Re: Output Drops on Gig Interface

Could we see "show int" output for the relevant interface, please?

What kind of LC is this? What is the fabric utilization? What is the
fabric switching mode?

--
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/
On Tue, May 4, 2010 at 19:09, naman sharma <naman.prep_at_gmail.com> wrote:
> Thanks all for your replies. Well it is 1 Gig and full duplex on both the
> side and it is not hardcoded. Flow control is off on both the sides for
> input and output traffic.
>
> So these 2 routers are in MPLs domain with one being PE and the other being
> P router and i see output drops on the PE router towards P router. PE router
> has mls qos enabled and right now the interface in the MPLS domain shows all
> the traffic in cos 0 and hence in Queue 1 and there is where i see the
> drops.
>
> Interface GigabitEthernet7/1 queueing strategy:B  Weighted Round-Robin
> B  Port QoS is enabled
> Trust boundary disabled
>
> B  Trust state: trust COS
> B  Extend trust state: not trusted [COS = 0]
> B  Default COS is 0
> B B B  Queueing Mode In Tx direction: mode-cos
> B B B  Transmit queues [type = 1p3q8t]:
> B B B  Queue IdB B B  SchedulingB  Num of thresholds
> B B B  -----------------------------------------
> B B B B B B  01B B B B B B B B  WRRB B B B B B B B B B B B B B B B  08
> B B B B B B  02B B B B B B B B  WRRB B B B B B B B B B B B B B B B  08
> B B B B B B  03B B B B B B B B  WRRB B B B B B B B B B B B B B B B  08
> B B B B B B  04B B B B B B B B  PriorityB B B B B B B B B B B  01
>
> B B B  WRR bandwidth ratios:B  100[queue 1] 150[queue 2] 200[queue 3]
> B B B  queue-limit ratios:B B B B  50[queue 1]B  20[queue 2]B  15[queue 3]B  15[Pri
> Queue]
>
> B B B  queue tail-drop-thresholds
> B B B  --------------------------
> B B B  1B B B B  70[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8]
> B B B  2B B B B  70[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8]
> B B B  3B B B B  100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8]
>
> B B B  queue random-detect-min-thresholds
> B B B  ----------------------------------
> B B B B B  1B B B  40[1] 70[2] 70[3] 70[4] 70[5] 70[6] 70[7] 70[8]
> B B B B B  2B B B  40[1] 70[2] 70[3] 70[4] 70[5] 70[6] 70[7] 70[8]
> B B B B B  3B B B  70[1] 70[2] 70[3] 70[4] 70[5] 70[6] 70[7] 70[8]
>
> B B B  queue random-detect-max-thresholds
> B B B  ----------------------------------
> B B B B B  1B B B  70[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8]
> B B B B B  2B B B  70[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8]
> B B B B B  3B B B  100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8]
>
> B B B  WRED disabled queues:
>
> B B B  queue thresh cos-map
> B B B  ---------------------------------------
> B B B  1B B B B  1B B B B B  0
> B B B  1B B B B  2B B B B B  1
> B B B  1B B B B  3
> B B B  1B B B B  4
> B B B  1B B B B  5
> B B B  1B B B B  6
> B B B  1B B B B  7
> B B B  1B B B B  8
> B B B  2B B B B  1B B B B B  2
> B B B  2B B B B  2B B B B B  3 4
> B B B  2B B B B  3
> B B B  2B B B B  4
> B B B  2B B B B  5
> B B B  2B B B B  6
> B B B  2B B B B  7
> B B B  2B B B B  8
> B B B  3B B B B  1B B B B B  6 7
> B B B  3B B B B  2
> B B B  3B B B B  3
> B B B  3B B B B  4
> B B B  3B B B B  5
> B B B  3B B B B  6
> B B B  3B B B B  7
> B B B  3B B B B  8
> B B B  4B B B B  1B B B B B  5
>
> B B B  Queueing Mode In Rx direction: mode-cos
> B B B  Receive queues [type = 1q8t]:
> B B B  Queue IdB B B  SchedulingB  Num of thresholds
> B B B  -----------------------------------------
> B B B B B B  01B B B B B B B B  WRRB B B B B B B B B B B B B B B B  08
>
> B B B  WRR bandwidth ratios:B  100[queue 1]
> B B B  queue-limit ratios:B B B  100[queue 1]
>
> B B B  queue tail-drop-thresholds
> B B B  --------------------------
> B B B  1B B B B  100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8]
>
> B B B  queue thresh cos-map
> B B B  ---------------------------------------
> B B B  1B B B B  1B B B B B  0 1 2 3 4 5 6 7
> B B B  1B B B B  2
> B B B  1B B B B  3
> B B B  1B B B B  4
> B B B  1B B B B  5
> B B B  1B B B B  6
> B B B  1B B B B  7
> B B B  1B B B B  8
>
>
> B  Packets dropped on Transmit:
> B B B  BPDU packets:B  0
>
> B B B  queueB B B B B B B B B B B B B  droppedB  [cos-map]
> B B B  ---------------------------------------------
>
> B B B  1B B B B B B B B B B B B B B B B B B  295660B  [0 1 ]
> B B B  2B B B B B B B B B B B B B B B B B B B B B B B  0B  [2 3 4 ]
> B B B  3B B B B B B B B B B B B B B B B B B B B B B B  0B  [6 7 ]
> B B B  4B B B B B B B B B B B B B B B B B B B B B B B  0B  [5 ]
>
> B  Packets dropped on Receive:
> B B B  BPDU packets:B  0
>
> B B B  queueB B B B B B B B B B B B B  droppedB  [cos-map]
> B B B  ---------------------------------------------
> B B B  1B B B B B B B B B B B B B B B B B B B B B B B  0B  [0 1 2 3 4 5 6 7 ]
>
> Now i can increase the queue limit but that will add delay to the packets
> sitting in the queue and can lead to other issues. Pls suggest.
>
> thanks
> naman
>
> On 4 May 2010 11:36, Marko Milivojevic <markom_at_ipexpert.com> wrote:
>>
>> You are absolutely right... If it's indeed GigE speed we're talking
>> about here. However, we only have the information that interface
>> itself is GigE, but as we know, we have those "10/100/1000" interfaces
>> - they are prone to this kind of thing.
>>
>> If it's GigE speed on the link, then I would personally look at QoS
>> and especially flow-control, as personally I had quite a few issues
>> with it and Cisco swouters.
>>
>> --
>> 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/
>>
>> On Tue, May 4, 2010 at 18:32, Ryan West <rwest_at_zyedge.com> wrote:
>> > Hey Marko,
>> >
>> >> -----Original Message-----
>> >> Sent: Tuesday, May 04, 2010 2:16 PM
>> >> To: Narbik Kocharians
>> >> Cc: itguy.pro_at_gmail.com; Kambiz Agahian; naman sharma; Cisco
>> >> certification
>> >> Subject: Re: Output Drops on Gig Interface
>> >>
>> >> On Tue, May 4, 2010 at 17:39, Narbik Kocharians <narbikk_at_gmail.com>
>> >> wrote:
>> >> > That is true, the end that is in half Duplex mode should get "late
>> >> > collisions" and the end that is in full duplex mode should get "CRC
>> >> > checks",
>> >> > whereas, a mismatch in Speed (Which i don't think that could be the
>> >> > problem
>> >> > that you are experiencing) should show as "NOTCONNECTED".
>> >>
>> >> Quite right, however, if duplex is not hardcoded, but speed is, it
>> >> would not be negotiated in most cases. Cisco used to default to
>> >> half-duplex in this case. I've seen quite a few issues caused by
>> >> configuring only parts of the speed/duplex pair.
>> >>
>> >> If any of them is set manually, negotiation is disabled. To negotiate
>> >> speed and duplex, both need to be set to auto.
>> >>
>> >
>> > It was my understanding that by default, all devices are supposed to
>> > perform autonegotiation as 802.3z does not specifically define a way to turn
>> > it off. B Also, Cisco devices do not support half-duplex Gig and the standard
>> > does not have support for it either. B With link negotiation turned off, the
>> > device with autonegotiation turned off will report up and the other side
>> > will be down.
>> >
>> > I have not tested all of these scenario's in great detail, so in
>> > practice it might differ slightly.
>> >
>> > -ryan
Blogs and organic groups at http://www.ccie.net