Download result-files for offline-processing?
25 Feb 2005 14:02:22 UTC
Topic 187990
(moderation:
Is there some way that i can download result-files for offline-processing? I got access to several big systems with lots of spare cpu-cycles which are not connected to anything which might look like the internet.
Ideas, anyone?
Cheer, Alex.
"I am tired of all this sort of thing called science here... We have spent
millions in that sort of thing for the last few years, and it is time it
should be stopped."
-- Simon Cameron, U.S. Senator, on the Smithsonian Institute, 1901.
Download result-files for offline-processing?
)
As far as I know the validator checks that the result returned came from the same host it was sent to so this wont be that easy to accomplish.
You may be able to it as follows:
Install BOINC on a machine with net access and download whatever units you think you might need (use preferences to get extra units if you want). Call this your 'router' host. You would then have to install BOINC on the offline server.. probobly matching the install on your router machine ie what directory it is installed on.
ONce you have that done, copy all the xml and project data into the offline servre's BOINC folder, startup BOINC on that box and let it rip!
You would need to reverse this process to get the results back to the BOINC servers. ie copy all project data and XML files back to the router PC. Then use the router machine to upload and report all the results to the servers.
I would assume BOINC will need to be fully shut down (not paused!) on both machines when moving the files.
Your 'router' machine may not be able to crunch at the same time... I don't know how it would work if you also wanted that one crunching work! things would get very confusing!
I will stress that I have never done this, I don't know if it will work and I don't know if the results produced will be OK!
... let us know if it works though! or if you have a better solution.
Wanna visit BOINC Synergy site? Click my stats!
Join BOINC Synergy Team
Paul, thanks for the
)
Paul,
thanks for the advice, i will definitively try it out - 8 machines with 32 CPUs each are (still) idle and quite hungry ;-) It would be a real waste and quite heretic if i wouldn't let 'em do what they really want... *chrunch* *munch*
I will keep you informed and tell you if it worked somehow.
Cheers, Alex.
"I am tired of all this sort of thing called science here... We have spent
millions in that sort of thing for the last few years, and it is time it
should be stopped."
-- Simon Cameron, U.S. Senator, on the Smithsonian Institute, 1901.
The important files in
)
The important files in general are the *.xml files in the BOINC directory, the projects/einstein.phys.uwm.edu and the slots subdirectories. Most of the files in the project directory are static, "sticky" data files that almost never change and thus only need to be transfered once. In particular this applies to the App executable, "sun", "earth" and the 11MB detector data file. Look at the timestamps to see which files need to be updated. And yes, make sure you shut down BOINC before playing with the files.
I suggest that you set up a separate BOINC directory on the internet machine for every off-line machine you want to run and learn how to use the command-line version of the client boinc_cli.exe which should keep its files in the directory where it's run in.
The server should only accept a Result from the machine it sent it to. However it doesn't care about where the calculation is actually done. But he will, of course, identify them with the benchmarks of the machine he contacted, while the cpu time is counted on the calculating machine. So don't be surprised if the claimed credit is way off in either direction.
Let us know whether it works.
BM
BM