This is worrying!
This sounds exactly the same as the problems we were having on our metaframe 1.8. We are shortly creating a new seperate W2k domain and installing a new Metaframe XPe farm. I was hoping this problem was going to die when I pulled the plug on the old metaframe server.
CNORRIS is on the right track. Try these to see if you get the same results when you can't connect:
Right click an idle session and click reset, then try to connect and it should be fine (let us know). OR
Log out (or reset) any user, then try to connect and it should be fine (let us know).
The thread of my original problem was:
thread48-266276 We spent a long time on this, we tried hotfixes, investigated licensing, and pulled out a lot of hair.
My gut feeling was the problem was something to do with WINS and IP resolution. We had recently added some W2k DNS servers and they were now running WINS on a new subnet.
I tried cleaning WINS out, but it was getting bad addresses from a partner. So I uninstalled it, but the partners kept talking to it, so I surgically removed it from the registry. The Citrix problem remained, so I built a new WINS DB and did not partner it at all. The problem seemed to get better, but it returned.
As this was a production server and I needed a fix, i decided to do a temporary botch job. I setup an NT Workstation with a schedule to launch a citrix session every few minutes. I modified the startup folder of that account to log out immediately. This works as long as the NTPC does not crash and provides a bit of breathing space.
If this problem is in Citrix XP I will seriously consider dumping Citrix, TS seems better now and there is always Netilla.
Please update this thread, I am keen to know how you get on.