cjslman wrote:
Maybe things are changing... it used to be that I was getting the OET WUs only once a day at a certain time (this went on for about 3 days). Since yesterday I have received groups of WUs (9-10) at different hours of the day. So far I have 27 WUs in the queue and 15 in PV jail.

CJSL

Crunching for a cure...

Sandvika wrote:


Hi Branjo, thanks for your tip. In the mean time I looked in my event log and noticed that whilst BOINC is running jobs it keeps on looking for more work regularly. On the other hand, as soon as it finds no work to do, it lengthens the polling interval each time that it can't find work and it quickly escalates to be hours. Consequently, in those brief periods when OET WUs become available, it is sleeping instead of looking for work! (Projects->Properties: Scheduling: CPU work fetch values) Clicking Projects->Update resets the polling interval, starting at 2 minutes. I started Googling and quickly found a discussion of the BOINC V6 vs V7 scheduler, which led me to editing the global_prefs_override config file and reloading it (Advanced->Read local prefs file).

Then I found your tip and discovered the values I'd set in the config file had been populated into the "network usage" preferences you mentioned Now I see them, it's clear what the fields are for, it just didn't occur to me that scheduler parameters would be set in the networking configuration It's a for usability when the config file is more intuitive than the user interface

I'll see how self-sustaining it becomes once it actually has OET work to do with the new settings. Ideally I don't want to oversubscribe the cores on my PC, so if I can keep it topped up to at least 2 days with either OET or UGM, I can change the device profile in use to get UGM once a day if it is running short and leave it on OET the rest of the time.

Cheers and