Setting Boinc to NOT download WUs and it removes the instruction
18 Dec 2022 17:17:42 UTC
Topic 228744
(moderation:
Odd problem here. I run a few projects from ClimatePrediction to WorldCommunityGrid as well as Einstein@home.
Windows laptop running Windows 11 22H2 and Boinc 7.20.2 with an Intel UHD Graphics 620 display adapter.
Every now and then I want to pilot what I'm crunching and when I set Einstein@Home to "Won't Get New Tasks" it stays that way for a bit and then removes the command and starts fetching new tasks again, even when WorldCommunity Grid is quite capable of providing tasks. It is a bit of an annoyance rather than serious, but I don't understand it.
On other laptops the command to not fetch new tasks works fine.
The task it has fetched are GPU related tasks I think:
Gamma-ray pulsar binary search #1 on GPUs 1.22 (FGRPopencl-intel_gpu)
Does anyone have any thoughts about this?
Is it a Boinc problem?
Thank you for any insight you can give.
The Real Weasle, How long
)
The Real Weasle,
How long is "a bit"?
You may want to set all tasks (CP, WCG, and E@H) to " No New Tasks " and wait for your predetermined "a bit" and see if E@H again restarts by itself. I don't believe this is a BOINC problem, more likely a Window11 problem.
Proud member of the Old Farts Association
Good advice there... Thank
)
Good advice there... Thank you. I will disable everything and see what happens.
I was actually running down a queue of Work Units so it might take a day to find out when it next needs work. I will report back.
The Real Weasle wrote: Good
)
Well, keep us posted. ;>)
Proud member of the Old Farts Association
The Real Weasle wrote: Odd
)
I have seen this before but only when every Project is set to No New Tasks then an odd Project or two will start sending tasks even though EVERY Project is set to No New Tasks. The only way I've found to stop it is to Suspend every Project then it stops getting tasks from any Project. But I've never seen it when at least one Project is allowed to get new tasks, even if none are available.
Personally it sounds like you found a bug in the software that happens sooo rarely it's mostly gone unreported, even by me.
This kind of error would have
)
This kind of error would have to be documented by the BOINC client emulator website so the David Anderson could review all the logs and see exactly what BOINC was doing.
You can start a simulation and see if you can reproduce it and then call the BOINC devs to its attention.
BOINC Client Emulator
He is running a new enough client to have the issues with max_concurrent and continuous work_fetch issue fix in place. These issue commonly occurred with several 0 resource share backup projects.
The client can get very confused often when it does its rr_simulation runs before each scheduler contact.
Keith Myers wrote: This kind
)
hmm very interesting
Okay, this problem has been
)
Okay, this problem has been going on sometime- I'm talking 6 - 7 months at least.
I stopped ALL projects from fetching tasks and it wound down to I think no Einstein tasks and one or possibly 2 WorldCommunity Grid and it did this to me: I'm wondering whether Boincstats has anything to do with it.
It unlocked WorldCommunityGrid and tried to fetch tasks from there, and finding none it went to Einstein.
I've re-blocked all tasks to not download and will see what happens again.
A bit later this evening I will look into the Emulator. I'm sorry of handy with computers, but things ran away with too much progress so I'm not much more than a simple user now.
EDIT - I don't think the lines about WCG having changed its URL is anything. It's just a problem they have at the moment.
19/12/2022 15:39:43 | World Community Grid | Finished upload of OPN1_0126709_02408_2_r630757638_0
19/12/2022 15:39:43 | World Community Grid | [file_xfer] Throughput 89591 bytes/sec
19/12/2022 15:39:45 | World Community Grid | Sending scheduler request: To report completed tasks.
19/12/2022 15:39:45 | World Community Grid | Reporting 1 completed tasks
19/12/2022 15:39:45 | World Community Grid | Not requesting tasks: "no new tasks" requested via Manager
19/12/2022 15:39:46 | World Community Grid | Scheduler request completed
19/12/2022 15:39:46 | World Community Grid | This project seems to have changed its URL. When convenient, remove the project, then add http://www.worldcommunitygrid.org/
19/12/2022 15:39:46 | World Community Grid | Project requested delay of 121 seconds
19/12/2022 16:08:31 | | Contacting account manager at https://bam.boincstats.com/
19/12/2022 16:08:33 | | Account manager: BAM! User: 18227, The Real Weasle
19/12/2022 16:08:33 | | Account manager: BAM! Host: 873224
19/12/2022 16:08:33 | | Account manager: Number of BAM! connections for this host: 1313
19/12/2022 16:08:33 | | Account manager: Dear founder of team Centro Meteorologico Lombardo - Italy, your team is invited to participate in the team challenges (http://boincstats.com/en/bam/teamChallenge/)
19/12/2022 16:08:33 | | Account manager contact succeeded
19/12/2022 16:08:34 | World Community Grid | Sending scheduler request: To fetch work.
19/12/2022 16:08:34 | World Community Grid | Requesting new tasks for CPU and Intel GPU
19/12/2022 16:08:35 | World Community Grid | Scheduler request completed: got 0 new tasks
19/12/2022 16:08:35 | World Community Grid | This project seems to have changed its URL. When convenient, remove the project, then add http://www.worldcommunitygrid.org/
19/12/2022 16:08:35 | World Community Grid | No tasks sent
19/12/2022 16:08:35 | World Community Grid | No tasks are available for OpenPandemics - COVID 19
19/12/2022 16:08:35 | World Community Grid | No tasks are available for OpenPandemics - COVID-19 - GPU
19/12/2022 16:08:35 | World Community Grid | No tasks are available for Help Stop TB
19/12/2022 16:08:35 | World Community Grid | No tasks are available for the applications you have selected.
19/12/2022 16:08:35 | World Community Grid | Project requested delay of 121 seconds
19/12/2022 16:08:40 | Einstein@Home | Sending scheduler request: To fetch work.
19/12/2022 16:08:40 | Einstein@Home | Requesting new tasks for CPU and Intel GPU
19/12/2022 16:08:42 | Einstein@Home | Scheduler request completed: got 9 new tasks
19/12/2022 16:08:42 | Einstein@Home | BOINC will delete file stochastic_fullG.bank (no longer needed)
19/12/2022 16:08:42 | Einstein@Home | Project requested delay of 60 seconds
19/12/2022 16:08:42 | Einstein@Home | [file_xfer_debug] Got server request to delete file stochastic_fullG.bank
19/12/2022 16:08:44 | Einstein@Home | Started download of p2030.20200618.G53.47-03.85.S.b2s0g0.00000_2440.bin4
19/12/2022 16:08:44 | Einstein@Home | [file_xfer] URL: http://einstein6.aei.uni-hannover.de/EinsteinAtHome/download/24b/p2030.20200618.G53.47-03.85.S.b2s0g0.00000_2440.bin4
19/12/2022 16:08:44 | Einstein@Home | Started download of p2030.20200618.G53.47-03.85.S.b2s0g0.00000_2441.bin4
19/12/2022 16:08:44 | Einstein@Home | [file_xfer] URL: http://einstein6.aei.uni-hannover.de/EinsteinAtHome/download/18e/p2030.20200618.G53.47-03.85.S.b2s0g0.00000_2441.bin4
19/12/2022 16:08:45 | Einstein@Home | [file_xfer] http op done; retval 0 (Success)
19/12/2022 16:08:45 | Einstein@Home | [file_xfer] http op done; retval 0 (Success)
19/12/2022 16:08:45 | Einstein@Home | [file_xfer] file transfer status 0 (Success)
19/12/2022 16:08:45 | Einstein@Home | Finished download of p2030.20200618.G53.47-03.85.S.b2s0g0.00000_2440.bin4
Well account managers add
)
Well account managers add another layer on top of the BOINC Client perturbing its actions.
I would remove all account managers first and see if the problem changes.
The Real Weasle wrote:Okay,
)
Try this, just for kicks.
Don't suspend Einstein, but just have no new tasks set. Suspend all other projects and then see what happens.
If you don't get Einstein to actively start running again, then it is not BOINC that has the problem, but one of the other projects overriding Einstein's settings. If it does start up Einstein again with the other projects suspended, then I definitely think you should do as Keith suggests.
Proud member of the Old Farts Association
BOINC users always forget an
)
BOINC users always forget an important BOINC feature and that is the REC or Recent Estimated Credit mechanism or algorithm that always runs in the client.
It tries to balance the awarded credit averages across all your projects. So when you suspend a project for a while against your other projects that continue to run, when you restart your suspended project, that project owes a "credit debt" against your current running projects.
So the client will fetch as much work as allowed by your cache levels and immediately start to run that work to pay off the credit debt as soon as possible.
Only when the debt is paid will the project observe your project resource allotments.
What you can do to shorten the payment debt payoff period is to change the default client averaging period for REC in the cc_config.xml file.
Change this parameter down in the Options section from:
<rec_half_life_days>10.000000</rec_half_life_days> to:
<rec_half_life_days>1.000000</rec_half_life_days>
This will pay off the credit debt much sooner because it is averaging all projects credit accumulations over a single day instead of over 10 days.