EIGRP memory high

[jawwadparacha]

Hello

Every time there is flap in the network. It increase memory utilization of EIGRP process. Below is the output for your reference. As you may see holding memory is very high on 28 Feb.

PID TTY  Allocated      Freed    Holding    Getbufs   Retbufs Process     

------------------------------------------------------------------------  

SWICORE1 31gen                                                            

564   0  617167404  608829620    8363504          0         0 EIGRP-IPv4  

.                                                                         

SWICORE1 28feb                                                            

564   0 2804776344 2725455364   70067488          0         0 EIGRP-IPv4  

.                                                                         

SWICORE2 31gen                                                            

565   0  527076104  515097500   11999576          0         0 EIGRP-IPv4  

.                                                                         

SWICORE2 28feb                                                            

565   0 2335677852 2253097496   73324688          0         0 EIGRP-IPv4  

How to fix this issue?

Regards

Jawwad

.                                                                         

[smogra]

What platform and IOS you are using?

Do you see any MALLOCFAIL messages in the logs?

[jawwadparacha]

This is Cisco 6509 with s72033-advipservicesk9_wan-mz.122-33.SXI10.bin IOS

I don't see any MALLOCFAIL errors in logs.

[Wilson Bonilla]

Hello.

I've seen many cases where EIGRP triggers high memory utilization, but the root cause is not really EIGRP, for example if there's  there's a layer 1 issue, bad cable or interfaces, interfaces with CRC errors, input errors,

a  spanning-tree loop in the network even high cpu utilization. many times we blame the routing protocol to be the issue but the fact is that there's something else behind EIGRP that triggers, EIGRP to flap, flood the network with queries ... and everything else that you already know.

Could you attach Show tech to the post please?

Regards.

Wilson B.

[jawwadparacha]

Hello

So, I think if found the issue.

In configuration, I found that "eigrp event-log-size 1000000" has been entered. Allowing large number of event to be stored. From Cisco documents, I see that this value by default is 500.

Do you think this might be the issue of memory increase incase of Eigrp state changes.

Regards

M. Jawwad Paracha

[Wilson Bonilla]

Hello.

That's a good catch, try the following

Clear the eigrp events:

Enable

Clear ip eigrp event

and then decrease the buffer size.

...and keep an eye on it, see how it goes after that configuration.

Regards.

Wilson B.

[jawwadparacha]

Hello

I gave Clear ip eigrp event command, although all events were deleted but holding memory still remained the same. What could be the issue?

Regards

M. Jawwad Paracha