Force report?
Message boards :
Number crunching :
Force report?
Message board moderation
| Author | Message |
|---|---|
   Send message Joined: 3 Dec 16 Posts: 4 Credit: 453,896 RAC: 0 |
Is it possible to get the running WUProp task to report what it has recorded before the 6 hour run time is completed? Will an abort cause it report? I am asking because I dual boot most of my systems, run VMs, reinstall OS, tinker, etc. This would be handy to force WUProp to complete the running task when I want to switch or terminate the OS and not have to wait or potentially lose data. No big deal, just curious. |
    Send message Joined: 29 Jul 11 Posts: 334 Credit: 1,241,519 RAC: 322 |
Would be a nice feature. It's 6 hours to report what's recorded or nothing.  
|
    Send message Joined: 22 Aug 16 Posts: 448 Credit: 2,093,579 RAC: 691 |
Tasks have a week deadline and they don't cancel. |
 DoctorNow    Send message Joined: 7 Apr 10 Posts: 224 Credit: 461,423 RAC: 0 |
Is it possible to get the running WUProp task to report what it has recorded before the 6 hour run time is completed? Will an abort cause it report? I guess that would be similar to the trickles of CPDN, which upload intermediate results from time to time. Technically surely possible, but hardly to program as far as I know, CPDN is the only project which has that. And for just 6 hours probably not worth the effort... Life is Science, and Science rules. To the universe and beyond Member of BOINC@Heidelberg My BOINC-Stats 
|
Michael Goetz   Send message Joined: 18 Apr 13 Posts: 168 Credit: 1,051,124 RAC: 134 |
Is it possible to get the running WUProp task to report what it has recorded before the 6 hour run time is completed? Will an abort cause it report? Primegrid uses trickles. :) And they're fairly easy to code. Want to find one of the largest known primes? Try PrimeGrid. Or help cure disease at WCG. 
|
|
Michael Goetz Send message Joined: 18 Apr 13 Posts: 168 Credit: 1,051,124 RAC: 134 |
Is it possible to get the running WUProp task to report what it has recorded before the 6 hour run time is completed? Will an abort cause it report? As a follow up to this, if I was going to do it... step 1) Read the boinc documentation for trickles. It's very simple. Your app calls an API, and a row gets inserted into the server's database. step 2) You write a cron on the server to handle the information sent by the trickle. For this, what I would do is just keep that trickle in the database until the associated task completes, either by success or failure. If the task succeeds, just erase the trickle. If the task fails, process the data in the trickle as if the task succeeded. No credit, but you still get the hours and WUProp gets some more information for its task properties repository. When should the task trickle? I would not have it trickle automatically. That would probably be far too much data to store on the server. But if you know you're going to kill the task and want to upload its data, you could manually trigger the trickle. The simplest way to do this would be to "touch" (i.e., create) an empty file (perhaps called "trickle.now") in the slot directory where the WUProp task is running. Periodically (e.g., once a minute) the app could check for the existence of trickle.now, and, if found, send a trickle, and then delete the trickle.now file. That's how I would do it. The easy part is changing the app. That's probably only a few lines of code. The hard (or harder) part is the cron on the server to process the trickles and the data contained within. Want to find one of the largest known primes? Try PrimeGrid. Or help cure disease at WCG.
|
|
DoctorNow Send message Joined: 7 Apr 10 Posts: 224 Credit: 461,423 RAC: 0 |
Primegrid uses trickles. :) Never knew... my bad. :-( On what subprojects are they used, I guess only on the long ones? Life is Science, and Science rules. To the universe and beyond Member of BOINC@Heidelberg My BOINC-Stats
|
|
Michael Goetz Send message Joined: 18 Apr 13 Posts: 168 Credit: 1,051,124 RAC: 134 |
Primegrid uses trickles. :) Most of them. Trickles are how we implement the variable deadlines. Everything new uses them, including everything LLR or Genefer. AP27 and WW also. PPS sieve hasn't been updated in a very long time and was never modified to use the system. The apps only trickle once a day, so you'll only see it happen on long-running tasks. We process about thirty thousand trickles each day. Want to find one of the largest known primes? Try PrimeGrid. Or help cure disease at WCG.
|
©2024 Sébastien