Windows

For months, even years now my company has had Windows98-2000 Terminal Server Clients connecting remotely through Cisco Firewalls using DSL/Cable and also through Cisco Routers via Frame Relay to 2000 application mode terminal servers. Some attach to TServers residing at our datacenter and some are in totally seperate domains residing at client locations. Recently some of the 98 clients have been unable to connect, getting the "client cannot connect....server too busy" error. This is only happening with VPN clients running win98 on client side. This issue is popping up in multiple locations, in our domain as well as in totally seperate domain/networks. No 2000/XP TS clients are being affected whatsoever. And no win98/frame relay clients are affected either.

The VPN's are still being established. You can ping the terminal servers. I have even ran port 3389 tests according MS KB article 187628 which executes successfully from the failing client workstations. Termserv CALS ARE available to the 98 clients - EVEN 98 clients coming through frame-relay connections can connect to the same 2000 terminal servers. No events are being logged whatsoever for "normal" Terminal services issues on the terminal server. What is difference with the communication of 98 machines? Are there tests that can be done to go deeper beyond "port pinging" in correlation with the necessary communication for Terminal services. Wouldn't the successfully connecting 2000 remote workstations be proving that the VPN tunnel is "open-enough" on the PIX's? Any help on this would be appreciated. I can't find no reported articles anywhere of this issue. Thanks