Comware

 View Only
  • 1.  Output queue

    Posted Jun 21, 2012 09:42 PM

    Is there anyway to view packet counters associated with the port output queues on the a-series? (I believe there are 8 numbered 0-7)

     

    Maybe I missed it in the manual...



  • 2.  RE: Output queue

    Posted Sep 04, 2012 06:07 AM

    I was wondering the exact same thing. The documentation, in my case for the A3600-SI, is not very clear about the different prioritization methods, and how they affect queues. So I would like to be able to see how packets are routed to the 8 queues, depending on my settings.

    Have you made any progress?



  • 3.  RE: Output queue

    Posted Sep 16, 2012 11:52 PM

    Unforunately I have not made any progress on queue counters. But I can get counters on my traffic behaviors for example:

     

    traffic behavior MARK_AF41
     remark dscp af41
     remark local-precedence 4
     remark drop-precedence 0
     accounting packet

     

    [switch]display qos policy interface GigabitEthernet 1/0/0/45

      Interface: GigabitEthernet1/0/0/45

      Direction: Inbound

      Policy: QOS_POLICY
       Classifier: CITRIX
         Operator: AND
         Rule(s) : If-match acl name CITRIX
         Behavior: MARK_AF41
          Marking:
            Remark DSCP af41
          Marking:
            Remark local precedence 4
          Marking:
            Remark drop precedence 0
          Accounting Enable:
            409833511 (Packets)

    I will offer up some of my learnings with the output queues. One not so well documented fact (at least on my 10508) is that output queues are only filled up using a local-precedence values (0-7). For example if I mark inbound traffic as af41 the switch will not change the output queue. What I need to do is mark my traffic inbound as af41 and also local-precedence 4 and drop-precendece 0 (seen in the config above). This will place the traffic in output queue 4 as it leaves an interface. I can then mess with the wrr or wfq settings on the outbound port to influence the flow of traffic.

     

    Now if the traffic is already marked as af41 BEFORE it gets to my switch the mapping to local-precedence 4 and drop precedence 0 will happen automatically as long as I trust dscp, and havent altered the default map tables.

     

    Hope that helps...

     

    Thanks,

     

    Dan