Your question appears to be slightly confusing, you refer to a duplicated HTTP event collector server, however the HTTP event collector is just another Splunk input, therefore the only way to run an active/active scenario would be 2 universal forwarders or 2 heavy forwarders. It would not make sense to run multiple http event collector ports on the same instance of Splunk.
The latter part of your question mentions two tokens on the SDK, I assume your again referring to the HTTP event collector ? You would use the same token on 2 forwarders if you were doing active/active, and then you would add additional tokens if required.
The inputs.conf documentation is here refer to the http event collector section. Or refer to the http event collector documentation
You will need something to load balance between the 2 servers, I use a load balancer server to distribute traffic to 2 heavy forwarders, however you could also do this using universal forwarders as per this Splunk blog post comparing heavy vs universal forwarders.
