input queue drops

[MARCELO MATURO]

I have an RSM router in a Catalyst 5509 module. There are several vlans and I have a lot of input queue drops in vlans interfaces. I have raised the hold-queue to 200 in each interface, but the problem continues.

Thanks. Regards

[daniel.bowen]

Ok, the two VLAN interfaces that are experiencing the throttles, are they the two most heavily populated (ie devices attached)?

I would suggest upgrading your IOS image to 12.0(8), although that will not resolve the problem as there are no bugs like this in that level of code.

It could be as simple as a Netware server getting very chatty at certain times in the day, possibly a backup being set off or something. I presume you run FastEthernet?

Also, if you have 600 devices on each VLAN, they are slightly over populated and you might want to consider splitting the VLAN into two smaller VLANs

Daniel,

[MARCELO MATURO]

yes, that are the most populated vlans. On the other hand there are people working for leave from Novell. I suppose this fact will improve the perfomance.

We are working too to eliminate transparent bridging between vlans.

Marcelo,

[michaeljmorris]

do a "show buffers". Are you getting failures for "no memory". You may need to tune your buffers :

http://www.cisco.com/warp/public/63/buffertuning.html

[MARCELO MATURO]

The result of show buffers is ok:

Buffer elements:

499 in free list (500 max allowed)

772756847 hits, 0 misses, 0 created

Thanks

[daniel.bowen]

It isn't a buffer problem - if you look at Vlan 5 stats, no packets were dropped due to no buffers

what version of code are you running?

[daniel.bowen]

Also, try to work out if there is some kind of traffic storm in your network occuring at certain times. If so, try to gather some stats from the RSM during this storm

sh proc cpu

sh proc mem

sh buffers

sh int "affected interfaces"

these commands should produce all the relevant info