After participating in SETI@home for a few years, I finally added World Community Grid and have been running both for about 3 months now. I always see tasks for both projects running or waiting to be run (I'm set to have 5 days of work at a time).

I just upgraded to BOINC 5.4.9 (from 5.2.13). But it's only showing Seti work -- no WCG work. They're both showing in the Projects tab, but when I go to the Tasks tab, it only shows a group of Seti works.

The upgrade seems to be only the event that drew my attention to the problem, not the cause of it. When I go to the WCG site, I see that the last upload it did from my computer was 6 days ago. (!?)

I tried clicking "Update" when highlighting the WCG project, but although it reported the scheduling request as successful, no new work was downloaded.

Here are the messages I get when I boot up (in case the helps to figure out the problem):

17/06/2006 12:33:13 PM|SETI@home|URL: http://setiathome.berkeley.edu/; Computer ID: XXXXXX; location: home; project prefs: default
17/06/2006 12:33:13 PM|World Community Grid|URL: http://www.worldcommunitygrid.org/; Computer ID: ZZZZZ; location: ; project prefs: default
17/06/2006 12:33:13 PM||General prefs: from SETI@home (last modified 2006-03-11 11:28:28)
17/06/2006 12:33:13 PM||General prefs: no separate prefs for home; using your defaults
17/06/2006 12:33:13 PM||Local control only allowed
17/06/2006 12:33:13 PM||Listening on port YYYYY
17/06/2006 12:33:13 PM||Suspending network activity - running CPU benchmarks
17/06/2006 12:33:16 PM||Running CPU benchmarks
17/06/2006 12:34:15 PM||Benchmark results:
17/06/2006 12:34:15 PM|| Number of CPUs: 1
17/06/2006 12:34:15 PM|| 906 floating point MIPS (Whetstone) per CPU
17/06/2006 12:34:15 PM|| 1815 integer MIPS (Dhrystone) per CPU
17/06/2006 12:34:15 PM||Finished CPU benchmarks

I'm not sure how to read most of that, other than generally. Should the Computer IDs in both projects be the same? They're not. I wasn't sure if it was wise to post the port number it is "listening to" in a public forum, so I used letters instead.

Why would it give my computer location for SETI as "home" but it is blank for WCG? And why would it say "local control only allowed"? I'm wondering if these might be indications of the problem. I only have one computer running BOINC.

When I try to Update the WCG project, it puts out these messages

17/06/2006 12:48:21 PM|World Community Grid|Sending scheduler request to https://secure.worldcommunitygrid.org/boinc/wcg_cgi/fcgi
17/06/2006 12:48:21 PM|World Community Grid|Reason: Requested by user
17/06/2006 12:48:21 PM|World Community Grid|(not requesting new work or reporting completed tasks)
17/06/2006 12:48:26 PM|World Community Grid|Scheduler request succeeded

Why would it say "not requesting new work"? That's exactly what I'm trying to do.

It also says (under Projects, right after I Update and after it puts out the above messages) that it is "deferring communication" for 10 minutes. It then counts down for 10 minutes but then nothing new happens -- no new messages, nothing. And still no WCG work.

Any ideas what the problem is? Thanks for the help.

Starhugger

You probably already have enough work on hand or possibly Seti is owed some CPU time. As soon as the debt is repaid and/or your work load is reduced you will get more work from WCG. This is not uncommon with multiple projects and a larger cache.

See Work-Fetch Policy and Work Scheduler for detail information.

edit
You may want to consider reducing your cache a bit. As a rule of thumb use this formulae - Connection time = Work days cache wanted / number of projects attached. With broadband you only need 1 day in case both project go down. In my case I use the default .1 since I'm attached to so many projects my computer is never idle. The Boinc long term debt calculation ensure that eventually my resource share ratio will be honored.

---

Boinc V 7.4.36
Win7 i5 3.33G 4GB NVidia 470

Thanks for replying, Aurora. I was thinking that I probably don't have my computer running quite as much lately as I usually have, although probably not much less. But I don't think 6 days ago was any particular corner on that. 3 weeks ago maybe, but not 6 days.

I currently have 4 SETI work packages listed. The earliest deadline is July 3rd and the latest is July 9th. Adding up the time to complete, in order to do all of them by the earliest deadline, I would only have to have the 'puter on for about 3 hours a day. No problem! My WCG stats report my average daily computer running time as 7 hours a day and I doubt I'm running much below that on average.

The only other thing I can think of that might have confused the scheduler is occasionally I have suspended BOINC when I'm running a bunch of other things and want to reduce my CPU usage. The previous version I was running (5.2.13), before I just upgraded today, didn't restart itself on reboot (the way older versions did) so I would sometimes discover that it was still suspended days after I had suspended it. (I'm really glad to see that the new version "snoozes" for an hour at a time and then restarts itself. This was becoming a real problem.)

Could that have slowed down the rate of downloaded work, leaving WCG out in the cold somehow? I still find it curious -- since I have the two projects set to run, WCG:SETI at 2:1, I would have thought WCG would be the one left to run, not SETI. Or that I would still have more WCG packets waiting than SETI.


Starhugger

You may want to consider reducing your cache a bit. As a rule of thumb use this formulae - Connection time = Work days cache wanted / number of projects attached.

... and ignore the "/ number of projects" when the bug of the server side scheduler (sending more than needed) gets corrected :-)

Norbert

---

I think same here.

I have 6 projects but since 13/6/06 only BBC shows up in Tasks and only BBC is running.

I use boinc 5.4.9
NOT the bbc client.

---

As Aurora Borealis suggested, the situation has corrected itself in my case. It took a while for the old stuff to run off, but it seems to be back on track now. I have a bunch of work for both of my projects waiting to be crunched now. I guess it just gets ahead of itself sometimes, complicated by inconsistencies in how often it is able to run. Thanks AB!

Starhugger

still only BBC climate...

I suspended it once and boinc downloaded work for the others. after finishing that work. only BBC remains again.

and BBC has two times more work done than the other projects together.

Well if nothing helps I kick BBC project.

---

still only BBC climate...

I suspended it once and boinc downloaded work for the others. after finishing that work. only BBC remains again.

and BBC has two times more work done than the other projects together.

Well if nothing helps I kick BBC project.

You may be running in EDF mode. Boinc is very conservative when it comes to meeting deadlines and may have calculated that it may have problems if it continues to follow your project shares. This happens on multi project systems for short periods of time for other projects also, but is much more noticeable when climate prediction projects are in the mix. Since the deadlines are so long they can spend much more time in the mode before Boinc calculates it is safe to download more work from other projects and follow your share preferences. It will pay back the projects that have been short changed when it can safely do so.
Are you getting any messages about being (nearly) overcommitted?

---

Boinc V 7.4.36
Win7 i5 3.33G 4GB NVidia 470

You may be running in EDF mode. Boinc is very conservative when it comes to meeting deadlines and may have calculated that it may have problems if it continues to follow your project shares. This happens on multi project systems for short periods of time for other projects also, but is much more noticeable when climate prediction projects are in the mix. Since the deadlines are so long they can spend much more time in the mode before Boinc calculates it is safe to download more work from other projects and follow your share preferences. It will pay back the projects that have been short changed when it can safely do so.
Are you getting any messages about being (nearly) overcommitted?

In fact, wow I didnt see those log lines although I went through them several times. It seems like you said:

2006-07-15 09:04:35 [---] Using earliest-deadline-first scheduling because computer is overcommitted.
2006-07-15 09:04:35 [BBC Climate Change Experiment] Restarting task hadcm3ln_0bsi_00354456_1 using hadcm3l version 508
2006-07-15 09:04:35 [---] Suspending work fetch because computer is overcommitted.

I guess I just have to wait then.

edit: I found some explanations of being overcommitted.

I guess it is because the program worries if it can keep the deadline.
My BBC deadline is 30/03/2007 and I have 10% done in 300 hrs. That means 112 days of work left to finish. (but pc is not always running) So it's right to concentrate on BBC. Otherwise it wont finish early enough.

Thanks for ur help!

---