App_Config Files Don't Play Nicely with Backup Projects

If this is common knowledge, pardon me; but I just discovered this fact this morning, so I figured I'd share.

If you're using an app_config file to limit the number of WUs that are crunched simultaneously on a project, and your other projects run out of work, you will NOT receive WUs from your backup project.

As an example, I'm running 1 Atlas WU/computer. I'm also trying to crunch Rioja and IGEM for goals. However, both of those projects ran out of work, leaving my machines dry. I had set up NFS as a backup, but that did not kick in, since I had enough Atlas WUs downloaded to cover all my threads. [NFS has been a great backup on machines where I've not been running Atlas.]

Yes, BOINC work fetch is not smart enough to understand that all tasks downloaded cannot run on all threads.

If only it'd be smart enough to rely on the nidle count for requesting work from the backup projects rather than what's in cache. :)