[c-nsp] Cisco 6509 LACP

[Mike Glass]

I do have this working now, it was a configuration on the vSphere side, I never configured a distributed switch before, but I created a Dswitch, then for each vlan had to create a DportGroup with IP hash load balancing, then on the DVUplinks I set it for vlan trunking for both vlans. Hopefully this is correct, it seems to be working fine though.


I have this duplicated for vlan 4








Thanks for all your help.

Mike


>>> Tóth András<diosbejgli at gmail.com> 2/11/2013 10:49 AM >>>

Hi Mike,


You can configure one unassigned IP address to each switch's "interface Vlan x", and check if they can ping each other when using a port-channel. This should not cause any harm if you use an unassigned IP. The configured IP then can be used for testing between the force10 and vmware, by doing a direct ping test between them, then between the 6500 and the vmware.


Best regards,
Andras





On Mon, Feb 11, 2013 at 6:36 PM, Mike Glass <MGlass at lccountymt.gov> wrote:

Here are the results
----------------
6509
----------------
#sh spanning-tree int port-channel 1---------------
Vlan Role Sts Cost Prio.Nbr Type
---------------- ---- --- --------- -------- --------------------------------
VLAN0002 Desg FWD 3 128.1665 P2p
VLAN0004 Desg FWD 3 128.1665 P2p
Cat6509# 
------------------------------------------------------------------------------------------------
-----------------
force10
-----------------
AIS-Prod-1#sh spanning-tree rstp int Port-channel 1
Port-channel 1 is designated forwarding
Edge port:no (default) port guard :none (default)
Link type: point-to-point (auto) bpdu filter:disable (default)
Bpdu guard :disable bpduguard shutdown-on-violation : disable
Bpdus sent 25200, received 135714
Interface Designated
Name PortID Prio Cost Sts Cost Bridge ID PortID
--------- -------- ---- ------- --------- ------- -------------------- --------
Po 1 128.2 128 18000 FWD 0 32768 0001.e88b.52b8 128.2
AIS-Prod-1#sh spanning-tree rstp int Port-channel 2
Port-channel 2 is designated forwarding
Edge port:no (default) port guard :none (default)
Link type: point-to-point (auto) bpdu filter:disable (default)
Bpdu guard :disable bpduguard shutdown-on-violation : disable
Bpdus sent 129288, received 0
Interface Designated
Name PortID Prio Cost Sts Cost Bridge ID PortID
--------- -------- ---- ------- --------- ------- -------------------- --------
Po 2 128.3 128 1800 FWD 0 32768 0001.e88b.52b8 128.3
-------------------------------------------------------------------------------------------------------------------
The Dell switch does has not have cdp, just lldp, on the cisco switch I see nothing in cdp for the interfaces, the dell lldp I do see the vmware interfaces.
I have switched to mode active with same results.
I have not configured address on the vlans, not sure if that will break anything in our existing system, If I do put ip address on a vlan on our production system will it cause any problems?
The force 10 vlans are tagged for vlans 4 and 2
interface Vlan 4
no ip address
tagged Port-channel 1-2
no shutdown
when I shut the port-channel off on either switch I see messages on both switches that eh protocol is down, so I think I am close but still cannot pass traffic from the a vm guest.
I also wonder if I already have the Vmware server hooked up to the 6509 on another interface that uses the same vlan could be causing some problems.
Thanks
Mike


>>> Tóth András<diosbejgli at gmail.com> 2/10/2013 1:42 PM >>>
Hi Mike,


First of all, how do you see that traffic is not passing? Could you please elaborate on how do you test it, from where to where are you trying to send traffic or ping tests? Does traffic pass if links are used as standalone (not being in a channel)?


If you configure an IP address for each vlan interface, can you ping between the 6500 and the force10? If yes, the problem might be between the vmware and force10 instead.


Are the force10 ports configured for dot1q trunk as well to be able to carry multiple vlans? Are they set with the same native vlan as the Cisco? (Default native vlan is 1)


Could you do a sh spanning-tree on both switches for the po interfaces to make sure they are not blocking? sh span int po1


Do you see the correct neighbors in cdp table? sh cdp nei


Can you confirm that the same issue is observed when using 'mode active'?


Best regards,
Andras