| 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: 5
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I know there are WUs for Win, Linux and Android and also for 32 bits and 64 bits. So, it is not a simple thing to share the WU queue between the clients though, it is not impossible to share between the same kind of clients. In my current understanding, it is not possible to share the queue because the client ID is assigned to WUs at the timing the WUs are downloaded.
I wonder if the WUs could be processed not by a client bases but a group of clients. It increases the flexibility of processing for the crunchers with multiple client configuration like me. I have been suffered with such as H/W crush or power loss a machine for many times for last 10 years. Now, I am about to leave the crunching due to my financial reason though, please consider about WU queue sharing. To crunch all the WUs I have received until the end of June, I need about 8 hours. When all the WUs I have are crunched up, I will not enable the new WUs for the all of my clients until the issue I have to get solved. Please improve the processing environment until then. Thanks. Kiyo. |
||
|
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
You'd have to make a case over at the Berkeley developers [this goes to the core of the distribution system so it would be a very major rewrite]
|
||
|
|
KerSamson
Master Cruncher Switzerland Joined: Jan 29, 2007 Post Count: 1684 Status: Offline Project Badges:
|
Hi Sekimoto,
----------------------------------------I agree with SekeRob, the functionality you mentioned has a major impact on the current boinc design. One possible workaround would be to cluster the hosts, requiring that CPU architecture and operating system are identical for all the clustered hosts. Many years ago, it was a boinc cluster approach based on Beowulf; I don't know what the current situation is. However, a new request needs some use cases for both enabling a design review and justifying implementation effort. In your specific case, I cannot see the related use case. Even if one or more use cases could be identified, the next step is to consider alternative approaches, for example: clustering. A redesign or an enhancement of the current implementation is only justified if the use case is valid AND if no alternative solution could be considered. At this time, WCG and Boinc provide already large flexibility for queue and host management. It is now your challenge to describe why queue sharing could be useful ![]() Cheers, Yves ---------------------------------------- [Edit 1 times, last edit by KerSamson at Jul 3, 2016 4:07:52 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks Rob and Samson.
Actually, it is not a big issue for me 'cause I am leaving in a couple of minute from the WCG. However, I think it is far better than current design. Just now, the last WU has been reported and my crunching has ended up. Thank you for your support and bye for a while ( I don't even know when I can come back though,,,). Good luck to WCG. Kiyo. |
||
|
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
knreed mentioned in another thread how a 'big' member was using Condor grid distribution, which I understand as being a front end that is programmed to queue and distribute tasks to systems that have time slices available within a pool of diverse devices. Google to learn more, when you're back from intermission.
|
||
|
|
|