nanog mailing list archives

RE: Router / Protocol Problem


From: "Justin M. Streiner" <streiner () cluebyfour org>
Date: Wed, 6 Sep 2006 11:19:56 -0400 (EDT)


On Wed, 6 Sep 2006, Mike Walter wrote:

Thanks for everyone's great input.  Here are answers to Justin's
questions.

#1 - 12.3.6a - 7204VXR (NPE400) 512MB - 200+ MB free
#2 - 12.2.15T5 - cisco 7204VXR (NPE225) - 256MB (I have a NPE400 - 512MB
I want to swap in) - 23MB Free (Issue?)

Full Routes from all peers.  No internal routing protocol as of yet, all
static routes.  Getting ready to implement OSPF.  I have not rebooted
the routers as a test.  I have CEF on both routers.  I have had some
customers complaining about slowness.

That NPE-225 could be the culprit. 23 MB free is workable, but definitely low. I see that router is running 12.2.15T5. T train IOS releases are where features are normally introduced first and are often more buggy than mainline IOS releases. I wouldn't be surprised if that release has some memory leaks that is exacerbating your issue. If you graph memory utilization with MRTG or some other handy graphing tool, memory leaks often show up as a pronounced 'saw-tooth' pattern when you look at the utilization over time.

Also, how badly is the CPU getting hit on that NPE-225? Since you're carrying full routes, I wouldn't be surprised to see the BGP Scanner process beating your CPU up every 60-ish seconds. What does the output of "show proc cpu hist" show you? If you have lots of spikes up to 100% peak utilization in the history graphs, then that can definitely be a source of latency.

jms


Current thread: