Debugging Cisco 2621 Routers

I've recently noticed problems with my 2621 router - it becomes unresponsive for about 10 minutes at a time, about 10 times a month. The router sits inbetween the outside world and servers which I routinely log in to - strangely, I've noticed that if I'm logged into my servers during these 10 minute outages, I remain logged in with no slowdown in responsivity. However, a ping from the server to the router hangs.

I'm not sure where to even being debugging the problem - I've enabled logging on the router but haven't noticed any messages which correlate strongly with the appearance of these events. Any suggestions on how to proceed would be greatly appreciated.

Is there a correlation with High bandwidth usage when the router fails? I've seen these older 2621s starting to hang at 25-30Mbit usage (Most of the time using a lot of processes in the CPU, slowly increasing to 99% and deeming the router unable to attend more processes, even pings).

My first suggestion will be to learn to turn on accounting if this is a feature of the router. If this router does not have accounting as one of the features then, turn on snmp monitoring and download any snmp monitoring tool (free) and turn it on, until the next 10 minute period happens. In your snmp tool you will have a good indication on what is happening the minute before and after the issue.

You never mentioned on your message how you solve your issue. Is it by restarting, or it goes away or how else?

Also. Try to troubleshoot if this is due to the external link or internal, this will give you a good starting point.

I hope this helps.

This sounds like it might CPU spike. IOS will prioritise the switching of traffic over your login, so your through traffic may well be fine when your session becomes unresponsive. You can check for CPU spikes using the "show processes cpu history command" which will give up to 72 hours of history. For example, below shows a spike I had a couple of days ago:

...
        11111 1      9 1 1111      1 1111 1 1221  111 1 1 1 1 111  11 1 11
    779900002804797760808320094575709000090605308800193708171813079107090189
100
 90                  *
 80                  *
 70                  *
 60                  *
 50                  *
 40                  *
 30                  *                       *
 20                  *                       **
 10 *********** ************** *********************************************
   0....5....1....1....2....2....3....3....4....4....5....5....6....6....7..
             0    5    0    5    0    5    0    5    0    5    0    5    0
                   CPU% per hour (last 72 hours)
                  * = maximum CPU%   # = average CPU%

If you're logged in when the spike is happening then you can use "show processes cpu sorted 5min" to get an idea of what's eating all the CPU. It might also be worth enabling NetFlow or Flexible NetFlow to check if your router is getting port scanned or something at those times.