I am hoping that my trials and tribulations will help someone down the line. For several weeks I had been trying to add SQL nodes to SAM with AppInsight for monitoring, all of which belong to a cluster. I studied the cluster creation from the Solarwinds AppInsight Application User Guide verbatim as well as "Google Foo'd" the living snot out of this topic.
So what was I experiencing? When I added a node (clustered), I would see all of the associated instances. However when added the net result was the same, performance counters not available. This was maddening for sure. I contacted Solarwinds support a few times and was able to pass them logs. We even went so far as doing a GoToMeeting on the Orion server to actually see the logs showing the counter information. I did all of the triage, checking the perfmons on the SQL server, removing them and adding them back. Creating a new "service account" for Solarwinds and making sure this account was properly added to each SQL server. Every single diagnostic hoop was jumped through and believe me, I got quite a behind scenes education on trouble shooting Orion.
I knew in my gut that this issue was communication related. Each time that I attempted to add an instance and assign it to AppInsight, I would always get some return. The server instances showed up and I got some basic monitors but none of the SQL centric ones. I began to take a different approach and did a netstat -a on isolating (1) SQL server with (3) instances on it. I got at definitive return on (1) instance with port 1433. This instance as all had a client assigned as visible through the cluster manager. Inspecting each of the instances in the cluster manager showed their alternated IP addresses. I checked all the A Records in DNS as well. This lead me to my solution.
I popped open the SQL configuration manager on this server (with the 3 instances) and inspected the IP setting for each. I found the IP's referred to in the A Records.....so really good stuff here. I then move all the way down to the Ports section at the bottom and there was the answer. Each had an alternate setting. From here I created a node for each client access name making sure that set up the port for each as static and viola. ALL and I mean ALL the performance monitors were polled and displayed.
Again I hope that this article helps someone with a similar problem. Bottom line is ALWAYS check the ports for each instance. Cheers!