Hi everyone!
We are struggling with our new Terminal Server setup and I am kindly asking for your help.
- 2 Remote Desktop Session Hosts (load balanced, 20 cores and 64 GB memory each) + 1 Broker/Licensing. No gateway, only local network access. SSO with AD credentials enabled and working. Currently working with local profiles (no roaming, UPD recently
disabled) for troubleshooting reasons. No MS Office installed. Mainly used software is a Terminal Server supported typical software for our business area.
- 1 Print Server (2 cores, 8 GB memory) with several new and old printers, latest drivers from the manufacturers.
All of them are Server 2019 DC 1809, patched up to January 2020 level.
Hardware base is vSphere 6.7 on a fresh HPE ProLiant set, shared SSD storage and lots of memory. There are no performance issues on a hardware base, also not on other VMs.
Printers get mapped into the Termial sessions via GPO ("Create" command as user policy with loopback mode, \\printserver\printername).
So far - so good. Most of the time all is working fine, users can print from the sessions and the sessions are performing quite well.
Since this setup has gone live a few days ago, we are experiencing intermittent issues in the sessions that have not been recognized during an extended testing phase. It seems that the Print Server's spoolsv.exe is hanging or sometimes crashing and during
this period all Terminal sessions on both hosts are freezing or hanging their applications ("Not responding"). Every session does have mapped the same printers, about 12 in total. Usually programs are freezing during work without
opening a printing dialog, but at the latest if a user tries to print something, the session is gone. It happens 1-5 times a day.
During that phase the RDSH are nominally not hugely loaded (10-20% CPU, 30% memory with ~30 users on each, so they are clearly overpowered), users are no longer able to log on or off ("Remote Desktop Service is busy"). After a couple of minutes
or if we restart the printer spoolers on both RDSH and Print Server, the system is running again and users, that have not tried to log off, can continue their session at normal speed. After another 5 minutes new sessions can being opened again, so the RD service
is no longer busy with itself.
The load on the Print Server is usually ~2-5% CPU, during a freeze it may go up to 20%. The Print Management console may be crashing also, or at least it is very slow.
We've tried to identify a faulty driver on the Print Server and have found one using the eventvwr that caused the spoolsv to crash, however the issue is still there. We've also removed all unused drivers from both RDSH and the Print Server.
We have enabled 2 new GPO settings for the hosts: "Always render print jobs on the server" and "Execute print drivers in isolated processes" without a noticeable change. On the Print Server all possible drivers are also running in Isolation (Mode 2), however not all of them are supporting this. The physical printers themselves are working fine, PC clients do have some of them installed locally on their machine without known issues.
What's unclear for me: It seems that the RD sessions are constantly communicating with the print server (according to the Resource Monitor), even when nothing is being printed. Therefore they immediately are slowing down once the Print Server is having an issue. Is this an usual behavior? Does it depend on which Default Printer the users are setting?
So the idea is now either to find the faulty driver/setting and/or to reduce the dependency on the print server, so even when it hangs or may be offline, the sessions should not hang up.
I appreciate every single hint that might help to identify the issue or to find a workaround.
Let me know if I missed any important information.
Many thanks and regards!