Hello.
I have a 4-ports EtherChannel on a pair of stacked Catalyst 2960XR, connected via a cross-switch LAG to a IBM EN4093.
I would like to make sure, the LAG we set up is correctly configured, before executing any physical fail-over test. Since I'm new on Cisco and another provider (IBM) is involved, I'm seeking a second opinion about the configuration made.
My two WS-C2960XR-48TS-I are stacked together in "native" mode, they serve as "top of the rack" switches in a small site, hosted in a Datacenter. I'm connecting to them a IBM PureFlex chassis's switch, EN4093 (single switch), configured by a IBM Specialist. The connection between the single EN4093 and the two 2960XR needs to be fast and redundant, so I set up a cross-switch EtherChannel. Actually, this LAG have 8 ports on 2960XR, but only 4 ports on EN4093. This is intentional, a second EN4093 will be added in the future and they will connect to 2960XR using all 8 ports. This way, the ports are reserved and already configured at 2960XR.
On 2960XR th configuration is fairly short ans simple, I take 4 ports from each switch, to create a 8-port EtherChannel and make it talk LACP:
interface port-channel 1
description "LAG 1 Servers"
interface range gi1/0/1-4,gi2/0/1-4
channel-group 1 mode active
I don't have the reference of the EN4093 configuration, it was configured by a IBM specialist under the requirement "4-port link aggregation with LACP IEEE 802.1aq protocol".
Something weired happened when we connected a cross-switch LAG on the Catalysts to a downlink switch at first: only 2 ports, located on the same Catalyst, was functional at time.
"show EtherChannel 1 detail" reported only 2 active ports on it's tail:
Ports in the Port-channel:
Index Load Port EC state No of bits
------+------+------+------------------+-----------
0 00 Gi1/0/1 On 0
0 00 Gi1/0/2 On 0
The configuration was functional and passed some simple fail-over test (we disconnected Gi1/0/1-2, it continued working on Gi2/0/1-2) and IBM specialist had to leave.
Actually, I was about to ask for help regarding this a s an issue (only 2 ports active at a time), but while collecting diagnostics I realized that all 4 ports are online. AFAIk, the IBM switch have not been touches since then, and I have made nothing to the Catalysts.
Questions:
1. The configuration I believe I have made, is it supported? According to the specification, WS-C2960XR-48TS-I supports cross-chassis EtherChannel, but I'm not sure if they mean something different from what I have done.
2. Does the configuration I have done corresponds to the requirement "LAG with LACP IEEE 802.1aq", or am I missing something?
3. According to the diagnostics I'm attaching, is my LAG configured correctly? Is there any other diagnostics worth to run, before I start actually to unplug things and see how redundancy will save the day?
Thank you.
