My good friend JP Poissant found a little configuration glitch the other day when building out a new farm. He found that when he tried to start the distributed cache service running on a second server in the farm, it failed to start. The service was listed but he was getting the "cacheHostInfo is null" error. Ultimately what he found is that the first cache host in the farm needs to be configured to allow ICMP traffic through the firewall. In JP's words: "To have Distributed Cache working on more than one servers, the first server with Distributed Cache needs to have its firewall set to allow for Inbound ICMP (ICMPv4). http://technet.microsoft.com/en-us/library/cc972926(v=WS.10).aspx. If the Distributed Cache is reconfigured to a dedicated cluster and the original server no longer has the Distributed Cache running, then, the first server of the cluster should have that firewall setting enabled too."
Many thanks to JP for tracking this down and more importantly, sharing it with the rest of us. 🙂