Uploads started failing sometime yesterday, 5/13/10. This has been confirmed on the forum. Result
turnaround time is up to a whopping 240.33 hours and it peaked at close to 300. There is no actual
indication whether the people that run the system actually know there is a problem, meaning it could
be broken until after the Tuesday shutdown. 

Per previous discussion, when Boinc notices more than a threshold number of failed uploads, it
refuses to do any more downloads. The assumption is, I suppose, that further work will not be
returned within the deadline. I have noticed that my other, slower computers (with fewer failed
uploads piled up) are still able to do downloads, at least until the failing uploads pile up to the
threshold. Perhaps this time I will actually observe what that threshold is. On a computer that
processes five units at once, this is harder to observe. 

My question is, is there any way to alter or patch that threshold, or otherwise trick Boinc into
downloading more work, at least until the upload problem is fixed? The deadlines are weeks away; I
suspect the problem will be fixed by then. Yes, I have considered using another computer as a
"zombie downloader", and transferring Boinc "packages" en mass to the faster computer. This could
even be done on one computer by archiving Boinc "packages" and loading them back later. This seems
like way too much of a PITA and waste of time to even consider, and could turn into a huge squashed
Boinc mess if not done properly. Maybe if someone creates a batch file to do this with a name like
"Boinchive" I will think about it. I suppose I should have just set a huge queue size whin I noticed
uploads failing (if not sooner). 

Surely, someone must know a way to patch, zap, or otherwise change, trick, or cajole that "failed
upload" threshold to make it possible to again download work until the uploads are fixed?

In the mean time, it is a great way to slow down the faster computers, and let the slower ones catch
up. :)