Remotely connecting vCenter Server to ESXi hosts

When supporting a number of small businesses, it is quite useful to connect a local vCenter server to manage clients’ remote ESXi hosts.  Setting this up for the first time can be frustrating, however, with the documentation being so comprehensive, paring it back to just what you need means it is easy to lose your way.  

The first thing to do is to work out the necessary ports.  For a site with a single host and an Essentials license, there is no need for vMotion, HA, or AD integration.  This simplifies matters, meaning the client’s firewall only needs the following TCP ports opened: 22, 80, 443, 902, 5988 & 5989.  In addition UDP port 902 should also be opened.  On your vCenter server, the list is shorter: UDP port 902 and TCP ports 902, 5988 and 5989.  The vCenter Server installation will have created all the necessary Windows Firewall rules, so only router/firewall NAT entries need to be created.

This then allows you to add the remote ESXi host to your vCenter server, although note that the latest version of vCenter (v5.1.0) doesn’t permit to you add ESXi hypervisors that use the free license.  

One thing to look out for that had me puzzled for a while is when, after successfully adding a host to your vCenter Server, it disconnects within seconds.  I traced this to the fact that that my server had two network cards installed and although it shouldn’t make any difference, it disconnected when the vCenter Server was bound to one but remained connected when bound to the other.  You can change the NIC binding in the vSphere client under Administration, vCenter Server Settings, Runtime Settings.  Note that where the Managed IP address is set, written in bold text is the following: “Does not require manual restart of vCenter Server”.  Don’t be fooled by this as I was.  After changing the value, stop all VMWare services and restart them – only then did my connection stop dropping after a few seconds.