SoftGrid 4.1 - Desktop Refresh Times Out - Error 10000005

I've had issues with SoftGrid today with all Desktop Refreshes failing and no applications launching. Unfortunately this happened while I was rebuilding one of the VAS servers so I wasted a lot of time trying to trouble shoot issues that could have been caused by removing this server.......

The errors I was seeing in the sftlog.txt file when trying to launch applications were:

[03/11/2008 08:00:28.370 AMGR WRN] {hap=1:app=CorelDRAW X3 13.0.0.739:tid=4FC}The SoftGrid Client lost contact with a server and entered disconnected mode.
[03/11/2008 08:00:28.385 JGSW ERR] {hap=1:app=CorelDRAW X3 13.0.0.739:tid=148C:usr=srvrjh}The SoftGrid client could not connect to stream URL 'rtsp://appstreaming.leeds.ac.uk:554/cdrawtst/cdrawtst.sft' (FS status: 1600190A-00000120).
[03/11/2008 08:00:28.385 SWAP ERR] {hap=1:app=CorelDRAW X3 13.0.0.739:tid=148C:usr=srvrjh}The client was unable to connect to a SoftGrid server (rc 1600190A-00000120)
[03/11/2008 08:00:28.432 TRAY ERR] {tid=C88:usr=srvrjh}The SoftGrid Client could not launch CorelDRAW X3 13.0.0.739.

An unexpected error occurred. Please report the following error code to your System Administrator.

Error code: 41112E-1600190A-00000120
This turned out to be nothing to do with the VAS rebuild and was actually associated with the provider policy. It seems that SoftGrid does not use native Windows authentication but instead employs LDAP. In our architecture LDAP often causes issues. We have two domains, a forest root and a child. The child domain is first in the alphabet.

Anyway, we performed some AD upgrade work at the weekend and the one domain controller in the child domain that the VAS servers could contact was offline (in case we needed to revert back the domain). We accepted the fact that users in this domain would not be able to authenticate to the VAS servers during this period, as we have no softgrid users in that domain at present. However what we did not anticipate is that because a group of users from the child domain was defined in the provider policy, SoftGrid would time out trying to contact the domain and it would break all desktop refreshes and all applications (that were not 100% pre-cached).

Microsoft states the issues we were seeing are related database problems, but I would suggest that before you do anything with your databases that you check your provider policy and look for any groups that have not been properly resolved.