| 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: Tasks reset to 0% |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 17
|
|
| Author |
|
|
Dennis Charles
Cruncher Joined: Dec 9, 2010 Post Count: 4 Status: Offline Project Badges: 
|
I had been running CEP for years now, but lately, I noticed that if I snooze the manager for some minutes just to do some other work, and when I get back to resume, the tasks that had been running for hours gets back to zero. Is there something that must be done to correctly have a checkpoint so that I will resume at the progress I left off? I saw an option in the computing preference that I am writhing checkpoints to disk every 30 seconds.
Please advise. Thank you, |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Welcome to the forums.
----------------------------------------Nothing has changed. You must have the "Leave application in memory when suspended" -On- for CEP2 to resume from where it was before a snooze or suspend. If this is not on, a snooze/suspend will unload the task from memory and then fall back to the previous checkpoint. This is with exception to the 1st checkpoint. Unless that one is reached, the task is always held in memory, so not understanding this dropping back to 0%, unless you powered off the computer before completing the #0 job or using a really old BOINC client. The 'Writing to disk AT MOST' setting is a request, if the science application can, not an enforcing. CEP2 checkpoints in this series only 8 times at the max ,over the full runtime, which can by up to 18 hours. Your badge on left, bronze, suggest you've successfully completed more than 14 days, but less than 45 days of computing for CEP2. If you have been running CEP2 for years, that would suggest without many good results. [Edit 1 times, last edit by Former Member at May 27, 2015 8:05:27 AM] |
||
|
|
zanmatteo
Cruncher Joined: Aug 11, 2014 Post Count: 4 Status: Offline Project Badges: 
|
Hi, same issue here, after seven-plus hours of crunching, CPE WU shows no checkpoint.
I had to turn off my PC, and when I restarted it the above WU is back to zero!  To my dismay I had to abort it.... Is there a solution to this WU behaviour? Can CPE force/do checkpoints more often? Otherwise I wan't be able to contribute to CPE anymore... Thanks in advance Matteo crunching in Italy  |
||
|
|
zanmatteo
Cruncher Joined: Aug 11, 2014 Post Count: 4 Status: Offline Project Badges:
|
@Sekerob
Just saw your post, I take it the checkpoints for CPE are saved every 1/8 of the total lengh shown in the "Time Left" colunm. Am I correct? Thanks Matteo chruncing in Italy :) |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
No, the first checkpoint takes the longest to reach [no regular intervals]. Depending on device speed it can be hours, can be 10-12 hours. General recommended for this project is 24/24 but I'd think, not to get too much progress loss, the computer is needed to be on ~12 hours a day.
If you hibernate the computer instead of powering down, on resume there is no loss, but you need to have that 'Leave application in memory when suspended' -On-. |
||
|
|
zanmatteo
Cruncher Joined: Aug 11, 2014 Post Count: 4 Status: Offline Project Badges:
|
Ibernation is actually a very good idea; didn't think of that at all.
I'll try that definitely. Still think the way CPE works the checkpoints, is akward for those like me with time limitations; but they surely have their reason for this. Thanks a lot for the quick and useful answer. Matteo crunching in Italy :) |
||
|
|
Yarensc
Advanced Cruncher USA Joined: Sep 24, 2011 Post Count: 136 Status: Offline Project Badges: 
|
the way CPE works the checkpoints, is akward for those like me with time limitations; but they surely have their reason for this. Since CEP has very large work units, the checkpoint files are very large (they can easily be over a gigabyte). If it checkpointed every minute or so it would make your computer freeze up from the disk usage. They decided to have the work units checkpoint at the end of each sub-task, so that only the results of that sub-task are checkpointed (as opposed to the potentially large checkpoint in the middle of that sub-task's simulation). SekeRob and the WCG tech's have tried requesting that BOINC get a feature to allow users to checkpoint on demand, but so far nothing. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
the way CPE works the checkpoints, is akward for those like me with time limitations; but they surely have their reason for this. Since CEP has very large work units, the checkpoint files are very large (they can easily be over a gigabyte). If it checkpointed every minute or so it would make your computer freeze up from the disk usage. They decided to have the work units checkpoint at the end of each sub-task, so that only the results of that sub-task are checkpointed (as opposed to the potentially large checkpoint in the middle of that sub-task's simulation). SekeRob and the WCG tech's have tried requesting that BOINC get a feature to allow users to checkpoint on demand, but so far nothing. Huh, did I? Alzheimer lite prevents me from remembering, but did propose running in a VM box, since that environment saves itself at intervals and can resume where left off [at an efficiency price]. Is this for the casual cruncher? Of course not, but some are doing it already. |
||
|
|
zanmatteo
Cruncher Joined: Aug 11, 2014 Post Count: 4 Status: Offline Project Badges:
|
@ Yarensc
Thanks for the cristal clear explanation; of course, it does make a lot of sense; surely if they could make it "easier" they would have. @ Sekerob Tried ibernation last night and it worked a charm. CPE started right back were it left. Obviously I had the "Leave application in memory when suspended" turned on.Thanks again for the good tip; for me the problem is solved.  Matteo crunching in Italy |
||
|
|
Yarensc
Advanced Cruncher USA Joined: Sep 24, 2011 Post Count: 136 Status: Offline Project Badges:
|
Huh, did I? Alzheimer lite prevents me from remembering hmm must have been someone else in the Alpha mailing list, maybe I was thinking of something else. @zanmatteo hopefully it doesn't give you any more issues, good luck! |
||
|
|
|