[c-nsp] PPP Mulitilink on 3620

[Jared Mauch]

On Tue, Feb 22, 2005 at 10:00:32AM -0600, Josh Duffek wrote:
> Does kinda sound like you are pushing too much through it...not sure
> though.  Look at the "sh proc cpu" and also look at the IP INPUT
> process.  Might also look at "sh int | i packets/sec" and see where that
> is...found this post:
> 
> https://puck.nether.net/pipermail/cisco-nsp/2004-November/014214.html
> 
> "3620 can do 2,000pps process switched and 20,000-40,000 in CEF
> depending
> on the alignment of the stars."

	I *strongly* recommend that people graph the cpu of your
routers.  It is highly valuable to track long-term usage of the device.

	here's a sample mrtg cofnig.

Target[Router.proc]: .1.3.6.1.4.1.9.9.109.1.1.1.1.3.1&.1.3.6.1.4.1.9.9.109.1.1
.1.1.4.1:public at routerip_or_name:
Options[Router.proc]: gauge,growright
Title[Router.proc]: Proc Usage
YLegend[Router.proc]: Proc Usage
PageTop[Router.proc]: Proc Usage
MaxBytes[Router.proc]: 100

	There's a few other criteria that I suggest people watch:

interface pps stats (watch for pps related DoS attacks/spikes)
router memory usage (watch for memory leak/fragmentation bugs)
supervisor cpu usage (watch for spikes, esp if doing l2+l3 w/ mls)
routes from peer (snmp pollable in recent IOS and JunOS)
chassis temp sensors (watch the temp of your datacenter/pop/room/basement)

	these will help you do the necessary forensics later when trying
to find out what type of cpu load is 'normal', or when a problem started
to happen.

	- jared

-- 
Jared Mauch  | pgp key available via finger from jared at puck.nether.net
clue++;      | http://puck.nether.net/~jared/  My statements are only mine.