I'm new to the BOINC client, so forgive me if this is dumb. I installed the other day and was attached only to SETI and things were working great. Then I tried adding Predictor with 50% resource allocation. Predictor went nuts and downloaded about 30 work units, but didn't actually do any work on any of them. In the meantime, SETI stopped cold. I detached from Predictor, but SETI begin giving me some strange error messages. I uninstalled BOINC entirely, then re-installed and conected to SETI. That was almost 24 hours ago and since SETI continually returns "no work available", which I hadn't seen once prior to the Predictor episode. Is this normal, or is something still goofy? Thank for your time....Jer

Hi,

WU supply has been rather spasmodic of late, Although I did download 2 units about three hours ago.




--
<p>Regards, Paul - Just slowly BOINCing along...</p>

-Sigh-

So you just tossed out a bunch of Predictor WUs huh? :-(

Depending on the deadline of the WUs, the P@Hs were most likely waiting for your S@Hs to complete.

And S@H has been down recently, so you trashed Predictor WUs for no reason.

<a> [/url]

remember it only runs one project at a time
ttyl
Jeff

Well now I feel like a complete idiot.....thanks for your time though. Fo rnow I guess I'll just stick to SETI....

I was having the same Issue, but I had a 10% Predictor to 90% Seti. Seti would not download any new work and Predictor downloaded like 20 units even though I had set the pref for it to only get a 1/2 days worth of work. I deattached from the Predictor project and seti downloaded work and all is good.

> I was having the same Issue, but I had a 10% Predictor to 90% Seti. Seti
> would not download any new work and Predictor downloaded like 20 units even
> though I had set the pref for it to only get a 1/2 days worth of work. I
> deattached from the Predictor project and seti downloaded work and all is
> good.
>
There is only one work queue. You cannot set Predictors queue one way, and S@Hs queue some other. They fill the SAME queue.

Ok if that's the case then what is the point of setting the amount of time each project gets if it only works on the current work unit in the list? Or am I still not understanding how thing works?

Ok if that's the case then what is the point of setting the amount of time each project gets if it only works on the current work unit in the list? Or am I still not understanding how thing works?

Ok if that's the case then what is the point of setting the amount of time each project gets if it only works on the current work unit in the list? Or am I still not understanding how thing works?

> Ok if that's the case then what is the point of setting the amount of time
> each project gets if it only works on the current work unit in the list? Or
> am I still not understanding how thing works?
>
The amount of time that each one gets is set by the resource share. When the queue is at the low water mark, or a processor is without work, BOINC will calculate all of the resource debts, and order the projects by this number. The project with the highets resource debt will be contacted for work. If that project can fill the queue, it does. If it cannot, then the next project is contacted to fill the remainder of the queue. The resource debt is calculated from the resource shares, and the exp_avg_cpu for each project. So when a steady state has been reached, S@H will get to fill the queue, followed by Predictor, then S@H ... assuming 50-50. If the usage is 90-10, then one of the projects will get to fill the queue about 9 times out of 10.

Jer, you have 4 computers listed under your acct, you need to merge them into one....

Bah!!! I am in it for the money