On Tue, May 4, 2010 at 16:22, naman sharma <naman.prep_at_gmail.com> wrote:
> Hi Group,
>
> I would like to discuss here issue i am seeing in my network. I am seeing
> output drops on our Gig interface while the utilization of the interface is
> just 10%. Would like to know if bursty traffic can cause this issue and is
> there a way to understand and troubleshoot this scenario.
>
> Also would traffic go to WRR queue after processing or before processing in
> 6724 Line card of 6500. I am just not clear if WRR are your hardware queues
> and if yes what is the difference between increasing the queue depth through
> hold-queue or WRR queue-limit command.
Other than checking speed/duplex (which is possible culprit), you are
seeing 10% utilization over which measuring interval - could it be
that microbursts are causing drops? Is some of this traffic EF/COS5
marked - is QoS processing enabled / have you enabled
priority-queueing? Do you have flow-control disabled/enabled - are you
receiving pause frames from the end-node?
Now, to answer your questions...
And yes, bursty traffic can very much cause this, hence my question
about the 10% utilization over which time period. You could be having
interface congestions in extremely short periods, which won't show on
5-minute intervals (sometimes even on 30 second ones).
WRR queue happens at the interface level, so it's the function
performed by the outgoing LC itself.
P.S. No need to open TAC case for trivial things. What else are CCIEs
for, if they can't troubleshoot simple things...
-- Marko Milivojevic - CCIE #18427 Senior Technical Instructor - IPexpert YES! We include 400 hours of REAL rack time with our Blended Learning Solution! Mailto: markom_at_ipexpert.com Telephone: +1.810.326.1444 Fax: +1.810.454.0130 Web: http://www.ipexpert.com/ Blogs and organic groups at http://www.ccie.netReceived on Tue May 04 2010 - 18:13:27 ART
This archive was generated by hypermail 2.2.0 : Tue Jun 01 2010 - 07:09:52 ART