Running Win 7, 64bit, 4 cores, running ver 7.0.25 .

My BOINC has been running for several hours now with only 1 task. With 4 cores and a low-water mark of .1 days, it seems like it should be looking to grab work from somewhere (I'm attached to 9 different projects). When I do an update to all 9 projects, it won't request work from any of them. This is a bug I assume? Or is there something I'm doing wrong?

Temporarily bumping up the Min Buffer will possibly set things in motion, then select project of priority and hit update. That said, BOINC was not going to let cores idle, but then 7.0.25 was overtaken already by 3 more iterative [test] releases for a long list of bug fixes.

--//--

Is, by any chance, one of the projects you're attached to SETI@home?

SETI is suffering from a configuration problem at the moment (and has been since around 21:00 UTC last night - 12 June 2012). As a result, all SETI tasks will be marked as running 'high priority', and no other projects will be scheduled.

Hopefully, this is a glaring enough boo-boo that it will be rectified in short order (we've notified them already). But, in the meantime, you can bypass it by upgrading to BOINC v7.0.28 from http://boinc.berkeley.edu/download_all.php - although this is marked as a 'may be unstable' version, it's actually better than v7.0.25 in several ways, and it's free of the bug triggered by SETI's new configuration.

It's true that 7.0.28 is better than 7.0.25. But there is still some progress to do.

I've switched 3 of my computers from 7.0.25 to 7.0.28 (1 Mac + 2 32-bit PCs). Each computer is connected to 4 or 5 projects with equal priority (100). The minimum workload is set to 2 days and the maximum workload to 3 days.

On all computers, the same active project (Einstein) has a large amount of work pending and seems to receive new WUs each time that 2 WUs complete. That sounds OK.

On 2 computers (out of 3 connected), another project (Climate Prediction) is running with one long WU (2 weeks). The 3rd computer has fully completed and reported its last WU and is desesperately waiting for a new one.

All other projects have 100% completed their pending WUs and are desesperately waiting for task completion and/or new WUs. For these projects, manual project updates can solve the lack of automatic requests, but then you have again to desesperately wait for new WUs. Project suspension and task abort may also give good results if forcing a request to the scheduler, but disconnect/reconnect seems to be the only secure way to get new WUs (with the risk of loosing WUs waiting for completion). And it works ! So these projects are not out of work.

It's true that 7.0.28 is better than 7.0.25. But there is still some progress to do.

I've switched 3 of my computers from 7.0.25 to 7.0.28 (1 Mac + 2 32-bit PCs). Each computer is connected to 4 or 5 projects with equal priority (100). The minimum workload is set to 2 days and the maximum workload to 3 days.

Boinc 7 has a new scheduler, at the moment you're asking it to cache a maximum of 5 days work, once it gets that amount it won't ask for more until it gets down to the minimum of 2 days,

Best to change your preferences to 4 days minimum, and 0.01 days additional, you'll get more frequent topups with that.

Claggy

I don't know what was planned, but it's clear that these figures don't apply individually to each project : for the time being, on my 3 computers running 7.0.28, only one project receives new WUs. Other projects have terminated crunching all WUs and uploaded the results, but are still waiting for WU completion.

I've changed the limits to 4 days + 0.1 day and I'll see what happens.

I'm progressing.

Each computer seems to have a "lucky project" which receives workload for all computer's cores. With my current settings (4 days + 0.1 day, 4 or 5 projects running with equal priority - 100), the workload for this project is always superior to 96h * number of cores (192 hours for a dual core, 384 hours for a quad core).
Please note that all my computers haven't the same "lucky" project.

Then 2 situations :
- if the WU deadline is short, tasks will run in high priority and the other projects will never receive new WUs. That's what happens to one of my computers.
- if the WU deadline is long enough, there is some time free for the other projects. They receive 1 WU at a time on a round robin mode, and it may take a very long time before getting automatic updates and new WUs for all projects. That's what happens to my other 2 computers.

I would like to suggest a SW modification.

There is a description of the bug in FAQ: BOINC 7.0.

The best thing to do is to let things go without any manual intervention.