Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
![]() |
World Community Grid Forums
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
No member browsing this thread |
Thread Status: Active Thread Type: Sticky Thread Total posts in this thread: 48
|
![]() |
Author |
|
Speedy51
Veteran Cruncher New Zealand Joined: Nov 4, 2005 Post Count: 1311 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Hope I'm posting this question in the right place. Is it possible to tell whether or not we are processing a task that is the leftover jobs from a task that has reached the 18 hour deadline? Other than the obvious giveaway of running for a shorter time.
----------------------------------------![]() |
||
|
Yarensc
Advanced Cruncher USA Joined: Sep 24, 2011 Post Count: 136 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
I believe when a work unit hits the time limit it is finished on the servers at Harvard, not sent back out to the grid. Otherwise you'd see work units with different numbers of tasks being run, or some just starting completed.
|
||
|
Speedy51
Veteran Cruncher New Zealand Joined: Nov 4, 2005 Post Count: 1311 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
I believe when a work unit hits the time limit it is finished on the servers at Harvard, not sent back out to the grid. Otherwise you'd see work units with different numbers of tasks being run, or some just starting completed. Thanks for your response. This raises one question if the jobs are not fully completed in the 18 hours are the remainder of the jobs run at Harvard? ![]() |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi everybody,
yes, we haven't sent out unfinished wus back to the grid (for various reasons), but rather retain the option to finish up interesting stuff on our in-house cluster. That's both simpler and more adequate to the problem. Best wishes Your CEP team |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Not sure of the reason for the 18 hr CPU time limit. I have 2 machines that should be fast enough to finish in that time, but have lost almost 4 days CPU time in the last few days. I have only had credit for 1 valid wu in that time.
----------------------------------------Unless I can find a cheap used Cray, I'll have to just concentrate all my resources on other projects. [Edit 2 times, last edit by Former Member at Nov 18, 2015 8:59:19 PM] |
||
|
Horse43
Cruncher Joined: Nov 18, 2015 Post Count: 10 Status: Offline Project Badges: ![]() ![]() ![]() ![]() |
Why the work unit packs are over 12 hours long?
|
||
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
Because... the processing time is -very- variable. Some synthetic molecule analyses take few hours, some 6-12 and more. Given there are 8 jobs in a task to get the simulation complete, the 18 hours was chosen as a point where almost all finish the whole set. [originally 12 hrs when there was lighter work].
|
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks for the reply SekeRob*.
----------------------------------------I had successfully completed a number of WUs when I started getting the larger WUs. Had I known (my fault) about the 18 hr limit, I would have just aborted the big ones. Thanks again. [Edit 1 times, last edit by Former Member at Nov 19, 2015 11:37:26 AM] |
||
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
Not really, all depends on whether your cruncher manages to get checkpoint 1 [Job #0]. Have had a few that never got to that point, but eventually decided the frequency was so low, it is okay that the occasional very tough arrived goes into nowhere in the capped time frame [Think credit is still granted for time, but even that I'd survive if not]. There's scripts running to keep the heaviest off the grid, but it's not perfect, and eventually as we had the chance to opt in, no harm if you opt-out again.
(Too many aborts causes a problem in the distribution system, one reason for those 'no work coming' complaints. My two crunchers work 24/24 uninterrupted on 2+1 each, seemingly always.) |
||
|
Yarensc
Advanced Cruncher USA Joined: Sep 24, 2011 Post Count: 136 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Had I known (my fault) about the 18 hr limit, I would have just aborted the big ones. Thanks again. If you're basing that on the Estimated Time column, its ofter quite inaccurate, so its usually safe to run the unit even if BOINC /thinks/ its going to take 20-30+ hours. |
||
|
|
![]() |