MAC-Miss Rate on ACE module

petersonb · ‎05-09-2008

What exactly does the MAC-Miss rate mean on the ACE? And if we are running out of resources for it, should I worry?

We have only implemented 1 production policy on the ACE module so far and we are already running out of resources for the mac-miss rate. All other resources look good.

Is this OK? Or is something wrong here?

Attached is the resource usage counters.

Thanks,

Ben

Gilles Dufour · ‎05-09-2008

When the ACE receives traffic for which it does not have an arp entry for either the source or destination, this is called a mac-miss and the fastpath agent needs to ask the slowpath agent to perform an arp request.

This communication is rate-limited.

With no mac entry for a src or dst, we drop the packet.

So, you should increase the resource.

Or review your design.

It's best to have the clients coming through a gateway (ie: the MSFC) instead of directly accessing the ACE.

This way only 1 mac entry is needed - the gateway.

You'll see a counter like this

switch/Admin# sho np 1 me-stats "-socm -v" | i mac

Drop [mac lookup fail]: 4 0

Gilles.

petersonb · ‎07-14-2008

Thanks for the info. we do have the routed design where all traffic into the front-end VIP is routed into that VLAN.

I tried to command you mentioned and here is what I got:

MSNGO-ACE-Admin-1/MSNGO-Infra-Services# sho np 1 me-stats "-socm -v" | i mac

Drop [mac lookup fail]: 0 0

If I type look at the MAC-Miss resources here is that counter:

MSNGO-ACE-Admin-1/MSNGO-Infra-Services# sh resource usage | i mac

mac-miss rate 214 1280 300 1100 834764

It just seems out mac-miss is rather high for some reason. The context is allowed to use 100% of the resources for the mac-miss. I am just not sure where the issue could be here.

Ben