gmail dot com>
 To:  m0n0wall at lists dot m0n0 dot ch
 Subject:  Re: [m0n0wall] Spontaneous reboot of 3.18
 Date:  Tue, 10 Nov 2009 18:21:20 -0500
On Tue, Nov 10, 2009 at 1:57 PM, Kimmo Jaskari <kimmo dot jaskari at gmail dot com> wrote:
> My m0n0 spontaneously rebooted not long ago; the only reason I caught
> it was that I happened to sit next to it when it did.
>
> Since I don't forward my logs anywhere at the moment, this is all I
> have. Posting it just in case anyone on the team is interested. The
> events are in reverse order (bottom to top).
>
> Nov 10 19:22:41 kernel: Rebooting...
> Nov 10 19:22:41 kernel: Automatic reboot in 15 seconds - press a key
> on the console to abort
> Nov 10 19:22:41 kernel: Cannot dump. No dump device defined.
> Nov 10 19:22:41 kernel: Uptime: 5d0h25m49s
> Nov 10 19:22:41 kernel: panic: kmem_malloc(4096): kmem_map too small:
> 34947072 total allocated
>

That's strange - you're running out of kernel memory. The only time
I've ever seen that is with a lot of CARP IPs (on pfSense obviously,
not m0n0wall, quite a while ago before we bumped it high enough to
accommodate the max number of CARP IPs).

That's not really helpful though, the only thing that tells is you
likely aren't having a hardware problem.

There are a few others who have reported the same, with much different
hardware.
http://www.google.com/search?q=%22kmem_map+too+small%22+site%3Am0n0.ch

Which leads me to think maybe ipfilter doesn't have the kernel memory
needed for larger state tables, though I'm really just speculating.
Does this happen to correlate with large numbers of active
connections?