Been pulling my hair out on this, seems to defy logic.
Environment: 3 nodes operating normal, adding a 4th node, of identical config. The 3 hosts have vmnic0 and 1 on vSwitch0, as uplinks for mgmt and vm traffic. Load balance method is set to Route based on IP hash. Both adapters are active. Switch has a port channel for each of these 3 hosts, with src-dst mac as the load balance algorithm. Traffic flows without issue, mgmt works fine, no performance issues on these 3 hosts.
Situation: Adding the 4th node, with the identical config as above. Connected to vcentre with one adapter connected. Applied the same vSwitch0 config to the 4th node, (MTU, hash, uplinks), configured the port channel and ports identical to the other 3 (own port channel). Management stays connected, pings drop. Restart the host, pings succeed, management/connection to the host fails. Shut down one port in the port channel, I can connect the host in vcentre again. Both ports up, no connection. I can however, connect to the host on its URL, just no connection in vmware.
Ive verified the port config on the switch, the vswitch is identical in its config. The only difference is the version of esxi hypervisor, the first 3 are 7.0.3, 21424296 and the 4th is 7.0.3, 21930508. Cant see how that has anything to do with this issue, but everything is identically configured.