cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1700
Views
5
Helpful
4
Replies

ICMP spikes every 50 pings

mbellears2
Level 1
Level 1

I've got a strange situation were every 50 pings, the response times jump to 300-400m/sec for 2-3 pings, then return to normal.

We have a 7206, running at ~10% CPU Utilisation, with ~140Mb RAM free performing 802.1q

The ping fluctuations occur when I ping local servers, client DSL connections and also external addresses.

Testing 5 different connections simultaneously (3 DSL, one local Server, and one external Internet address) reveals the fluctuation occurs at exactly the same time on all connections.

ip-directed broadcasts are disabled on every interface.

Any assistance is greatly appreciated.

Regards,

MB

1 Accepted Solution

Accepted Solutions

tsadmin
Level 1
Level 1

Hello;

We went through this after upgrading our ISP connection to accept full BGP routing tables. The issue is related to periodic updates on the router.

Once we enabled CEF, the problem went away.

See this CCO link for more details:

http://www.cisco.com/en/US/customer/tech/tk365/tk80/technologies_tech_note09186a00801c4f48.shtml

View solution in original post

4 Replies 4

kban-if
Level 1
Level 1

MB, when you say local server does "local" mean same segment or broadcast domain? I would assume broadcast domain, but you never know? Also do you see the same thing when you run the test from different subnets? I'm thinking maybe there is an issue on the local segment/b-cast domain. Can you try the same test from a different subnet?

We have a /20, and the icmp fluctuations occur when pinging any of the addresses within this /20 (And also when pinging external addresses).

It does not occur when pinging devices in the same subnet (Eg. /24)

So pinging from xxx.xxx.x5.1 -> xxx.xxx.x6.1 will produce the fluctuations, but pinging from xxx.xxx.x5.1 -> xxx.xxx.x5.2 will not produce the fluctuations.

Thanks for the reply.

Regards,

MB

tsadmin
Level 1
Level 1

Hello;

We went through this after upgrading our ISP connection to accept full BGP routing tables. The issue is related to periodic updates on the router.

Once we enabled CEF, the problem went away.

See this CCO link for more details:

http://www.cisco.com/en/US/customer/tech/tk365/tk80/technologies_tech_note09186a00801c4f48.shtml

That was the problem! - We had recently started accepting the full BGP routing tables from one of our upstreams.

BGP scanner was causing high-cpu once a minute.

Enabling cef appears to have fixed the issue!

Thanks very much.

Regards,

MB