oops, SORRY Ady ... got annoyed and just posted ... forgot what PC I was on.

5mins later it did pick up some more work.

LOL - You know insanity has set in when you find yourself manually forcing a results upload (on my dual core machines I will sometimes have two to report, two in processing, and two in the input queue) in order to be included in an upcoming stats run!

Ah well, it DOES help to read the BOINC wiki ( here ) but sometimes it makes my head hurt However, there does not seem to be much in the way of connections between uploading the result of running a WU, reporting the result and getting new WU's. Once BOINC finishes a WU, it looks like it will upload the results at that time (unless you go mucking around and prevent that or the server can't receive it for any of a multitude of reasons). The reporting of the results is a separate process. Thus it is quite possible that an event where results get reported to the server can in fact report multiple results at the same time.

Getting new WU's is another separate process. Big surprise here, if you complete a WU, it will then download a new WU to start crunching. Where it can get interesting is when you use "Connect to network about every N days" to download multiple WUs. Basically, it seems that it figures out that, when BOINC thinks it should go download a WU, it figures out that, given the WUs you currently have and this profile setting, you need an amount of work that is basically the difference between the two. If you look at the wiki info for the To Completion Column , the value that initially gets sets there is an estimate of how long the WU will take to process on your machine. If you ever browse thru the stuff on the Messages tab of the BOINC Manager, you may notice message about tasks being paused, a benchmark being run and the tasks being resumed. Just like the old UD Agent periodically benchmarked your machine, BOINC does too. I assume that, just like with the UD agent, if your machine is REAL busy doing other things when the benchmark runs, then that can impact all of this since a WU could get estimated to take a lot longer that it really does if the load on your CPU when the benchmark runs is unusually high.

I'm still missing a few things clearly as I can't help but notice that, if three WU get downloaded, they ALL have the same initial value in the To Cmpletion column. However, if I add up all the values in that column (for the WCG project that is - some folks run multiple projects), that seems to always be a number that is bigger that the "Connect to network about every N days" value converted to hours by just enough that one less WU would take the total below that. I've also seen that these initial values seem to be somewhere very close to my other running or already downloaded WUs. Not sure I REALLY want to understand this in much more detail though.

I've used English. However, whether it's comprehensible English is a whole other question!

It does pay to read thru the wiki and do a few searches on basic things. Good info there!

I put into production today some changes to the server that should reduce the chances of this happening. Please post if you have seen this occur again - espicially if it has not retrieved work within an hour.