| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
| World Community Grid Forums
|
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 7
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Can one do a sneaker net with BOINC on a machine with no internet connection? (With all that BOINC's RCP "self-communication" is it even possible?)
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
You really do have the RPC stuff totally wrong.
Yes, BOINC can be run as a sneaker-net, but it's not particularly easy. I don't recommend it. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Didactylos:
Yes, BOINC can be run as a sneaker-net, but it's not particularly easy. Is it done the same way sneaker net is done with the UD Agent? That is: shut down BOINC; create a different folder, say WCG-SneakerNet-1; install BOINC there; connect; download; shut down the sneaker net version; copy to portable USB drive; copy to old machine; start original BOINC on the machine with internet connection; start sneaker net BOINC on older machine; when work unit (?? job ??) is finished on older sneaker net machine -- shut it down; copy folder to portable USB drive; stop BOINC on the machine with internet connection; copy sneaker net folder to the machine with internet connection; start sneaker net BOINC; upload/download; stop sneaker net BOINC; copy to portable USB drive; copy to old machine; start original BOINC on the machine with internet connection; start sneaker net BOINC on older machine; repeat; and voila you're in business? If not, how is it done? |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
That ought to work, perhaps with a few refinements such as suspending computation on the connected host.
BUT: the two computers must be compatible. If they are not, then either it will fail altogether (the work will be aborted by the disconnected host) or the results will always be invalid. What constitutes "compatible" is hard to say. At the least, the computers must both have the same CPU manufacturer. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thank you.
At the least, the computers must both have the same CPU manufacturer. Cr@p! ![]() |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7848 Status: Offline Project Badges:
|
My best advice is to try it. Watch your results. If you get all or mostly invalids quit the experiment because you will be wasting time. However, if all or most of them turn out valid go for it. Good luck.
----------------------------------------Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
Protego
Cruncher Joined: Apr 26, 2007 Post Count: 33 Status: Offline |
I have tried a different approach.
----------------------------------------When a WU is not in progress, so it is either fully completed or not started yet, it is very small compared to when it is in progress. So I have manually removed some work units from the Internet machine by copying the relevant files and then manually removed the enties from the configuration files inside BOINC. This works OK, but the BOINC at start up check the queue, and the project server then re-fill the Internet machine with the missing work units. Here we find the first problem: The project server keep track of your queue of not started WU:s, and will resend these same WU:s to you. I, of course, would prefer new (different) WU:s ! I solve this by setting selected WU:s to "Pause this WU". Then we find the next problem: The BOINC and the Server cannot agree upon when to send new WU:s to this machine, if some WU:s are on Pause. So when non-pause WU:s are completed, no new WU:s are sent, even if there is still time left to complete both some new WU:s and the set that are on Pause. Installing and running the WU:s on the off-line machine is straitforward, but it get kind of tricky keeping a work queue filled also on this machine. The reson is that the life span of the FA@H WU:s is about nine days, much too short for this kind of manual upload/download. In fact both macines will likely go empty after a while, due to that the project server simply refuse to send more work. A good thing is that when the Internet machine have been refilled with the completed WU:s, you can just leave it connected, and it will itself report the WU:s and obtain more work. You don't have to sit in fromt of the machine and .... waiting for 00:05:01 ... (not reporing or requesting more work) .... Cannot complete in time as your machine is online only 7.5654% of the time and ... waiting for 00:02:01 .... and so on!! Currenly I don't run this, I cannot find the time and effort to update every 2-3 days. I can update every 10-14 days, but then the WU:s would be expired. So the off-line machines are empty. .. And now my 96 MB RAM machine, that run Win98, is no longer accepted either, as, if you run XP, 96 MB is not enouth to run FA@H. So now that one is also empty. ![]() ![]() |
||
|
|
|