[f-nsp] Packet loss to ve on MLX

Brad Fleming bdflemin at gmail.com
Fri Jun 29 12:12:24 EDT 2012


On Jun 13, 2012, at 12:08 PM, Brian Taylor wrote:

> We have an MLX that has two devices that send a ping every second to their
> gateway as an heartbeat.
> Both of these are directly plugged in to copper Gigabit ports on separate
> ni-mlx-20-port-1g-copper modules.
> They are untagged into the same VLAN with a router interface on the same
> MLX.
> There are times that the VE delays ping responses by up to 5 seconds.
> Overall load on this chassis is under 10%.
> 
> Has anyone seen any ICMP issues with a VE on an MLX or anything else that
> might explain this?


Is the 10% load on the management module or the line card(s)? What does a "show cpu lp" give you? What about a "show cpu ave all 5"?

We have about 17 XMR4000s with 20port 1G SFP cards and don't notice any problems with ICMP toward VE interfaces regardless of number or slot location of ports. We're still running 5.1 IronWare.

-brad



More information about the foundry-nsp mailing list