| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: The Clean Energy Project - Phase 2 Forum
Thread: CEP2: And wasted another 12 hours of computing time! |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 12
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
E203199_ 864_ C.28.C22H13N3OSSi.00212913.4.set1d06_ 2-- 640 Valid 9/25/11 07:54:20 9/27/11 06:45:17 12.00 322.7 / 117.1
E203199_ 864_ C.28.C22H13N3OSSi.00212913.4.set1d06_ 0-- 640 Valid 9/15/11 08:08:28 9/20/11 12:04:03 12.00 79.1 / 95.1 E203199_ 864_ C.28.C22H13N3OSSi.00212913.4.set1d06_ 1-- 640 Valid 9/15/11 07:46:44 9/26/11 13:14:25 12.00 119.8 / 102.5 The 'No Reply' decided it was time to turn up more than a 30 hours late, whilst my computer was acting the repair man, crunching off-line and doing the job within 48 hours. In future, all 'No Reply' repairs will be put in suspend for 36 hours and allow the servers to abort them should the grace rule allow these 'Too Late' to still show. To think that mine was nearly done with all 16 jobs and the first 2 not getting 12 jobs done. What a waste! --//-- |
||
|
|
anhhai
Veteran Cruncher Joined: Mar 22, 2005 Post Count: 839 Status: Offline Project Badges: 
|
Sekerob, couldn't you suggest to the "powers that be" to change the policy about sending out repair jobs? Something like have the server wait 24 hrs after the no reply deadline before sending out a repair job? I personally always thought it was silly to send out repair jobs right away like that.
---------------------------------------- |
||
|
|
toss
Senior Cruncher New Zealand Joined: Jan 3, 2007 Post Count: 220 Status: Offline Project Badges: 
|
Sekerob, couldn't you suggest to the "powers that be" to change the policy about sending out repair jobs? Something like have the server wait 24 hrs after the no reply deadline before sending out a repair job? I personally always thought it was silly to send out repair jobs right away like that. I have long thought a 10 day cache to be over the top. If the max cache were limited to say 8 days but the 10 day reporting time retained and strictly imposed, then I think there would be a big reduction in make up work required and Sek's example would be resolved. Cheers |
||
|
|
anhhai
Veteran Cruncher Joined: Mar 22, 2005 Post Count: 839 Status: Offline Project Badges:
|
toss, that is not the only cause of the late reply. people take vacation and shut off their systems. When they turn it back on, boinc will abort WUs that are already late if they haven't been started. But if they have, they will just continue at normal.
---------------------------------------- |
||
|
|
toss
Senior Cruncher New Zealand Joined: Jan 3, 2007 Post Count: 220 Status: Offline Project Badges:
|
Agreed anhhai. There are numerous late causes. Despite running small cache's I have caused a few WU's to time out in the past as a result of machine failure.
However I am aware a lot of people run max cache and my proposition would still have a significant impact I think. |
||
|
|
rilian
Veteran Cruncher Ukraine - we rule! Joined: Jun 17, 2007 Post Count: 1460 Status: Offline Project Badges: 
|
Since project does not need completed WU ASAP, 24 or N hours wait before resend no-reply job is really wise idea! Imagine how many CPU-years this can save on the GRID
---------------------------------------- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Another problem is finishing a task, not uploading/reporting it, and then going on holiday for 10days. Seen several posts where a user did this and got no credit. A big put off to the newcomer.
|
||
|
|
rilian
Veteran Cruncher Ukraine - we rule! Joined: Jun 17, 2007 Post Count: 1460 Status: Offline Project Badges:
|
the "report results immediately" option should be ON in BOINC by default...
---------------------------------------- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
No, that option is for those that reimage daily, such as with DeepFreeze. BOINC is designed to minimize contacts with servers that involve the scheduler... work fetches and clearance of Ready to Report, which are combined when possible. Anyone going on vacation can hit the *Update* button and abort any unfinished tasks. Even newcomers;>)
--//-- |
||
|
|
mikey
Veteran Cruncher Joined: May 10, 2009 Post Count: 826 Status: Offline Project Badges: 
|
Sekerob, couldn't you suggest to the "powers that be" to change the policy about sending out repair jobs? Something like have the server wait 24 hrs after the no reply deadline before sending out a repair job? I personally always thought it was silly to send out repair jobs right away like that. One idea Seti was working on a LONG LONG time ago was to send all repair jobs to only machines that send units back within 24 hours on average. I do not think it worked out but maybe you guys can figure out how to make it work. As a cruncher I don't care what kind of units I get as long as they work and I get credit for crunching them. I have been doing this a LONG time, pre Boinc, and know that that doesn't always work out. But I have always thought that giving credits to those that return units late is counter productive to sending them out to other pc's too. If I return a unit late but before the repair job is returned I get credit for crunching and the repair job is cancelled or returned for no credits, that just seems silly. Someone isn't getting credit for crunching and the person who was late learns nothing! If a person consistently gets no credit because they turn in units late then they will earn how to fix it. But to reward them for returning the unit anyway, seems silly. If they are going to reward the original person they should only give them 1/2 credits maybe and the person that got the repair job full credits.   |
||
|
|
|