Hi there, looking for help with a 3 node Harvester cluster.
"I have two Harvester nodes in the same rack with cross-card LACP bonds connected to a Cisco vPC leaf pair. Even though these are two separate servers, the LACP Partner Key being returned by the switch is identical for both nodes. This causes a ‘Last Man Standing’ scenario where the second node to power up kills the first node’s connectivity. To me this confirms that the switch-side ports for both servers are accidentally assigned to the same vPC ID / Port-Channel, but my network colleague claims there are no errors in the Cisco ACI configuration.
Any one here with Harvester nodes living i ACI with 802.3ad bonds?
Good morning, I’ve seen your post and we’d be interested to know how you’ve set it up. We’ve just set up two 802.3d LACP servers with ACI at the same location. We’re planning to set up another two at a different location. We’re concerned about how the Harvester management VIP behaves with ACI when it needs to migrate between nodes.
Many thanks
The trouble I had with bonds configured on ACI was du to misconfigured profiles in ACI. It now works flawlessly. Actually I’ve decide to go ahead with a single 4-uplink LACP bond (and not split for separate bond for storage).
I’m not sure I understand your concern with the cluster VIP. It appears as a secondary /32 IP address on one of the the Harvester control nodes.