Etherchannel config not working

whiteford · ‎03-10-2009

Hi,

I am trying to create an etherchannel between a Cisco 2950 and 3550 in a lab first. I want to use 2 ports on each switch as a trunk for load balance and redundancy.

If I just have one fa0/1 from each switch connected the PC's in fa 0/24 on each switch (VLAN 10) can ping each other as soon as I add the second trunk from fa0/2 on each switch the pings stop and I get the spanning tree blocking one of the ports:

01:11:14: %SPANTREE-7-RECV_1Q_NON_TRUNK: Received 802.1Q BPDU on non trunk FastE thernet0/3 VLAN1.

01:11:14: %SPANTREE-7-BLOCK_PORT_TYPE: Blocking FastEthernet0/3 on VLAN0001. Inconsistent port type.

Please take a look at my simple configs

whiteford · ‎03-10-2009

First mistake:

Received 802.1Q BPDU on non trunk FastE thernet0/3 VLAN1.

I used the wrong port, however does my config look ok for load balance?

As I'm getting this error now:

01:30:08: %PM-4-ERR_DISABLE: channel-misconfig error detected on Fa0/2, putting Fa0/2 in err-disable state

jorge.calvo · ‎03-10-2009

Hello,

Regarding the misconfiguration you have on the channel-group check:

1. You are using the same trunking protocol on both sides, dot1q in this case.

2. The flow-control is the same on both sides. Flowcontrol is off by default, thus the line should not be there.

interface Port-channel1

switchport mode trunk

flowcontrol send off <---

3. Enable trunk only on the Port-channel.

About the load-balance, you left it by default which is source mac-address. The load balancing will depend on the kind of flow that will traverse your Port-channel.

By the way, did you get your ping to work?

glen.grant · ‎03-10-2009

You can't bring them up one at a time . Also my self I would use " channel group X mode desirable non-silent but that is just a preference which lets the switch negotiate the etherchannel. Do a shut on the port channel itself not the interfaces . Then bring up the "port channel" (no shut) which should bring up both interfaces at the same time. Do this for both sides . When you do 1 at a time it treats it as an individual trunk link and then you have a mismatch from the far side of the link and a spanning tree loop so it shuts down 1 port , when you bring the "port channel active " it puts both ports into the port channel and should eliminate the spanning tree blocked port condition .

whiteford · ‎03-10-2009

Shutting down the port channel and bring it back up fixed it and all is working, I can ping other the devices across the switches even when I pull out the a trunk port or shut it down, so by default is it load balancing?

Can I tell which of the trunks is getting used most?