Cisco Router very strange problem - please help

[bojan.vujic]

Hello everybody,

I’m writing here, because I need you help with Cisco Routers. For already few days, I’m trying to solve strange issues regarding to Router performance, and now I don’t have any more Ideas, where ales to look for solution.

This is situation: Outdoor Wireless P2MP professional grade network (non Cisco), with cc 60 users, and about 50 -60 mbps, of traffic. All AP are connected to Cisco 2960 TT, and that Switch is connected on Cisco 2801 with HWIC-4ESW. No special configuration, regular routing between few subnets, few VLAN, maybe 10 rate limit ACL, really nothing special. This scenario works fine, until few days ago when performance issues with Cisco 2801 start to occur. During working time, CPU utilization of router was 50% and more and that was, by my opinion, reason of performance issues. I was thinking that I need router with higher bandwidth because 2801 is over his limit.

After that, I bought Cisco 1941 Sec-K9, and I move HWIC-4ESW from old 2801 to 1941 Sec-K9, do the same configuration, but when I plug 1941 to the network I have very high packet drop between router and switch which is near each other, and which was connected over HWIC-4ESW. Situation was much worse then before, even I found that 1941 have better performance (I look in this document http://www.cisco.com/web/partners/downloads/765/tools/quickreference/routerperformance.pdf). After that I remove HWIC-4ESW, and try with sub interface instead of regular trunking and vlans, but same problem occur. After that I try same with Cisco Routers 2811, and 2911 and same problem occur all the time.

Strange think is that with old router 2801 there is no problem between Router and Switch, and all works fine until utilization of bandwidth start rising. When is not peak time, all works fine. In document above All new routers seems to have higher bandwidth and better performance than old 2801, but still not working..

With new routers there is problem all the time. Old routers 2801 have ISO version 12 and new one IOS version 15.

Please advice because I don’t know what else I can do to solve this issue.

Thanks in advance.

P.S

There is lot of devices, so I didn’t paste any configuration, but if it is necessary I will upload.

[Francesco Pettene]

Hello Bojan

what was keeping the cpu busy? You could investigate on 2801:

show processes cpu sorted

Someone could have started sending small packets lowering overall

throughtput, or multicast packet thats process switched and may lead

to very high cpu usage.

Beside, is there any particular MTU/MSS value set? That maybe

wasn't copied to the new routers provoking the drops?

You could also try a

show interface **/**

where the drops occur on the new routers.

[bojan.vujic]

Hello Francesco,

Thanks for help, tomorrow I will try to setup new router again, and I will monitor current 2801 to see what make CPU busy, and then I will let you to know.

Is it possible that MTU is different on all of this different models of routers?

I check Interfaces when drops start, there is no strange thinks like interface resets or similar errors, all seams fine.

How I can monitor is there small packets which can causes this kind of problems?

[Francesco Pettene]

MTU is a standard value in eth interfaces (1500 bytes), if not otherwise configured.

To monitor what's getting into your router you could setup a SPAN session on you

2960, just to sniff and analyse that traffic with some tools as Wireshark.