The problem with all of that is that it is now almost a year and WCG has not been able to function without issues. Any testing or determination of the overall system just gets more complicated and thus drawn out when too many folks think too selfish and as I put it "are hoarding WUs". There are people who have proclaimed proudly that they have caches of 5+ days and then lament "they don't have any problems" when everyone else is having issues with download errors.
It just doesn't allow for a proper assessment of the basic functionality. And we are far from having things "in running order", despite WCG Towers proclaiming otherwise...

Ralf

50 here, as well!

Is it really the cruncher's fault that boinc allows a 10d cache setting, but more importantly, the server allows such cache to fill up? Sounds more like the a problem on the project end, which is allowing anyone to get virtually as many tasks as they want rather than limiting it to a ~2 per thread basis to make it "fair" for everyone.

Besides, there are valid reasons to run big queues anyways.

I'm sure there are legitimate reasons to run a 10 queue, such as if you have system which chews through work units where a shorter queue leaves the system with idle threads. it is unfortunate that there are some who may use a large queue without realizing the it is too big and some work units will get either server aborted due to not being started on time or they finish some too late to be counted. They may be part of the set and forget crew who feel as long as their system is not idle life is good. Fortunately, BOINC reissues the work unit and it gets crunched, although not as soon as it could and there is some wasted energy and crunching involved.
Anytime there is a system with this many volunteers, not everyone necessarily follows all the conventions of efficiency, but they do contribute.
I'm pretty sure they could do some work on back end to minimize this behavior, but if the number of late units is small enough, it would probably not be worth the effort.
All that being said, the supply issue is still a bit spotty and I have a few threads which are idle due to the almost ever present "no work available " messages. Hopefully, given time, my 2 day queues will be fully replenished.
Cheers

Sgt. Joe; no slap at you here, but for average users this is all akin to metaphysics.

Regarding most efficient way to run or hold on WU's...I've been here almost 3 years and have no idea of those most efficient methods. The average volunteer won't either. We load and go what we can, when we can, on whatever hardware we've got available.

I'm not running high-end servers, or some of the other 'uber-hardware' some are, so limit my downloads to about 32 WUs across 3-4 projects at any time (so about 0.6-0.7 days worth max at one time....sometimes less with WCG issues choking off downloads or uploads.

If holding so few jobs is hoarding I'm at a loss to justify even participating to run 60-80 WUs per 24 hour period while the 'power crowd' is holding onto far more work than even they can process in the same time period. My lone machine runs 23/7 (1 hour rest break), and every morning I can send back my small contribution, as well as through the day if WCG is running right, and I do my part to link up, send/receive WU's, and go about life.

If anything the high-end crowd need to not hold so many jobs at one time, that other more humble users can take a bow once in a while.