You could take a peek at the ULS logs and see if anything is outputting related to what you are experiencing.  Sometimes the logs give better and deeper details the messages displayed on the pages.

Thanks

Mike

Jeff,

That seems like some odd behavior. While this does not fix what you may be seeing, have you tried using the filter action inside the workflow instead of a conditional start.

I'm not saying that the conditional start isn't a way to go, but you have a few different things going on and one may be causing the other to fire weird. The timing for conditional start and closing the form is a disconnected issue for Nintex and one that I would suggest like Mike to look at your logs.  The workflow starting shouldn't affect the browser closing out the form as the workflow is firing off in the background.

One thing to note about page closing times, SharePoint will cache workflows when they run. So the first time they run after an App Pool recycle will take longer than successive runs.

Being on premise, how is your farm configured? Do you have a separate app server from the web front end? For the servername listed in the error, what role does it play?

I agree, these two things solve more problems than I can list.

I'm going to mark this as answered. Wasn't a very specific question to begin with, and I'm no longer confident that the long delay on a form submission is correlated with a Nintex conditional test. I'm still curious about how Nintex would have the previous value of a field (to test the condition) though.