I am working on monitoring connection servers with REST API right now and have some issues with counting sessions. For each connection server REST API shows nearly twice more sessions then there should be. For example: in the admin page dashboard session count shows around 2000 sessions but if I dig a bit further and open connection servers monitoring tab I can see there are around 550 sessions on each of them. We have six servers and their session sum is greater than 3300. We do not even have that much users and there can not be so much sessions on each connection server. Even at night each connection server has around 300 sessions which is ridiculous because there are as much as 200 night users at maximum for the whole VDI and connection servers session count should be 30 or 40. We also have six UAGs in front of our connection servers and their active sessions count (in edge service session statistics) is mostly right: it is 300-360 sessions per UAG which is an okay number. But why does it show thousands (tens of thousands sometimes!) of inactive sessions? One of my UAGs showing 322 active sessions and 5077 total sessions right now! It is 4755 inactive, 3807 Blast (WOW!) and 264 tunnel (seems okay). And this is just ONE of my UAGs. Another UAG edge service stat claiming there are 12528 inactive sessions (why keep them?) which is as much as x6 total user count. Surprising, isn’t it?
So I want to ask if this is an expected behavior? What should I do to get a correct session count and how am I supposed to react at these numbers? With all these strange numbers my VDI works fine, but they still bother me.
My Horizon setup looks like this: firewall -> external L7 load balancer -> six UAG’s -> internal L7 load balancer -> six connection servers. There are a number of internal clients connecting directly to internal load balancer bypassing UAGs but 95% connections are external.
Versions are: connection servers 7.12, UAGs 3.9.1.