| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: Best way to move work to another machine without losing progress (linux)? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 9
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I'm looking at getting WCG running on Slicify. For those of you who don't know us, you can think of it like the Amazon EC2 Spot Market - a market for temporary cloud processing.
The main issue I can see is that our machines are only available for typically a few hours at a time. Ideally I'd like to be able to set up a script so that once an hour or so, it backs up the calculations in progress to cloud storage. That way you could just fire up another instance, download the work done so far, and carry on from where you left off. What is the best way to go about this? If I just backup the directory while boinc is running is this going to corrupt the files? Is there a graceful way to terminate boinc and have it save state, and then resume after the backup? thanks Steve |
||
|
|
twilyth
Master Cruncher US Joined: Mar 30, 2007 Post Count: 2130 Status: Offline Project Badges: 
|
I'm just taking a shot in the dark here but you might be able to accomplish this just by taking the BOINC directory from the old Program Data directory (normally hidden) and overlaying it on to the new one.
----------------------------------------However there might be some hardware specific information in there so some tweaking might be required but others here have done this manually so they should be able to provide better guidance.   |
||
|
|
deltavee
Ace Cruncher Texas Hill Country Joined: Nov 17, 2004 Post Count: 4894 Status: Offline Project Badges: 
|
I'm not sure what you are proposing is possible. What some people do is download workunits onto machine A, move them to machine B for processing and then move them back to machine A for uploading. I think that if you upload them from machine B there will be problems. If I remember correctly there was a time that some crunchers would try to copy the same completed workunit to multiple machines for uploading in order to cheat on points. The safeguards in place to prevent that may also prevent you from doing what you propose. Of course I could be wrong about this so good luck on your efforts.
----------------------------------------[Edit 2 times, last edit by deltavee at Mar 8, 2014 8:31:58 PM] |
||
|
|
cowtipperbs
Advanced Cruncher Joined: Aug 24, 2009 Post Count: 78 Status: Offline Project Badges:
|
Try looking at this thread. I know its about Windows, but might work.
----------------------------------------https://secure.worldcommunitygrid.org/forums/...ead,35016_offset,0#418912  |
||
|
|
twilyth
Master Cruncher US Joined: Mar 30, 2007 Post Count: 2130 Status: Offline Project Badges:
|
I don't want to put words in the OP's mouth, but my guess is that he/she is mainly interested in transferring incomplete wu's.
----------------------------------------I'm not sure how it works in Slicify when you get kicked of a machine, but I'm assuming that you have time to do an orderly shutdown. If so, it should be possible to force uploading and reporting of the completed wu's. If the second machine actually completes the work on a wu, hopefullyl the server wouldn't think it's questionable, but I wouldn't be surprised if it did - as you say. It's also possible that the software itself would notice the change and shut you down even before you got to the server. However I did a search on 'boinc moving work units' and found some hits that look promising. |
||
|
|
deltavee
Ace Cruncher Texas Hill Country Joined: Nov 17, 2004 Post Count: 4894 Status: Offline Project Badges:
|
On Slicify you get kicked off the server as soon as someone offers more for it's use. All your files are deleted and gone. Poof.
https://ripple.com/forum/viewtopic.php?f=18&t=5580 |
||
|
|
twilyth
Master Cruncher US Joined: Mar 30, 2007 Post Count: 2130 Status: Offline Project Badges:
|
Really? I had no idea. It seems strange that you couldn't reserve some minimum amount of time but I guess they have their reasons for doing it that way. Thanks.
---------------------------------------- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks for all the replies and links.
Some of the info I found is quite old (6-8 years old), so difficult to know whether things have changed in the meantime. Looks like I'll just need to do some more investigation and try it out. Will post later if I get anything to work successfully thanks Steve |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7846 Status: Offline Project Badges:
|
I'm not sure what you are proposing is possible. What some people do is download workunits onto machine A, move them to machine B for processing and then move them back to machine A for uploading. I think that if you upload them from machine B there will be problems. If I remember correctly there was a time that some crunchers would try to copy the same completed workunit to multiple machines for uploading in order to cheat on points. The safeguards in place to prevent that may also prevent you from doing what you propose. Of course I could be wrong about this so good luck on your efforts. I agree with deltavee. I seem to remember there is a machine ID or something similar attached to the WU once it has been started. At any rate it seems like a management nightmare to keep track of all this on multiple machines. Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
|