Unusual output discards with 3750 running 12.2(50)SE and above. How to troubleshoot and real-life impact?

Unanswered Question
Sep 27th, 2010

Hi,

I've noticed some unusual behaviour with output discards being demonstrated by our 3750 switch stacks that are running IOS 12.2(50)SE and above (IP Services and IP Base). Our edge switch estate is made up of 3560s and 3750 stacks of which the bulk are running 12.2(25)SEEn which reflects the IOS version used at the time of a major refresh of the network a couple of years ago. New sites and switches since added to the network have been configured with a more recent version of that IOS train as the V2 variants of the 3560/3750s come pre-loaded with 12.2(50)SEn.

I've noticed that those 3750s running a 12.2(50)SE variant are reporting a large number of output discards and total output drops on every live FastEthernet port whereas those running 12.2(25)SEEn do not. Unlike 12.2(50)SEn, when you enter the 'sh interfaces counters errors' command on a 3750 running 12.2(25)SEE2 it does not have the 'OutDiscards' column and when you enter the 'sh mls qos interface fastEthernet n/0/n statistics' command 12.2(25)SEE2 does not show the individual output queues statistics but a 'sh int' still shows the total output drops which on a 12.2(25)SEE2 switch is usually zero but are in the tens of thousands on switch stacks running 12.2(50)SEn?

Running the 'sh mls qos interface fastEthernet n/0/n statistics' command on those switch stacks showing the output discards shows that the total output drops value is made up of the individual 'output queues dropped' values and all the drops seem to be from 'queue 2 threshold 3'?

I'm in the process of reviewing our IOS standards and configuration templates with a view to upgrading the edge estate to take advantage of new features such as EnergyWise but I don't want to introduce a potential problem into the network by doing so - hence this post. If our current QoS config (based on the 12.2(25)SEE release) is no longer compatible with later releases I'd like to understand why. Alternatively, it could just be a bug and the switch is erroneously reporting these discards?

I've attached a couple of files - one showing an example of the output 'errors' and another showing how the switches are configured from a QoS perspective.

Any help or advice would be gratefully received.

Thanks in advance

Matthew

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Average Rating: 0 (0 ratings)
crknipe82 Tue, 04/12/2011 - 13:52

Im wondering if you ever found a resolution to this? I am having a similar issue on a 3750 running 12.2(40)se. With my issue it jumps from a large number to 0 and back.

Actions

Login or Register to take actions

This Discussion

Posted September 27, 2010 at 5:28 AM
Stats:
Replies:1 Avg. Rating:
Views:1460 Votes:0
Shares:0

Related Content

Discussions Leaderboard

Rank Username Points
1 15,007
2 8,155
3 7,730
4 7,083
5 6,742
Rank Username Points
140
72
69
65
45