Impossible to get new work since 2 days, Normal or not?
log in

Advanced search

Message boards : Windows : Impossible to get new work since 2 days, Normal or not?

1 · 2 · Next
Author Message
The Brain QC
Send message
Joined: 7 Jun 10
Posts: 4
Credit: 13,110,703
RAC: 0
Message 8841 - Posted: 15 Jul 2010, 1:22:42 UTC

All is in the title, since two or three days i can't get any ne Wu from the server, is there a problem somewhere or my Boinc installation has problems.

My config if it can help :

Windows 7 Ult. x64
Core 2 Quad q6600
Cuda card Gtx 470
4Gb Ram

Profile Slicker
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 11 Jun 09
Posts: 2525
Credit: 740,580,099
RAC: 1
Message 8846 - Posted: 15 Jul 2010, 5:30:17 UTC

What other projects are you connected to?
What version of BOINC are you using?
Are you using the opt app?
Have you been able to get work in the past with that exact same setup?
When you turn on sched_op_debug in the cc_config.xml, is BOINC asking for work and not getting any from the server, or is it not asking for any?

Profile mikey
Avatar
Send message
Joined: 11 Aug 09
Posts: 3242
Credit: 1,688,902,610
RAC: 5,971,309
Message 8856 - Posted: 15 Jul 2010, 11:59:10 UTC - in response to Message 8841.

All is in the title, since two or three days i can't get any ne Wu from the server, is there a problem somewhere or my Boinc installation has problems.

My config if it can help :

Windows 7 Ult. x64
Core 2 Quad q6600
Cuda card Gtx 470
4Gb Ram


You do know that Collatz is having ISP issues right now? You seem to have plenty of units right now so I hope this is why.

The Brain QC
Send message
Joined: 7 Jun 10
Posts: 4
Credit: 13,110,703
RAC: 0
Message 8870 - Posted: 16 Jul 2010, 14:52:28 UTC

Thank you for answers, seems all is in order now.

S@NL-Johnny5
Send message
Joined: 31 Oct 09
Posts: 4
Credit: 8,554,491
RAC: 0
Message 8955 - Posted: 23 Jul 2010, 17:06:22 UTC

i have the same described problem. it's going on for a pretty long time now. by estimate it could be the same moment as the original poster.

win7 ultimate x64 with 4gb ram
core 2 duo E6600 running @ 3ghz
Nvidia GTX480 1,5gb

my account is configured to only get cuda work

here is the collatz related log:

23-7-2010 16:52:42 OS: Microsoft Windows 7: Ultimate x64 Edition, (06.01.7600.00)
23-7-2010 16:52:42 Memory: 4.00 GB physical, 8.00 GB virtual
23-7-2010 16:52:42 Disk: 119.14 GB total, 19.06 GB free
23-7-2010 16:52:43 NVIDIA GPU 0: GeForce GTX 480 (driver version 25721, CUDA version 3010, compute capability 2.0, 1503MB, 778 GFLOPS peak)
23-7-2010 18:57:25 Collatz Conjecture update requested by user
23-7-2010 18:57:28 Collatz Conjecture Sending scheduler request: Requested by user.
23-7-2010 18:57:28 Collatz Conjecture Not reporting or requesting tasks
23-7-2010 18:57:30 Collatz Conjecture Scheduler request completed

Profile Crunch3r
Volunteer moderator
Project developer
Project tester
Avatar
Send message
Joined: 30 Jun 09
Posts: 219
Credit: 7,515,635,101
RAC: 11,492
Message 8964 - Posted: 24 Jul 2010, 0:26:01 UTC - in response to Message 8955.

i have the same described problem. it's going on for a pretty long time now. by estimate it could be the same moment as the original poster.

win7 ultimate x64 with 4gb ram
core 2 duo E6600 running @ 3ghz
Nvidia GTX480 1,5gb

my account is configured to only get cuda work

here is the collatz related log:

23-7-2010 16:52:42 OS: Microsoft Windows 7: Ultimate x64 Edition, (06.01.7600.00)
23-7-2010 16:52:42 Memory: 4.00 GB physical, 8.00 GB virtual
23-7-2010 16:52:42 Disk: 119.14 GB total, 19.06 GB free
23-7-2010 16:52:43 NVIDIA GPU 0: GeForce GTX 480 (driver version 25721, CUDA version 3010, compute capability 2.0, 1503MB, 778 GFLOPS peak)
23-7-2010 18:57:25 Collatz Conjecture update requested by user
23-7-2010 18:57:28 Collatz Conjecture Sending scheduler request: Requested by user.
23-7-2010 18:57:28 Collatz Conjecture Not reporting or requesting tasks
23-7-2010 18:57:30 Collatz Conjecture Scheduler request completed


You should stop hitting that update button and simply let boinc do it's "thing"... Update does not trigger a work fetch... on the contrary, you'll increase the "back off" time by doing that.

____________

Team BOINC United.Join Science that matters.

S@NL-Johnny5
Send message
Joined: 31 Oct 09
Posts: 4
Credit: 8,554,491
RAC: 0
Message 8968 - Posted: 24 Jul 2010, 8:03:23 UTC - in response to Message 8964.


You should stop hitting that update button and simply let boinc do it's "thing"... Update does not trigger a work fetch... on the contrary, you'll increase the "back off" time by doing that.

okay.
and how long would you say it takes until boinc automatically requests for work? because i'm not hitting the button the second there is no work... only after i notice there is no work for several days.
shouldn't it trigger when there is no work at all in the task list? why would boinc not download any taks when it can do about 50 in a day?

Profile mikey
Avatar
Send message
Joined: 11 Aug 09
Posts: 3242
Credit: 1,688,902,610
RAC: 5,971,309
Message 8972 - Posted: 24 Jul 2010, 13:57:35 UTC - in response to Message 8968.


You should stop hitting that update button and simply let boinc do it's "thing"... Update does not trigger a work fetch... on the contrary, you'll increase the "back off" time by doing that.

okay.
and how long would you say it takes until boinc automatically requests for work? because i'm not hitting the button the second there is no work... only after i notice there is no work for several days.
shouldn't it trigger when there is no work at all in the task list? why would boinc not download any taks when it can do about 50 in a day?


Your problem is that you are running several projects and Boinc is trying to balance that along with your account settings of percentage of work for each project, etc, etc, etc. Boinc is a program that is trying to do its thing and we users keep fiddling with it, forcing it to do what we want it to do. I am as guilty as anyone, EVERYDAY I force an update on each of my machines, not exactly what Boinc wants, but it makes ME feel better!

S@NL-Johnny5
Send message
Joined: 31 Oct 09
Posts: 4
Credit: 8,554,491
RAC: 0
Message 8976 - Posted: 24 Jul 2010, 16:51:56 UTC - in response to Message 8972.

Your problem is that you are running several projects and Boinc is trying to balance that along with your account settings of percentage of work for each project, etc, etc, etc. Boinc is a program that is trying to do its thing and we users keep fiddling with it, forcing it to do what we want it to do. I am as guilty as anyone, EVERYDAY I force an update on each of my machines, not exactly what Boinc wants, but it makes ME feel better!

that would be likely.

nonetheless. when i disable all the other projects it would take work from the only one remaining, right?

Profile mikey
Avatar
Send message
Joined: 11 Aug 09
Posts: 3242
Credit: 1,688,902,610
RAC: 5,971,309
Message 8981 - Posted: 25 Jul 2010, 11:19:10 UTC - in response to Message 8976.

Your problem is that you are running several projects and Boinc is trying to balance that along with your account settings of percentage of work for each project, etc, etc, etc. Boinc is a program that is trying to do its thing and we users keep fiddling with it, forcing it to do what we want it to do. I am as guilty as anyone, EVERYDAY I force an update on each of my machines, not exactly what Boinc wants, but it makes ME feel better!

that would be likely.

nonetheless. when i disable all the other projects it would take work from the only one remaining, right?


Sort of, it also depends on your Long term Debt, or LTD. Boinc has a very complicated formula it uses t oget work when crunching for multiple projects. You can check out your LTD by using Notepad or something like it in Windows to VIEW your client_state.xml file. It is a hidden file so you have some trouble finding it but BE CAREFUL not to edit it while Boinc is running and even more careful to ONLY edit ONE LINE as a messed up file can cause Boinc to crash!!!!! My LTD line looks like this:
<long_term_debt>0.000000</long_term_debt>
you will have to look for it mine is in the Project section.

Mine is all zeros because I am only crunching for one project on this pc. What you CAN do is shut down Boinc and then edit the file changing the numbers to something like 1.000000, save the file and then when you restart Boinc, assuming all went well, all your projects will get work again. BUT if you are the least bit worried, LEAVE IT ALONE and Boinc WILL fix it on its own, it may just take awhile to do it. Boinc is designed to be a set it and forget it program, it can takes weeks or months to even itself out.

S@NL-Johnny5
Send message
Joined: 31 Oct 09
Posts: 4
Credit: 8,554,491
RAC: 0
Message 8982 - Posted: 25 Jul 2010, 12:22:32 UTC - in response to Message 8981.

ok thanx.
i'll stay away from the file. but at least i have a bit more understanding now.
for the most part i'm fine with anything on the pc, as long as i understand it.

Profile mikey
Avatar
Send message
Joined: 11 Aug 09
Posts: 3242
Credit: 1,688,902,610
RAC: 5,971,309
Message 8991 - Posted: 26 Jul 2010, 11:35:30 UTC - in response to Message 8982.

ok thanx.
i'll stay away from the file. but at least i have a bit more understanding now.
for the most part i'm fine with anything on the pc, as long as i understand it.


Okay that works

Profile Slicker
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 11 Jun 09
Posts: 2525
Credit: 740,580,099
RAC: 1
Message 8995 - Posted: 26 Jul 2010, 14:14:33 UTC

There is also a function built into the command line version that lets you reset or change the debts without having to stop and restart boinc or mess with the XML file.

boinccmd.exe --set_debts URL1 std1 ltd1 [URL2 std2 ltd2 ...]

where:
URL1 = universal resource locator of the project (e.g. the web address)
std1 = short term debt 1
ltd1 = long term debt 1

(I can live with the URL references, but why do Americans feel the need to make everything into an abbreviation or acronym? Can't they just learn to type?)

i.e.
boinccmd.exe --set_debts project_address short_term_debt_1 lone_term_debt_1 [...]

Anyway, an example would be:

boinccmd.exe --set_debts http://boinc.thesonntags.com/collatz/ 0.0 0.0

Normally, you would set the debts for ALL projects at once since the sum of all the short term debts should always be zero, and the sum of all the long term debts should be zero. They can be positive or negative. Negative infers you've got too many WUs. Positive infers you need more.

If you want to see what's happening, install BOINCView as it shows the debts in the projects tab.

Odd-Rod
Send message
Joined: 11 Jul 09
Posts: 528
Credit: 3,118,537
RAC: 241
Message 8997 - Posted: 26 Jul 2010, 15:09:00 UTC - in response to Message 8995.

If you want to see what's happening, install BOINCView as it shows the debts in the projects tab.

Indeed! I have learnt a lot about Boinc using Boincview. It's a pity it's no longer developed.

Profile mikey
Avatar
Send message
Joined: 11 Aug 09
Posts: 3242
Credit: 1,688,902,610
RAC: 5,971,309
Message 9008 - Posted: 27 Jul 2010, 11:14:50 UTC - in response to Message 8995.

There is also a function built into the command line version that lets you reset or change the debts without having to stop and restart boinc or mess with the XML file.

boinccmd.exe --set_debts URL1 std1 ltd1 [URL2 std2 ltd2 ...]

where:
URL1 = universal resource locator of the project (e.g. the web address)
std1 = short term debt 1
ltd1 = long term debt 1

(I can live with the URL references, but why do Americans feel the need to make everything into an abbreviation or acronym? Can't they just learn to type?)

i.e.
boinccmd.exe --set_debts project_address short_term_debt_1 lone_term_debt_1 [...]

Anyway, an example would be:

boinccmd.exe --set_debts http://boinc.thesonntags.com/collatz/ 0.0 0.0

Normally, you would set the debts for ALL projects at once since the sum of all the short term debts should always be zero, and the sum of all the long term debts should be zero. They can be positive or negative. Negative infers you've got too many WUs. Positive infers you need more.


Thanks for that Slicker!

[B^S] Elphidieus
Send message
Joined: 24 Jan 10
Posts: 77
Credit: 39,017,655
RAC: 0
Message 9032 - Posted: 30 Jul 2010, 10:51:34 UTC
Last modified: 30 Jul 2010, 10:59:20 UTC

I've just got a downgrade on GC from 5870 to 5770, and now I'm having problems getting new workunits...

My LTD is at 0.0, cache is set to 10 days, and I don't have another other projects running except Collatz, but it just refused to download more WUs...

Any ideas...?

Thanks...

Profile mikey
Avatar
Send message
Joined: 11 Aug 09
Posts: 3242
Credit: 1,688,902,610
RAC: 5,971,309
Message 9034 - Posted: 30 Jul 2010, 11:04:40 UTC - in response to Message 9032.

I've just got a downgrade on GC from 5870 to 5770, and now I'm having problems getting new workunits...

My LTD is at 0.0, cache is set to 10 days, and I don't have another other projects running except Collatz, but it just refused to download more WUs...

Any ideas...?

Thanks...


Did you try resetting the Project?

[B^S] Elphidieus
Send message
Joined: 24 Jan 10
Posts: 77
Credit: 39,017,655
RAC: 0
Message 9036 - Posted: 30 Jul 2010, 12:01:07 UTC - in response to Message 9034.
Last modified: 30 Jul 2010, 12:22:38 UTC

I found the damn problem...! My online preferences got messed up somehow...

And why must the dreadful 5-minute wait for every scheduler requests...? Every 5 minutes is equivalent of 750 credits on a 5870.

Profile Slicker
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 11 Jun 09
Posts: 2525
Credit: 740,580,099
RAC: 1
Message 9037 - Posted: 30 Jul 2010, 13:08:20 UTC - in response to Message 9036.

I found the damn problem...! My online preferences got messed up somehow...

And why must the dreadful 5-minute wait for every scheduler requests...? Every 5 minutes is equivalent of 750 credits on a 5870.


Computers that are not set up correctly will trash 150 WUs in under 3 seconds. If they can connect every 5 seconds instead of 5 minutes, that equates to 1,800 workunits trashed per minute. The normal server workload is 34 WUs per minute. When two or three people with bad setups do that at the same time, there will be no work for anyone since there is no way for the server to keep up with 5,400 WUs per minute. Once it starts getting backed up, then every machine asks more often for work making it worse and worse.

It is possible when the drivers are really screwed up or the machine is overclocked to the extreme, that a machine will actually think the results are valid when they are not. Put multiple GPUs in a machine with an invalid configuration and a single machine can bring down the entire project.

In just the past 24 hours, there have been 17,681 workunits error out. That works out to 11.6% of all WUs are trashed by improper setups. Add to that the 2.5% that never get returned (people abandon the project, detatch w/o aborting, or suspend it whenever MW comes back online) and that means one out of seven workunits have to be resent which creates even more work for the server.

So, while it is inconvenient for a few, it is for the benefit of all that the settings are the way they are.

Profile mikey
Avatar
Send message
Joined: 11 Aug 09
Posts: 3242
Credit: 1,688,902,610
RAC: 5,971,309
Message 9048 - Posted: 31 Jul 2010, 11:11:01 UTC - in response to Message 9037.

In just the past 24 hours, there have been 17,681 workunits error out. That works out to 11.6% of all WUs are trashed by improper setups. Add to that the 2.5% that never get returned (people abandon the project, detatch w/o aborting, or suspend it whenever MW comes back online) and that means one out of seven workunits have to be resent which creates even more work for the server.

So, while it is inconvenient for a few, it is for the benefit of all that the settings are the way they are.


So Slicker do you put those people that are in that 11.6% of us on a decreased workunit fetch schedule or do you let Boinc handle it? I guess what I'm asking is what do you do about people that consistently trash units? Do you send them an email telling them to check their setups, possibly giving them tips on what to check? Do you just reduce their total amount to some number sooner or later approaching zero? Or do you just ignore it and figure it is the cost of running a project? Inquiring minds are curious.

1 · 2 · Next
Post to thread

Message boards : Windows : Impossible to get new work since 2 days, Normal or not?


Main page · Your account · Message boards


Copyright © 2018 Jon Sonntag; All rights reserved.