port-channel 11 is down (No operational members)

[CiscoBrownBelt]

So I receive port-channel 11 is down (No operational members) when creating a port-channel for server ports on a pair of Nexus switches in a VPC domain. Here are the configs for the ports (same on both switches). Any help guys?

interface Ethernet1/1                                                           

  speed 1000                                                                    

  description Server1                                            

  switchport mode trunk                                                         

  switchport trunk native vlan 2                                             

  switchport trunk allowed vlan 3                                           

  spanning-tree port type edge trunk                                            

  channel-group 2 mode active                                                  

  no shutdown   

                                                                 

interface Ethernet1/2                                                         

  speed 1000                                                                    

  description Server1                                            

  switchport mode trunk                                                         

  switchport trunk native vlan 2                                             

  switchport trunk allowed vlan 3                                           

  spanning-tree port type edge trunk                                            

  channel-group 2 mode active                                                  

  no shutdown   

[marce1000]

 - Did you configure the port channel interface too (po11), (with the same characteristics) as the interface menbers ?

M.

[CiscoBrownBelt]

Yes.

For some reason I had to delete int po2 the go under interfaces, create channel-group 2, and now it came up.

Now my problem as to why I am doing this in the first place, the server connected to these ports/po2 I can only ping one host now not even the other. This pair of switches is joined to a VPC domain 1. Any ideas?

[Reza Sharifi]

Hi,

Since you configure the switch side as trunk, you need to do the same thing on the server side if not communication will not work. For testing and validations, you can change the trunk port to access port and then test to make sure all hosts in that one vlans are reachable.  If this works, you can then change the switch port back to trunk and make sure the server is also configured as trunk.  If the server is going to be in one VLAN only, you don't need trunk at all.

HTH

[CiscoBrownBelt]

Actually I create the ports to join to a port-channel however it stays down, port-channel2 is down (No operational members). Any idea why?

[CiscoBrownBelt]

Actually just got them to come up. I just configured channel-group 2 without the "active" parameter.

[Reza Sharifi]

So, when you create a Portchannel between 2 switches using LACP, make both switches active and

when you create a Portchannel between a switch and a server using LACP, make the switch active and keep the server as passive.

HTH

[CiscoBrownBelt]

Awesome thanks!

I am not sure how servers are configured but I will make sure that is done.

[CiscoBrownBelt]

Thanks bro! Ok so if I have 2 ports on both VPC Peer switches connected to the same server, configure those port as port-channel and vpc X,

e.g.

int 1-2

mode trunk

channel-group 2 mode active

int port-channel 2

vpc 2

then make sure the server ports configuration on the server is configured as passive?

[Reza Sharifi]

That is correct.  In your example, since the switch is trunk the server also has to be configured as trunk.

HTH

[CiscoBrownBelt]

Ok so a problem I am experiencing is the port-channel for the servers goes down if I do "channel-group 2 active". If I just do "channel-group 2 (no lacp show in port-channel summary) it goes down. Any idea what may be wrong on server end?

Also, if the 2 VPC peer switches are configured for Active/Active, should HSRP under the SVIs let's say 2 have the same priority or should SVI 2 on VPC swith1 be 105 and VPC switch 2 be 100?

[CiscoBrownBelt]

So in a topology where 2x 95k Nexus will be Core, and 4x 93k Nexus will be access, is it best to join the 2x 9504 in a VPC domain let's say VPC domain 10, then 1st pair of 93k as VPC Domain11, next pair VPC domain 12.

Just do all SVI on the cores, only 1 SVI for manangement on Access switches, then configure trunks between access and core switches if you only have 1 uplink?