Message boards :
Number crunching :
Requesting 0 seconds of work
Message board moderation
Author | Message |
---|---|
Send message Joined: 6 Apr 08 Posts: 2018 Credit: 100,142,856 RAC: 0 |
I have MilkyWay set up with a resource share of 1000 (90.91%) and climateprediction (my 'backup' when MW sends no WUs) at 100 (9.09%). The trouble is, on one of my PCS, it won't request any MW work. 01/01/2009 11:08:49|Milkyway@home|Sending scheduler request: Requested by user. Requesting 0 seconds of work, reporting 14 completed tasks But when I suspend climateprediction, it does; 01/01/2009 11:09:09|Milkyway@home|Sending scheduler request: Requested by user. Requesting 1734956 seconds of work, reporting 0 completed tasks Any ideas why MW won't play with climateprediction? |
Send message Joined: 12 Nov 07 Posts: 2425 Credit: 524,164 RAC: 0 |
Is your debt off in Boinc? Doesn't expecting the unexpected make the unexpected the expected? If it makes sense, DON'T do it. |
Send message Joined: 6 Apr 08 Posts: 2018 Credit: 100,142,856 RAC: 0 |
|
Send message Joined: 27 Aug 07 Posts: 647 Credit: 27,592,547 RAC: 0 |
Maybe you should suspend your CPDN model shortly and then MW will fetch work again. ;-) PS. Happy New Year, btw! :-))) Lovely greetings, Cori |
Send message Joined: 12 Nov 07 Posts: 2425 Credit: 524,164 RAC: 0 |
Do you have the same settings on your other computers? If so then it would be something else. Long term debt is part of what allows which project to get work when it is owed work. It is located in the Boinc folder called client state, open with notepad (or such), search for 'long', you will see numbers in the middle of the line. Large + numbers means that project has ran a lot and owes the others time, - numbers means that project hasn't ran much and is owed more time. It can be changed, but boinc needs to be closed first. Doesn't expecting the unexpected make the unexpected the expected? If it makes sense, DON'T do it. |
Send message Joined: 12 Nov 07 Posts: 2425 Credit: 524,164 RAC: 0 |
@ Ice, in your sig. that's the One ring right? Doesn't expecting the unexpected make the unexpected the expected? If it makes sense, DON'T do it. |
Send message Joined: 7 Jun 08 Posts: 464 Credit: 56,639,936 RAC: 0 |
I have MilkyWay set up with a resource share of 1000 (90.91%) and climateprediction (my 'backup' when MW sends no WUs) at 100 (9.09%). This is normal, and the reason is that the Resource Share affects not only what your host will do when it comes time to fetch new work, but also sets the actual time slices each project gets for execution (both the length and frequency). With such a high bias towards MW, this has most likely put the CPDN task into a schedule bind, and thus the CC will give it priority until it gets back into it's deadline window. As a result, MW will get cut off from DL'ing new work until the CC drops out of HP/EDF even though it is the primary. The other advice for forcing it to DL new work for MW works, but only makes the situation worse overall. The bottom line is the best way to run high bias Resource Shares is to just leave the CC alone and let it take care of things on its own (IMHO). <edit> BTW, given that CPDN has such a long deadline, it also requires a relatively long runtime for all hosts compared to other projects. So even though this would seem to make it a good choice for a backup project, this is not the case and is probably the worst one to choose. You only need to stop and think about what you are observing now, and then consider what will happen in the case where you have no backup task onboard, and shouldn't have one to to very large negative LTD, and then your primary goes offline long enough to open up sufficient overall cache slack to require new work to refill it. The moral of the story here is that for High Bias backup scenarios, you want to choose a project with relatively short running work and a relatively loose deadline tightness factor compared to the primary. Alinator |
Send message Joined: 6 Apr 08 Posts: 2018 Credit: 100,142,856 RAC: 0 |
|
Send message Joined: 12 Nov 07 Posts: 2425 Credit: 524,164 RAC: 0 |
The one that links to www.ic3.me Doesn't expecting the unexpected make the unexpected the expected? If it makes sense, DON'T do it. |
Send message Joined: 6 Apr 08 Posts: 2018 Credit: 100,142,856 RAC: 0 |
Thanks for the advice. I actually want to run CPDN as an extra project rather than just a backup, so will take Cori's advice and occassionally suspend CPDN to get MW to request work. I'm not too bothered about it since it is only happeneing on the one pc, and once the CPDN WU has finally done I will probably delete the Boinc installtion and start with a fresh one. Thanks again banditwolf, Alinator and Cori. |
Send message Joined: 6 Apr 08 Posts: 2018 Credit: 100,142,856 RAC: 0 |
|
Send message Joined: 12 Nov 07 Posts: 2425 Credit: 524,164 RAC: 0 |
Thanks for the advice. I actually want to run CPDN as an extra project rather than just a backup, so will take Cori's advice and occassionally suspend CPDN to get MW to request work. Or put 'no new work' on one and then 'allow work' when you want to run the other. Thats how I go between MW and Rosetta. I haven't liked how boinc switches and choose to do it myself, since I only use one pc. Doesn't expecting the unexpected make the unexpected the expected? If it makes sense, DON'T do it. |
Send message Joined: 7 Jun 08 Posts: 464 Credit: 56,639,936 RAC: 0 |
Thanks for the advice. I actually want to run CPDN as an extra project rather than just a backup, so will take Cori's advice and occassionally suspend CPDN to get MW to request work. Agreed. Whereas the CC scheduling and work fetch policies work reasonably well in most 'mainstream' setups, the options you have available for when you want to tailor the behaviour to unusual or special circumstances leave a lot to be desired! ;-) Call me crazy, but I want my cake and eat it too. :-) Alinator |
Send message Joined: 12 Apr 08 Posts: 621 Credit: 161,934,067 RAC: 0 |
Whereas the CC scheduling and work fetch policies work reasonably well in most 'mainstream' setups, the options you have available for when you want to tailor the behaviour to unusual or special circumstances leave a lot to be desired! ;-) I won't call you crazy ... but ... Dr. Anderson says no ... as always ... Though with version 6 something you can do multiple selects and that allows some better control over the way things can be run if you don't mind micromanaging ... |
©2024 Astroinformatics Group