Has anyone seen or have a resolution for TFS215085 and TFS215074?

TFS215085: An error occurred while connecting to agent X

TFS215074: Failed to start build agent on 'X' because another build was unexpectedly in progress. Investigate build agent status to determ开发者_运维知识库ine the cause.

I am queuing up a large number of builds most of which run fine, occasionally a build will fail with these messages and the agent will become unreachable. Often it will automatically reset it self and become enabled again and builds will continue on that agent. Sometimes it will stay unreachable, but if I manually change its status back to enabled it will start processing builds again.

There are no TFS messages or other messages in the application logs related to this. The web IIS logs for the webservices give no indication there as well.

This is for Team Foundation Server 2008 SP1 running on both Windows Server 2008 and Windows Server 2003. I have turned logging on when this has failed, but there is no additional information there as well.

Any suggestions or hidden switches for managing the build agent that would resolve this issue would be greatly appreciated.

This error occurs since there is no official synchronization between the TFS application tier and the build machine(s) other than the server queue state. For instance, if a build is issued a stop command but doesn't actually stop on the machine, the server (in 2008) will mark it as stopped after a timeout period no matter what the outcome which immediately allows the next build in the queue to start. When this build attempts to start, it will fail with the error message above (TF215074) since we only allowed one build to run on an agent at any given time and this constraint has been violated. Is this issue happening often? Do you know how to reproduce it?

Patrick