[c-nsp] FW: Cisco blade switch config

[Kevin Berry]

Etherchannel now shows only the two uplink ports, as below:

Group  Port-channel  Protocol    Ports
> ------+-------------+-----------+-------------------------------------
> ------+-------------+-----------+------
> ----
> 1      Po1(SU)         LACP      Gi0/20(P)   Gi0/22(P)
>
>You original show etherchannel had too many ports in the portchannel. Have
you fixed this yet?

-----Original Message-----
From: Andrew Miehs [mailto:andrew at 2sheds.de] 
Sent: Wednesday, February 20, 2013 5:03 PM
To: Kevin Berry
Cc: <cisco-nsp at puck.nether.net>
Subject: Re: [c-nsp] FW: Cisco blade switch config



Sent from a mobile device

On 21/02/2013, at 8:11, "Kevin Berry" <kevin.berry.70 at gmail.com> wrote:

> Yes, I am aware. My HP simply was configured something like:
> 
> Config t
> Trunk 21-22 trk1 lacp
> Trunk 37-38 trk2 lacp
> Exit
> 
> Then I made sure that those trunks were TAGGED for each VLAN.
> 
> The HP switch, appears correct. I believe, even by review Then for 
> Cisco, I built Etherchannel LACP from ports. My etherchannel summary 
> now shows that gi 0/20 & 0/22 are in use and bundled in port channel. 
> So that seems fine.


> My etherchannel for those ports was built for channel 1, or PO1. 
> Config
> reads:
>> interface Port-channel1
>> description HP CORE
>> switchport trunk allowed vlan 1-3,100 switchport mode trunk 
>> spanning-tree portfast trunk spanning-tree vlan 4-99 cost 20 !
>> interface GigabitEthernet0/1
>> switchport trunk allowed vlan 1-3,100 switchport mode trunk speed 
>> 1000 channel-group 1 mode active spanning-tree portfast !
>> interface GigabitEthernet0/2
>> switchport trunk allowed vlan 1-3,100 switchport mode trunk speed 
>> 1000 channel-group 1 mode active spanning-tree portfast
> 

You original show etherchannel had too many ports in the portchannel. Have
you fixed this yet?

> Now, this being the case, would you not expect these ports to carry 
> traffic on those vlans (1,2,3,100) ?
> If I have a ESX host in gi 0/1 and it has vm's on 2 of those vlans, 
> and 2 virtual switches are configured, one for each ip segment/vlan, 
> then each VM should send traffic to my uplink ports, in this case gi 
> 0/20 & 0/22, which will pass all those vlans on to the next switch?
> 
> It is odd to me, as port gi 0/1 needs to be able to actually 
> accommodate multiple vlans, so to me that would be a trunk. Then, to 
> combine multiple ports to move traffic, I need the etherchannel (with 
> LACP talking to the HP). It seems this all works fine as long as only 
> 1 vlan is applied. AS soon as the expectation is for more than one vlan,
it ceases.
>