Not getting new work units

Message boards : Number crunching : Not getting new work units

To post messages, you must log in.

AuthorMessage
Profile Silverdrake
Avatar

Send message
Joined: 9 Oct 11
Posts: 10
Credit: 2,552,771
RAC: 0
Message 75481 - Posted: 26 Apr 2013, 10:48:15 UTC

My problem is that I have received no WUs at all for more than a week. I try manually updating and it mostly says it's neither reporting or requesting, but when it does request new work, I get 0 tasks. Other projects are running fine.
ID: 75481 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 75488 - Posted: 26 Apr 2013, 15:41:00 UTC
Last modified: 26 Apr 2013, 15:43:17 UTC

This would tend to mean either your machine is too busy with current tasks for other BOINC projects, or perhaps your current configuration is not allowing enough memory for tasks to run.

What are your resource shares with other projects on the same machine? What Operating System are you running? What BOINC version are you running?
Rosetta Moderator: Mod.Sense
ID: 75488 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Silverdrake
Avatar

Send message
Joined: 9 Oct 11
Posts: 10
Credit: 2,552,771
RAC: 0
Message 75492 - Posted: 26 Apr 2013, 22:24:32 UTC - in response to Message 75488.  
Last modified: 26 Apr 2013, 22:35:29 UTC

Computer data: https://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=1485667
Boinc verion 7.0.28
Resource share is evenly divided: https://boinc.bakerlab.org/rosetta/show_user.php?userid=432794
Boinc-allocated HD space: 20GB (currently using less than 100MB.)
Boinc can use 100% of processors, CPU and GPU while in the computer is in use, and 80% of CPU time.

I added Poem just the other day because I'd been getting nothing but Seti WUs (having same problem with SIMAP). The computer usually runs only two projects at a time, and it's been only the last few weeks that I've not been getting full loads of WUs from all (3 att) projects.
ID: 75492 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
mikey
Avatar

Send message
Joined: 5 Jan 06
Posts: 1895
Credit: 9,169,305
RAC: 3,857
Message 75499 - Posted: 27 Apr 2013, 11:12:10 UTC - in response to Message 75492.  

Computer data: https://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=1485667
Boinc verion 7.0.28
Resource share is evenly divided: https://boinc.bakerlab.org/rosetta/show_user.php?userid=432794
Boinc-allocated HD space: 20GB (currently using less than 100MB.)
Boinc can use 100% of processors, CPU and GPU while in the computer is in use, and 80% of CPU time.

I added Poem just the other day because I'd been getting nothing but Seti WUs (having same problem with SIMAP). The computer usually runs only two projects at a time, and it's been only the last few weeks that I've not been getting full loads of WUs from all (3 att) projects.


That's the problem...you added Poem and Boinc, on your pc, now thinks it OWES Poem crunching time so Rosetta is stopped until Poem 'catches up'. To get more Rosie units you could change the percentage settings, setting for example Rosie to 100%, Poem to 50% and Seti to 25%. That would mean Rosie gets twice as much crunching time as Poem which in turn would get twice as much crunching time as Seti, and Boinc would then figure out how to make that happen. Those numbers are changed on each projects website under your account and then computing preferences. Just give it a couple of weeks to balance itself out and then change the numbers as to your desires to help each project.
ID: 75499 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 75506 - Posted: 27 Apr 2013, 15:16:18 UTC

for example Rosie to 100%, Poem to 50% and Seti to 25%


Mikey means each respective resource share, not percentage. The percentage is then computed by dividing the resource share for a project in to the total of all of your attached project resource shares.
Rosetta Moderator: Mod.Sense
ID: 75506 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Silverdrake
Avatar

Send message
Joined: 9 Oct 11
Posts: 10
Credit: 2,552,771
RAC: 0
Message 75517 - Posted: 28 Apr 2013, 2:45:39 UTC - in response to Message 75506.  

Thanks, Mikey and Mod.Sense. I'll give that a try after I work through the scads of Seti WUs I've received (set it to NNT, already, wondering if Boinc was reading my CPU time as full because of them).

It's just that I added Poem AFTER I stopped getting any WUs from Rosetta (and SIMAP).
ID: 75517 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
mikey
Avatar

Send message
Joined: 5 Jan 06
Posts: 1895
Credit: 9,169,305
RAC: 3,857
Message 75523 - Posted: 28 Apr 2013, 11:05:01 UTC - in response to Message 75517.  

Thanks, Mikey and Mod.Sense. I'll give that a try after I work through the scads of Seti WUs I've received (set it to NNT, already, wondering if Boinc was reading my CPU time as full because of them).

It's just that I added Poem AFTER I stopped getting any WUs from Rosetta (and SIMAP).


Ahh you didn't say you had "scads" of Seti units, that could be the problem too, it 'owes' Seti crunching time so is trying to do that by not crunching Rosie units for a day or so. IF you just leave things alone it WILL fix itself, that could a week or more though depending on your particular settings. Everytime Seti goes down and you don't have enough of a cache of units to see you thru it, then when Seti comes back online you will 'owe' Seti some crunching time. Unfortunately you cannot set a project by project cache level, so it is a one size fits all thing, that can be difficult when each project has different agendas.
ID: 75523 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 75525 - Posted: 28 Apr 2013, 15:27:24 UTC

...wondering if Boinc was reading my CPU time as full...


Right, the BOINC Manager tries to only request work when it has some degree of confidence that the work it already has will be completed on time. It is trying to balance numerous priorities with your settings and your debts and your resource shares. To further complicate things, when you add another project the resource shares of the others are all proportionately reduced.

One simple way to set a project's debt back to zero is to mark it for no new work, crunch through and report back all of the work, then detach from the project and reattach.
Rosetta Moderator: Mod.Sense
ID: 75525 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Silverdrake
Avatar

Send message
Joined: 9 Oct 11
Posts: 10
Credit: 2,552,771
RAC: 0
Message 75542 - Posted: 29 Apr 2013, 21:33:28 UTC
Last modified: 29 Apr 2013, 21:33:59 UTC

Adjusting the resource shares to Rosetta and SIMAP having double those for Seti and Poem, and NNTing Seti, I got Rosetta and SIMAP units. Now it won't get any Seti or Poem units, though.

It glutted up on Seti units because I stopped getting any Rosetta or SIMAP WUs. I added Poem because I had nothing but Seti. I still have no idea why I stopped getting R and S WUs. The servers were running and showing available WUs.
ID: 75542 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
mikey
Avatar

Send message
Joined: 5 Jan 06
Posts: 1895
Credit: 9,169,305
RAC: 3,857
Message 75548 - Posted: 30 Apr 2013, 11:09:50 UTC - in response to Message 75542.  

Adjusting the resource shares to Rosetta and SIMAP having double those for Seti and Poem, and NNTing Seti, I got Rosetta and SIMAP units. Now it won't get any Seti or Poem units, though.

It glutted up on Seti units because I stopped getting any Rosetta or SIMAP WUs. I added Poem because I had nothing but Seti. I still have no idea why I stopped getting R and S WUs. The servers were running and showing available WUs.


Boinc has been around a VERY long time now and is extremely complicated in the way it tries to figure out where to get new units from. It is not just a click and get process, there is alot of formulas involved and everytime you add a new project all the old calculations it was working on prior to that go out the window and they all get redone, all at pc speeds of course. Basically if you have 2 projects each running at 100% priority and then add a 3rd project also at 100% priority then each project will go from an average of 50% to 33%, meaning the original projects don't need work right now because there is a new player in the game and it needs to get caught up. So even though units are available you won't get any because you have new priorities on YOUR pc.
ID: 75548 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Silverdrake
Avatar

Send message
Joined: 9 Oct 11
Posts: 10
Credit: 2,552,771
RAC: 0
Message 75576 - Posted: 7 May 2013, 3:28:38 UTC - in response to Message 75548.  
Last modified: 7 May 2013, 3:30:44 UTC

Boinc has been around a VERY long time now and is extremely complicated in the way it tries to figure out where to get new units from. It is not just a click and get process, there is alot of formulas involved and everytime you add a new project all the old calculations it was working on prior to that go out the window and they all get redone, all at pc speeds of course.

I get that it changes when I add new projects. But when I'd been happily crunching 3 projects for years and suddenly STOPPED getting any WUs from 2 of them, both of which showed available work, that is what I DON'T get.

I upgraded to Boinc 7.0.64 (after finishing and resetting everything).

I found this under Properties for each project:
Scheduling priorities
Rosetta -0.25
Poem -0.80
SIMAP -0.12
Seti -6.43

All minuses, with SIMAP as the least-so. Yet when it requested work, the log said:
5/6/2013 6:28:57 PM | boincsimap | Not requesting tasks: project is not highest priority
Even though SIMAP was set co-equal to Rosetta with the highest resource shares.

I'm just sitting here looking at those scheduling priorities and going ??????????????????! :-
ID: 75576 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
mikey
Avatar

Send message
Joined: 5 Jan 06
Posts: 1895
Credit: 9,169,305
RAC: 3,857
Message 75578 - Posted: 7 May 2013, 11:21:12 UTC - in response to Message 75576.  

Boinc has been around a VERY long time now and is extremely complicated in the way it tries to figure out where to get new units from. It is not just a click and get process, there is alot of formulas involved and everytime you add a new project all the old calculations it was working on prior to that go out the window and they all get redone, all at pc speeds of course.


I get that it changes when I add new projects. But when I'd been happily crunching 3 projects for years and suddenly STOPPED getting any WUs from 2 of them, both of which showed available work, that is what I DON'T get.

I upgraded to Boinc 7.0.64 (after finishing and resetting everything).

I found this under Properties for each project:
Scheduling priorities
Rosetta -0.25
Poem -0.80
SIMAP -0.12
Seti -6.43

All minuses, with SIMAP as the least-so. Yet when it requested work, the log said:
5/6/2013 6:28:57 PM | boincsimap | Not requesting tasks: project is not highest priority
Even though SIMAP was set co-equal to Rosetta with the highest resource shares.

I'm just sitting here looking at those scheduling priorities and going ??????????????????! :-


Look further down under properties and you will see Duration Correction Factor(DCF), this is where Boinc looks to see where to get the next batch of work from first. The project with the HIGHEST number should get more work then the lowest numbered projects, as it has the most difference from the number 1.0. If it can't get any work from that project it will go down the line until it finds a project that will send it work, changing the DCF as it does. You CAN manually change the numbers but it involves editing files, if you really want to do that I can help you but be aware ANY time you do that kind of stuff things can easily go from bad to worse! Boinc is designed to 'fix it' itself over time, it can take days, weeks or even months to do exactly what you want it to do, but eventually it will be pretty darn close.
ID: 75578 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Silverdrake
Avatar

Send message
Joined: 9 Oct 11
Posts: 10
Credit: 2,552,771
RAC: 0
Message 75584 - Posted: 8 May 2013, 19:17:43 UTC - in response to Message 75578.  

Look further down under properties and you will see Duration Correction Factor(DCF), this is where Boinc looks to see where to get the next batch of work from first. The project with the HIGHEST number should get more work then the lowest numbered projects, as it has the most difference from the number 1.0. If it can't get any work from that project it will go down the line until it finds a project that will send it work, changing the DCF as it does.

Now I'm even more confused. Rosetta's DCU is 0.9882. None of the others even have one listed.

It just reported all my SIMAP WUs completed, did the "not the highest priority" bit, again, and dled more Rosetta WUs when the ones I already have haven't even been started yet -- they're all due after the SIMAP and Poem units.

So did it decide that I've done enough SIMAP WUs for now, so it won't dl any more?
Why is Rosetta the only project that even has a DCU?
ID: 75584 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
mikey
Avatar

Send message
Joined: 5 Jan 06
Posts: 1895
Credit: 9,169,305
RAC: 3,857
Message 75589 - Posted: 9 May 2013, 11:13:45 UTC - in response to Message 75584.  

Look further down under properties and you will see Duration Correction Factor(DCF), this is where Boinc looks to see where to get the next batch of work from first. The project with the HIGHEST number should get more work then the lowest numbered projects, as it has the most difference from the number 1.0. If it can't get any work from that project it will go down the line until it finds a project that will send it work, changing the DCF as it does.

Now I'm even more confused. Rosetta's DCU is 0.9882. None of the others even have one listed.

It just reported all my SIMAP WUs completed, did the "not the highest priority" bit, again, and dled more Rosetta WUs when the ones I already have haven't even been started yet -- they're all due after the SIMAP and Poem units.

So did it decide that I've done enough SIMAP WUs for now, so it won't dl any more?
Why is Rosetta the only project that even has a DCU?


Without being there I can't say exactly, but one thing goes on is that the deadlines are NOT important until and unless you get too close to them and Boinc thinks you will miss them. THEN Boinc will switch to running tasks in high priority mode, normally units run in a first in, first out(fifo) order. Except that the time to swap between projects affects Boincs swapping between projects and will affect the fifo order.

All projects should have a dcf, duration correction factor, it is the bottom line under properties. I have 4 listed on this pc and all 4 have a number in that line.

I think part of your problem is that you are trying to run 4 projects on a dual core pc. Since you can't tell Boinc to run this project on this core and that project on that core, Boinc will often run the same project on both cores. This causes more problems the more projects you try to run on a single pc. My suggestion would be to pick 2 projects and sent the other 2 to no new tasks and when all the tasks are gone from those other 2 projects suspend them. Then run just 2 projects for 6 months, do the no new tasks thing again and then enable the other 2 projects and crunch them for 6 months. You will get the same result that you are trying for now, but your pc won't be as overwhelmed and confused with all the different deadlines, lack of workunits etc, etc. The other option would be to pick up a 2nd pc, a second hand one would be fine, and crunch the second 2 projects on it. Even if you don't know how to build a pc you can pick some used ones up for under 150 bucks complete with the OS on it. You can run it without a screen, keyboard or mouse once it is set up by using a free program like UltraVNC like I do. I have 15 pc's running here at home and couldn't live without it. Once setup with a password it basically lets you log into each pc remotely and it is like you are sitting in front of the pc controlling it.
ID: 75589 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
TJ

Send message
Joined: 29 Mar 09
Posts: 127
Credit: 4,799,890
RAC: 0
Message 75772 - Posted: 18 Jun 2013, 21:50:47 UTC
Last modified: 18 Jun 2013, 21:59:16 UTC

I am not getting new work from Rosie, while 7 cores of my 2 Xeons are idle.
Albert is running on the GPU with 0.5CPU.
What is happening?
Rosetta is set to accept new work, 6 cores are allowed to crunch, free memory and disk space enough.

Edit: I used the hard way, remove and connect again. It did the trick after 5 more minutes and manually requesting more work.
Greetings,
TJ.
ID: 75772 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
mikey
Avatar

Send message
Joined: 5 Jan 06
Posts: 1895
Credit: 9,169,305
RAC: 3,857
Message 75776 - Posted: 19 Jun 2013, 10:31:24 UTC - in response to Message 75772.  

I am not getting new work from Rosie, while 7 cores of my 2 Xeons are idle.
Albert is running on the GPU with 0.5CPU.
What is happening?
Rosetta is set to accept new work, 6 cores are allowed to crunch, free memory and disk space enough.

Edit: I used the hard way, remove and connect again. It did the trick after 5 more minutes and manually requesting more work.


I have done that too in the past, there is a bug somewhere!
ID: 75776 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Fredrik

Send message
Joined: 11 Jun 07
Posts: 1
Credit: 159,674
RAC: 0
Message 76260 - Posted: 10 Dec 2013, 10:39:12 UTC

Same here, I was not receiving any tasks but after modifying Tools > Computing Preferences > Disk Usage settings - increasing GB available to BOINC, then hitting "Update" work started to download.
ID: 76260 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : Not getting new work units



©2024 University of Washington
https://www.bakerlab.org