[f-nsp] Multicast causing high CPU

[Christian MacNevin]

Thanks a lot, it sounds like it's pretty cut and dried :)

I'm pretty sure they wouldn't want to do hardware flooding. That's  
basically putting the switch in dense mode and asking for all sorts of  
trouble unless I'm
mistaking how you mean that.

Cisco has info in troubleshooting CPU on everything. Does Foundry's  
documentation not cover it? I must admit i found it difficult to find  
anything quickly with
Google..



On Aug 19, 2008, at 8:26 PM, Niels Bakker wrote:

> * christian.macnevin at gmail.com (Christian MacNevin) [Tue 19 Aug  
> 2008, 19:16 CEST]:
>> I've got a customer who's seeing very high CPU utilisation whenever  
>> multicast is turned on across his BigIrons. I'm sorry I can't be  
>> more specific than this, it's just about all the info I have.
>
> Assuming that with BigIron you mean IronCore or JetCore - these  
> forward multicast over the CPU by default.  On Jetcore you can  
> enable hardware flooding of broadcast/multicast on a per-VLAN basis  
> or in later software releases with the cpupro command based on  
> recent average CPU load.
>
>
>> They also have cisco 3750s and 65s which seem to be doing fine.  
>> Before I can get a chance to look at the topology, I'm thinking  
>> about maybe rpf checking owing to a split l2/l3 architecture, or  
>> possibly just the RP functionality being too much for it  
>> (decapsulating PIM registers).
>
> That's definitely going over the CPU, I'm not sure that h/w flooding  
> will even help you...
>
>
>> Are there many multicsat users on this list? And for that matter,  
>> are there any good high cpu utilization troubleshooting guides out  
>> there for foundry?
>
> For a six to ten year old platform?
>
>
> 	-- Niels.
>
> -- 
> _______________________________________________
> foundry-nsp mailing list
> foundry-nsp at puck.nether.net
> http://puck.nether.net/mailman/listinfo/foundry-nsp