Troubleshooting Etherchannel

This topic is to discuss the following lesson:

Hi Rene,

Can you point to any material on L2CP? It would be great if you have any documents on the same?

Regards,
Ananth

Hello Ananth Maruti

Currently there is no post concerning L2CP. However, until Rene gets something up and running concerning L2CP, you can take a look at this:

This document desscribes L2CP configuration for the ASR 901.

I hope this has been helpful!

Laz

Hi,

I had some questions concerning the port-channel load-balancing based on dst-ip | dst-mac | src-dst-ip | src-dst-mac | src-ip | src-mac. I was wondering if you could explain that a little further for me.

also have question on clear error on the port channel and things of that nature.

thanks again for a great topic.

Hello Christopher

The port channel load balancing feature that you are referring to is the port-channel load-balance method command. This command sets the load distribution method that will be used to distribute traffic among the ports in the bundle.

The available methods are the following:

* dst-ip—Load distribution on the destination IP address
* dst-mac—Load distribution on the destination MAC address
* dst-port—Load distribution on the destination TCP/UDP port
* src-dst-ip—Load distribution on the source XOR destination IP address
* src-dst-mac—Load distribution on the source XOR destination MAC address
* src-dst-port—Load distribution on the source XOR destination TCP/UDP port
* src-ip—Load distribution on the source IP address
* src-mac—Load distribution on the source MAC address
* src-port—Load distribution on the source port

Essentially each of the above methods makes the specific parameter go over the same link every time. That is, if there is traffic to many different destination IP addresses going over the bundle, packets with the same destination IP address will always use the same link in the bundle. These must be used with caution because if, for example, you use the dst-mac method and all traffic going over the EtherChannel bundle has the same destination MAC, then ALL traffic will go over only one physical link of the bundle, drastically reducing the efficiency of the bundle! This can easily be the case if you have a default router on the other end of the EtherChannel. Make sure you have enough variety of IPs/MAC/Ports on each end to justify your load balancing method.

You can find more information about load-balancing over EtherChannel from Cisco here and more about this specific command at Cisco’s command reference.

I hope this has been helpful!

Laz

thank you for the info!!

I didn’t find the command to clear any port channel error though. Any ideas ?

Hello Christopher.

Can you be more specific? Where are the portchannel errors indicated?

Thanks

Laz

Hi @chriscowboyfann,

You can clear the Port-channel interface counters, which will include clearing the error counters, by using the following command:

CiscoCat2970#clear counters Port-channel 1 
Clear "show interface" counters on this interface [confirm]y

Verify with:
CiscoCat2970#show interface Port-Channel 1 counters errors

Port        Align-Err     FCS-Err    Xmit-Err     Rcv-Err  UnderSize  OutDiscards 
Po1                 0           0           0           0          0            0 

Port      Single-Col  Multi-Col   Late-Col  Excess-Col  Carri-Sen      Runts     Giants 
Po1                0          0          0           0          0          0          0 

If you haven’t already, you may wish to read the thread “etherchannel-on-cisco-ios-catalyst-switch” here

Kind regards,
Jon

I figured it out thank you !!!

1 Like

thank you :slight_smile: I appreciate the help

1 Like

Hello Rene,

I am having an issue with LACP etherchannel (2 interfaces in switch stack of 2 switches) connected to a UCS server, the ports as below:

- int Gi1/0/1 is up up LACP mode Active
- int Gi2/0/1 is up up LACP mode Active
- once I configured the port channel 1, the interface Gi2/0/1 goes up/down status
- PO 1 is down down
- Gi1/0/1 is up/up
- All my config right, no err-disabled no errors in switch log, both interfaces speed & duples are the same.
- 1 Po1 (SU) LACP Gi1/0/1(I) Gi2/0/1(w) when using etherchannel summary output

any suggestion what to look for? thanks in advance.

Hello Wisam

There are two issues going on here at the same time. One is configuring a cross-stack etherchannel configuration and the other is configuring a UCS server to function correctly as an etherchannel endpoint.

Concerning the cross-stack etherchannel configuration, make sure you are using the minimum IOS necessary to support it. Also, make sure the platform supports it as well. This document describes the requirements for cross stack etherchannel on the 3750 series switches for your reference.

Secondly, make sure the UCS server is configured correctly for its NIC cards to create the etherchannel. More information about how to configure this can be found at this Cisco document.

Finally, looking at the output of the etherchannel summary command, you can see that Gi1/0/1 goes into stand-alone mode (I) and the Gi2/0/1 interface goes into waiting to be aggregated (w) mode. In most cases, this has to do with the config on the server side of the link. Often it is a misconfiguration that makes the second NIC card not be recognised, and thus it puts the first link into stand-alone mode.

Before doing anything else, if you have a spare switch on hand, or even spare ports on a production switch (careful how you do this!) you can attempt to create an etherchannel with these two links and see if it connects. If it does, you config on the switch side is OK and the issue is with the NIC cards.

Keep us posted on your results! I hope this has been helpful!

Laz

Hi , I have issue here … if there is Etherchannel between 2 Switch the bandl 2 cable … and working fine … no I will send Continuous Ping through it … during that Ping I will shutdown one link in the Etherchannel … that link included high port priority number which is all traffic using that … the Ping will be Time it out !! The question why another link not co we that ?

Hi @Dhari.kahled,

Thanks for your question. When you administratively shutdown one link, does the PortChannel still remain operationally Up? If a switch detects an EtherChannel bundle has less than the specified minimum number of links Up, it will go Down automatically and your ICMP echo (ping) traffic will be lost.

Kind regards,
Jon