Port-channel between dual sup32s on 6500 question.

Unanswered Question
Jul 8th, 2008
User Badges:

We use a 6500 with redundant Sup-32-GE-3B running IOS 12.2(33)SXH2 as our access-layer switch.


There are two uplinks from this 6500 to the core which are L3 port-channels. They are spread between both sup-32s for redundancy.


We are having some strange things happen on the network, print spoolers quit working, DHCP sometimes is flaky, etc, and as I work through the troubleshooting exercise I noticed that the port-channels ports are not configured exactly the same. This may or may not be contributing to the strange problems, but I thought I'd ask.


The difference in the ports on one sup-32 compared to the other sup-32 is that auto-qos was applied to one sup and not the other. PO-1 consists of interfaces 5/1, 5/2, 6/1 and 6/2. (By the way, the WRR configs were put on all the ports on module 5 because 'auto-qos' was configured on port 5/9 because of a VoIP device on that port which is no longer used) Here is the config of one of the port channels:


interface Port-channel1

description BVU RS6506A-CD-MDF

ip address xxx.xxx.xxx.xxx 255.255.255.252

ip flow ingress

ip pim sparse-mode

load-interval 30

mls qos trust dscp


interface GigabitEthernet5/1

description Po1 to RS6509A

no ip address

wrr-queue cos-map 2 1 1 2

wrr-queue cos-map 3 5 3 4

wrr-queue cos-map 3 7 6 7

rcv-queue cos-map 1 1 0

rcv-queue cos-map 1 2 1

rcv-queue cos-map 1 3 2

rcv-queue cos-map 1 4 3

rcv-queue cos-map 2 1 4

rcv-queue cos-map 2 2 5

rcv-queue cos-map 2 3 6

rcv-queue cos-map 2 4 7

mls qos trust dscp

channel-group 1 mode desirable

!

interface GigabitEthernet5/2

description Po1 to RS6509A

no ip address

wrr-queue cos-map 2 1 1 2

wrr-queue cos-map 3 5 3 4

wrr-queue cos-map 3 7 6 7

rcv-queue cos-map 1 1 0

rcv-queue cos-map 1 2 1

rcv-queue cos-map 1 3 2

rcv-queue cos-map 1 4 3

rcv-queue cos-map 2 1 4

rcv-queue cos-map 2 2 5

rcv-queue cos-map 2 3 6

rcv-queue cos-map 2 4 7

mls qos trust dscp

channel-group 1 mode desirable

!

!

interface GigabitEthernet6/1

description Po1 to RS6509A

no ip address

mls qos trust dscp

channel-group 1 mode desirable

!

interface GigabitEthernet6/2

description Po1 to RS6509A

no ip address

mls qos trust dscp

channel-group 1 mode desirable


BVU-SW6509A-AU-W303#sh etherchannel summary


Group Port-channel Protocol Ports

------+-------------+-----------+-----------------------------------------------

1 Po1(RU) PAgP Gi5/1(P) Gi5/2(P) Gi6/1(P)

Gi6/2(P)


Could the fact that the ports on module 5 have different wrr-queue settings than the ports on module 6 cause instability in the port channel that these ports are in?


I plan on removing auto-qos from module 5 to clean things up, but I'm just curious if that could be causing troubles.

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
letsgomets Tue, 07/08/2008 - 10:53
User Badges:

It is best practice to have your port channel interface and all members of the port channel to have identical configurations. Additionally you want to always configure your port channel members via the port channel interface so that your configurations are consistent across all members.


I would try matching the configurations on each member and see how it works from there. What is on the other end of the port channel? It is important to make sure the load balancing method is identical on both sides of the port channel. The defaults between cisco devices are not all the same. Example:


3750 default load balacing is source mac while 6500 is source/destination IP.


Good luck!

Actions

This Discussion