7204 and hold-queue

UniversityofArts · ‎09-24-2007

Hi,

I'm having issues on a 7204 VXR router that i'm having problems solving and would be grateful for some advice please.

The router has four FE ports, 1 connects to a tier 1 IP transit provider, 2 connect to internal WAN links to other sites and the 4th interface is trunked to multiple external Internet Exchange Points and a spare WAN link.

Every now and again the router gets affected by i believe something attached to the trunked interface. It causes the router to drop packets to all connections on this interface and causes routing protocol flaps with EIGRP and BGP which in turn ramps up the CPU to 100%.

This happened again today and you can see from the interface the output drops that have occurred:

Input queue: 0/75/2382/1269 (size/max/drops/flushes); Total output drops: 2375.

My understanding is that the input queue of 75 is shared by all vlans on the sub interface so i am now wondering that this input queue is far too low, i'd like to check if i increase the queue size what affect that it might have on the router and other interfaces?

Many thanks

Paddy

Joseph W. Doherty · ‎09-24-2007

I suspect your input queue drops might be caused by the BGP scanner, if you're dealing with Internet routes. If so, had a similar problem on a pair of Internet connected routers, a 7204 and a 7304. Found a Cisco paper that sugested increasing input queue to let CPU catch up. Did increase the inbound hold queue, drops decreased, no adverse effect as far as I can tell.

Didn't have any issue with routing protocol flaps, so don't know whether it would help with that. Perhap adjustment to the scheduler values if changing the hold queue doesn't help.