I'd like some clarifications to bug 588412 and some information as to a work around (upgrading to SQL Server 2012 is not an option).
In bug 588412 Microsoft states that "We believe we found the cause of the problem. One of the system API we use to help resolve path can return different error codes depending on the network configurations and what are available". What sort of network configurations cause these problems? Also the phrase "... and what are available" looks like it's missing at least one word after "what".
In our particular setup we're configuring input & output locations via a configuration table in the database. The dtexec program is triggered by a .cmd script which sets some environment variables and passes some parameters to dtexec. The .cmd script is started via a SQL Agent job.
As to a work around around: We've tried adding a net use command (which maps the UNC path to a drive letter) to the .cmd file and running the job step using a proxy. Developers who work for our client have routinely used UNC paths (albeit using config strings stored in a dtsConfig file) so the trick with the net use in the .cmd script seems like an awful lot of gyrations to go through to make it work. Sites such as http://www.sql-questions.com/microsoft/SQL-Server-Integration-Services/34273760/ssis-package-hangs-when-running-simple-execute-process-task.aspx mention "adding \\server to the Local
Intranet sites zone in IE". Exactly how farreaching (Just IE, login session, system wide) are the effects of that change in a server environment?
FYI: I think VMWare is being used for virtualization