Etherchannel Weidness
I created 2 Etherchannels between a pair of switches. I decided to use LACP on one (Po1) and make the other a static Etherchannel (Po2 - "mode on"). I noticed that the configuration from the constituent ports had been set on Po1 as has always happened in the past, but Po2 had only the basic configuration applied by default to ports (dynamic trunking instead of the static trunking I had configured on the constituent ports). I corrected the configuration on the Port-Channel, but now noticed the Port-Channel and its ports were err-disabled.
Thinking this possibly a result of mismatched configs [due to the previous issue,] I tried cycling the ports (shut, no shut), but immediately received console messages that they were being once again err-disabled for "channel-misconfig". I double checked and the configuration on both sides was identical. I finally disabled ("shutdown") Po1 and then cycled Po2's ports again. Sure enough, they came right up and stayed up. I would expect STP to place one of the Port-Channels in the blocking state, but not err-disable it. This must be some weird interaction between [Static] Etherchannels and STP. One post I found online for someone who ran into the same issue stated that you should just use LACP or PAgP, but that doesn't really address why a correctly configured Static Etherchannel was err-disabled rather than simply being placed in the Blocking state.
[EDIT] Perhaps, the constituent ports being in the STP Blocking state caused the switch to believe that the Etherchannel was misconfigured?
Thinking this possibly a result of mismatched configs [due to the previous issue,] I tried cycling the ports (shut, no shut), but immediately received console messages that they were being once again err-disabled for "channel-misconfig". I double checked and the configuration on both sides was identical. I finally disabled ("shutdown") Po1 and then cycled Po2's ports again. Sure enough, they came right up and stayed up. I would expect STP to place one of the Port-Channels in the blocking state, but not err-disable it. This must be some weird interaction between [Static] Etherchannels and STP. One post I found online for someone who ran into the same issue stated that you should just use LACP or PAgP, but that doesn't really address why a correctly configured Static Etherchannel was err-disabled rather than simply being placed in the Blocking state.
[EDIT] Perhaps, the constituent ports being in the STP Blocking state caused the switch to believe that the Etherchannel was misconfigured?
R&S: CCENT → CCNA → CCNP → CCIE [ ]
Security: CCNA [ ]
Virtualization: VCA-DCV [ ]
Security: CCNA [ ]
Virtualization: VCA-DCV [ ]
Comments
-
Node Man
Member Posts: 668 ■■■□□□□□□□
What model of switches? Is it possible that they only allow 1 etherchannel at a time? -
theodoxa
Member Posts: 1,340 ■■■■□□□□□□
What model of switches? Is it possible that they only allow 1 etherchannel at a time?
3550 .......
[EDIT] I'm going to try it again later with Wireshark and Debugging to see what exactly is happening and in what order.R&S: CCENT → CCNA → CCNP → CCIE [ ]
Security: CCNA [ ]
Virtualization: VCA-DCV [ ] -
Dieg0M
Member Posts: 861
In my experience, switches get a little confused when you change settings on them (as in L2 or L3). The best way to do it to avoid this is to shutdown the port-channel and configure both sides and re-enable the port-channel. I understand that in a production environment this is not always possible but in a lab environment (like the CCIE), this is the fastest way to do it in order to avoid weird issues like this.Follow my CCDE journey at www.routingnull0.com -
Iristheangel
Mod Posts: 4,133 Mod
Agreed with Diego. Shut down the ports, default the interfaces, configure and then bring them up. Some older switches have problems unless you shut the interfaces and default the ports. I had the same problem with some old old 3750's on 12.2 IOS