[c-nsp] CEF fun in SXF
Gert Doering
gert at greenie.muc.de
Tue Dec 13 02:58:51 EST 2005
Hi,
On Mon, Dec 12, 2005 at 03:16:25PM -0800, Roman Sokolov wrote:
> Monday, December 12, 2005, 12:42:07 PM, you wrote:
> GD> ... is there a way to influence the PPS rate for this rate limiting?
> Yes. sh mls rate-limit. router#mls rate-limit ?
Hmmm, I'm not sure that applies. This seems to be just "special" stuff,
not "generic unicast packets travelling through the box".
> GD> We recently ran into another interesting effect, where an out-of-memory
> GD> problem triggered "%FIB-2-FIBDOWN: CEF has been disabled due to a low
> GD> memory condition." -- the box wasn't very busy anyway, so the MSFC should
> GD> have managed to keep up. Instead, we saw packet loss like crazy, which
> GD> fits your description.
> If you have disabled cef because of memory issues, then may be you have to
> use same solution like for gsr's cef :) i mean, exception memory minimum
> processor xxx. Fast reboot sometimes can help much, and of course you have
> to monitor number of routes in your routing table.
Well, the main problem was a quite unexpected memory leak in the BGP
process. The box has more than sufficient memory for the number of routes
it's carrying (only 170k "Internet" routes, and some 200 VRF routes),
but as BGP was leaking like hell, we got caught by surprise.
Fast reboot isn't available on Sup720 - the box will always do a
very-slow-reboot... especially with the huge SXE and SXF images :(
> GD> So it would have been nice to say "yes, we know the box will keep up
> GD> with the current load, and tonight we will reboot it" instead of having
> GD> to reboot at office hours...
> It's not ideal world :)
Sure, but that doesn't keep my from trying to improve things :)
gert
--
USENET is *not* the non-clickable part of WWW!
//www.muc.de/~gert/
Gert Doering - Munich, Germany gert at greenie.muc.de
fax: +49-89-35655025 gert at net.informatik.tu-muenchen.de
More information about the cisco-nsp
mailing list