etherchannel showing a suspended port

[Michael Schindler]

I have 2 3750's stacked connected with 4 etherchannel trunk lines to a stack of 4 2960's.  Show spanning-tree shows what I would expect.  1 of the etherchannels is forwarding and the other 3 are in the blocking state.  As I dig deeper I have noticed:

All port channels on the 3750's are configured:

Interface port-channel7

Switchport trunk encapsulation dot1q

switchport mode trunk

The ports that are configured as members

Interface GigabitEthernet1/0/7

  Switchport trunk encapsulation dot1q

  switchport mode trunk

  channel-group 7 mode on

Interface GigabitEthernet2/0/7

Switchport trunk encapsulation dot1q

  switchport mode trunk

  channel-group 7 mode on

A very basic configuration that by default would allow all vlans.  2 of the port channels seem to be fine.  On the other 2 the command

show int po7

returns that only Gi1/0/7 is a member not Gi2/0/7

A show int gi2/0/7 switchport shows it as "Suspended Member of Bundle Po7"

A show int gi2/0/7 trunk shows the port status as "trunking" and Vlans allowed on trunk "none"

The port gi2/0/7 is being suspended because of a allowed vlan mismatch.  I cannot find a way to change this, none of the normal things I would do will change this .  The commands

switch(config-if)# switchport trunk allowed vlan all

switch(config-if)# switchport trunk allowed vlan add vlanid

do not yield any results.  The port continues to so none under allowed vlans and thus remains suspended.  (Native vlan is 1 in all trunks)

The ports on the other end of the trunks all look fine, all vlans allowed and no members suspended.

Any help very much appreciated.

Mike Schindler

[Leo Laohoo]

The port gi2/0/7 is being suspended because of a allowed vlan mismatch.

Remove the allowed VLANs from the configuration and put it back in with the correct VLAN numbers.

[Michael Schindler]

Thanks Leo

I have tried a number of things and nothing seems to change the vlans allowed from none.  As for your specific suggestion.  I have not specifically allowed any vlans.  Everything is default (vlans allowed 1 - 4094) .  Are you suggesting the command switchport trunk allowed vlans remove 1 - 4094?  If this is accepted then a switchport trunk allowed vlans all?

Thanks again

Mike

[InayathUlla Sharieff]

Mike,

Here is the deal:-

1- Remove the PO, Default interfaces both the ends.

( conf t# default int x/x)

2- 1st check if the interface connecting between each other are coming up and showing it as trunk? If yes then proceed with Step 3.

3- Shut down the physical interfaces. Create the Port-channel apply the basic config and then apply the po config on the physical interfaces. (Physical interfaces will inherit the config from its PO).

4- Bring the interfaces back online and update us the result.

Regards

Inayath

[Michael Schindler]

Thanks to both of you.

I am going to try Leo's suggestion first and I will post the before and after results of the show int trunk command.

A little extra information.  I am working with Cisco on a couple of things and they had me restart our core switch (the 3750's mentioned above).  When I did this I found that almost all of my etherchannel trunks failed to come up.  One critical one connected our vpn\direct access environment (a single 2960).  This one failed entirely.  I restarted the 2960 and the etherchannel trunk came up and functioned correctly.  I went and restarted all the other access layer switches (another single 2960, and 2 2960 stacks with 4 members each).  After the restart all ether channel links were fully functional.  I left the bank of switches from my original post as it came up so I could trouble shoot it.

[Michael Schindler]

Ok Things are getting weirder.  I mentioned earlier That I have been working with cisco on this as well.  Their suggestion was to use LAcP on the cross stacked etherchannel links.  This weekend I started changing the etherchannel inks to LAcP.  I started with on port on one member of a stack of 2960's.  About 30 seconds after enabling LAcP on the access switch I lost connectivity with ALL switches (I lost access to our entire network).

Needless to say I was not thrilled to have to jump in the car and drive to the office Sunday morning.  When I got there and connected to the console port on the core 3750 stack I quickly realized that all of my ether channel links were down.  To get the network backup I did the same routine as before. Stopped the core switch and then restarted each access switch to reestablish the etherchannel links.

WHY would my core switch drop ALL of it's etherchannel links when 1 etherchannel port on one member of on stack was configured with LAcP?  At worst I would expect the link to go down until I configured the neighboring port on the core switches.  I would not expect all the ports to be affected??

Any thoughts

[devils_advocate]

Have you checked to make sure the spanning tree root bridge for all your vlans is the core switch?

[Michael Schindler]

The spanning tree configuration was a key factor in choosing what port to begin with.  So just before I started I checked to make sure the core was the root.  The etherchannel I was started with was in the blocking state, I wanted to make sure that if I messed up I was affecting the least impactful interface.

[Leo Laohoo]

Ok then, let's do some simple troubleshooting. 

Remove the two physical links from the etherchannel/PO.  And then post the complete output to the command "sh interface trunk" taken from the two switches.

[Michael Schindler]

I may have misread your instructions LEO.  I zeroed in on the word physical and unplugged the cables.  Now rereading your directions I am pretty sure you wanted my to unbundle the physical ports from the etherchannel.  For what it's worth I will attach the results.  I did a show int trunk before I unplugged the cables and the same command after I unplugged them.

When I plugged in the cables, guess what?  Everything came up green.  Now the etherchannel is fully functional with all members of the etherchannel allowing all vlans.!!