[c-nsp] Mysterious 7513 reboot

[Rodney Dunn]

Jon,

I'll bet that 12.2S code doesn't have the memory enhancements
for BGP convergence that went in to 12.0S. So if your peers
flapped you may have used a LOT of transient memory and when
it got low it could never recover. That may show up as fragmentation
but it's a bit different. You can usually tell if you look at 'sh mem stat'
and see if the "lowest" column is pretty low.

Rodney

On Wed, Apr 26, 2006 at 11:37:44AM -0400, Jon Lewis wrote:
> On Wed, 26 Apr 2006, Chris Cappuccio wrote:
> 
> > You're running out of memory, you should move back to 12.0 if you aren't already
> > there
> 
> He's probably not really out of memory, but has badly fragmented memory. 
> We just had a similar event yesterday caused by a VIP gone bad.  Due to 
> [insert IOS bug speculation], after a VIP crashed several times, our RSP 
> memory was so badly fragmented that BGP sessions couldn't stay up (we'd 
> run out of memory and they'd get reset) and the router wouldn't allow me 
> to do a redundancy force-switchover.  Under "normal" conditions, this 
> router has 96MB free (running 12.2S).
> 
> 
> ----------------------------------------------------------------------
>   Jon Lewis                   |  I route
>   Senior Network Engineer     |  therefore you are
>   Atlantic Net                |
> _________ http://www.lewis.org/~jlewis/pgp for PGP public key_________
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/