not enough work units
19 Dec 2005 16:13:46 UTC
Topic 190410
(moderation:
was running eistein and rosseta which ran out of work and came to crashing halt, rosseta server would not download new work units, even after the server was fixed, but i was wondering if the problem is with bionic as it keeps telling me the computer is over commited, but runs out of work in a 16 hour cycle, tried changeing
days held in cache but no joy.
not enough work units
)
What projects are attached? What are their resource shares? What is your current cache setting? When it says you are overcommitted, what project does it begin working on? What messages do you get when you click on "update" for Rosetta? Please go to "your account", then "view computers", and view the computer in question - copy the bottom part of that screen, from "last time computer contacted server" down, and paste in here.
This does NOT sound like you have a problem, unless you have bad information stored for the host for things like efficiency, time BOINC is allowed to run, etc. "Overcommitted" means just that, you have some work on the system that may not meet deadline. As long as this is true, it will not get work for other projects.
RE: What projects are
)
hi bill, thank you for your intrest in my problem, have sort of worked it out, ( i was flipping back and forward in windows magic disapearing calander and an email checking my new work roster, and closed the system calander on the wrong date, found this out as both bionic and AVG anti virus "downloading in the background"
both promptly cracked up with messages interal database out of date , and bionic refusing to download new work "overcommited" so these files must be time stamped on download, not at the home server. regards stella
If I understand it right -
)
If I understand it right - you changed the system date on your PC by accident, and work was downloaded while the date was wrong. Since then, you have corrected the date.
It is probable that the BOINC scheduler is very confused. The deadline for returning a result is included in the result, hard-coded. The "time sent" is included with the result. The problem here is that these are converted from UTC to your local date/time for use in the displays. If you look at "Report deadline" in BOINC Manager, then find the same WU on the server, one is in "local time" and the other in UTC time.
I suspect BOINC thinks that you are going to miss the deadlines (or already have) on the work you have in the cache, and has entered "Earliest Deadline First" mode. Until the results that are present have been completed, it's not going to get any new work. So, the best thing to do is just let it work it's way through these, and see if everything then goes back to normal.
The only concern I have is that there are several fields on the server where information about the "efficiency" of your computer are stored; they're down at the bottom of the "View computers" page, and include % time BOINC running, % time connected, etc. I have seen cases where one of these will go from 0.98 to 0 because of something similar to what has happened to you. This basically tells BOINC to only do one result at a time, because your computer is incredibly "slow". While this will eventually solve itself, if in a few days you are still not up and running normally, please look at those numbers, and if any of them are "extremely low", post back here and we'll help you reset them.
Hi bill, thanks for the
)
Hi bill, thanks for the insight.