We have a home-grown batch application generating print jobs via low-level C++ Win32 API calls. The application is running in desktop mode on an instance of Server 2008 R2 in 32-bit mode. The print queue is hosted on a different Server 2008 R2 server. The application is printing correspondence to a print queue that is load-balanced to several TC/PIP printers. Each print job is only a few pages, but there are hundreds to thousands of them. The application only calls CreateDC(...) once and then reuses that device context for each print job. When the issue occurs, we can tell from instrumentation in the application logs that even though the actual print job only took a few seconds to spool, the final call for each print job to EndDoc(hDC) takes exactly 7.5 minutes to complete. We have a suspicion that the issue may be related to when one or more of the printers configured to this queue or even another queue on the same server is in some error state like jammed, out-of-paper or toner, etc. Cancelling and restarting the application only helps briefly then the problem comes back. Rebooting both the client & server also only helps briefly, but it seems that finding and fixing the offending physical printer is the only permanent cure.
Anybody got any ideas?