Questions and Answers :
Getting started :
How do I defer results getting reported?
Message board moderation
Author | Message |
---|---|
red-ray Send message Joined: 24 Jun 99 Posts: 308 Credit: 9,029,848 RAC: 0 |
At the moment my systems as reporting results every 5 minutes. Is it possible to defer reporting 'till there are either more than a specified number to report or a specified time has passed please? I am typically reporting as below. I tried adding <report_results_immediately>0</report_results_immediately> but this did not help. In a tryical day I see 97 MB uplaoded and 286 MB downloaded. The amount uploaded is more than I was expecting and am wondering if reporting results every 5 minutes could be the reason it's so large. I am using BOINC x64 V7.0.25. As I recall 6.12.34 used to do the same. 29/04/2012 17:10:46 | SETI@home | [dcf] DCF: 6.892625->6.352469, raw_ratio 1.491066, adj_ratio 0.216328 |
BilBg Send message Joined: 27 May 07 Posts: 3720 Credit: 9,385,827 RAC: 0 |
29/04/2012 17:38:03 | SETI@home | Reporting 4 completed tasks, not requesting new tasks Since BOINC makes Scheduler requests only to Report completed tasks (and not requesting new tasks) BOINC thinks it will not have Internet connection in the next XX days and is in a hurry to report while it can. This depends on "Maintain enough tasks to keep busy for at least 10 days" (if the deadline is in the next 10 days BOINC will report immediately) So change your cache from e.g. 10+10 to 8+12 or 5+15 days. (the values >10 for the second setting are not possible with the GUI (BOINC Manager) but it's possible to edit directly: <work_buf_additional_days>XX.000000</work_buf_additional_days> in global_prefs_override.xml ) Also the setting "Switch between tasks every XYZ minutes" may affect this if set to big values (e.g. 15000 minutes) Â - ALF - "Find out what you don't do well ..... then don't do it!" :) Â |
John McLeod VII Send message Joined: 15 Jul 99 Posts: 24806 Credit: 790,712 RAC: 0 |
29/04/2012 17:38:03 | SETI@home | Reporting 4 completed tasks, not requesting new tasks The task switch interval is part of the computation of the computation deadline, but not part of the computation of when to actually report. BOINC WIKI |
red-ray Send message Joined: 24 Jun 99 Posts: 308 Credit: 9,029,848 RAC: 0 |
Thank you Bil, I have set <work_buf_additional_days>20.000000</work_buf_additional_days> which will help getting work when the DCF is stupid. 30/04/2012 00:45:28 | SETI@home | [dcf] DCF: 2.139742->8.482084, raw_ratio 8.482084, adj_ratio 3.964069 I have connect every 8 days set to force shorties to get run as they arrive so I have enought WUs cached to keep busy when the server is down. I will see what happens if I set zero tomorrow. |
BilBg Send message Joined: 27 May 07 Posts: 3720 Credit: 9,385,827 RAC: 0 |
I have connect every 8 days set to force shorties to get run as they arrive so I have enought WUs cached to keep busy when the server is down. I will see what happens if I set zero tomorrow. With BOINC 7.0.25 "zero" for the "connect every/Maintain enough tasks to keep busy" is not good, it will not ask for work at all. (with 8+20 it will wait for the cache to be < 8 days, then will ask for 20 days) John McLeod VII say "The task switch interval is part of the computation of the computation deadline, but not part of the computation of when to actually report." so you can use this setting "to force shorties to get run as they arrive" instead. Try "Switch between tasks every 14400 minutes" (which should make the shorties run at "high priority") Â - ALF - "Find out what you don't do well ..... then don't do it!" :) Â |
red-ray Send message Joined: 24 Jun 99 Posts: 308 Credit: 9,029,848 RAC: 0 |
I will try <cpu_scheduling_period_minutes>14400.000000</cpu_scheduling_period_minutes> and see what happens. I have now set <work_buf_min_days>0.000000</work_buf_min_days> but it's still reporting results every 5 minutes. |
BilBg Send message Joined: 27 May 07 Posts: 3720 Credit: 9,385,827 RAC: 0 |
Is it possible that you have tasks that are over or very near the deadline? (and you don't forget to "Read local prefs file" after manual edit?) And "reporting results every 5 minutes" is not coupled with asking for tasks? Â - ALF - "Find out what you don't do well ..... then don't do it!" :) Â |
red-ray Send message Joined: 24 Jun 99 Posts: 308 Credit: 9,029,848 RAC: 0 |
Looking at http://setiathome.berkeley.edu/results.php?hostid=6379672&offset=1573&show_names=0&state=1 4 Jun 2012 | 23:53:56 UTC is the soonest. The GPUs will totally empty the cache in four days and the CPUs in similar. Remember for my systems the 400 + 50 limits are stupidly small. |
red-ray Send message Joined: 24 Jun 99 Posts: 308 Credit: 9,029,848 RAC: 0 |
I have discovered this is a "feature" from changeset [trac]changeset:23571[/trac]. The word bug comes to mind. |
BilBg Send message Joined: 27 May 07 Posts: 3720 Credit: 9,385,827 RAC: 0 |
You may propose changes/new value (-1) for <report_results_immediately> (I don't know how, maybe direct email to David Anderson http://setiathome.berkeley.edu/view_profile.php?userid=432107 He posted his email here: http://setiathome.berkeley.edu/forum_thread.php?id=61471&nowrap=true#1037452 ) As you know now it is: <report_results_immediately>0|1</report_results_immediately> 0 == Default value/behaviour (report results on several events) 1 == Force report_results_immediately Propose: -1 == Force Do_NOT_report_results_immediately (report results ONLY on events, e.g. "24 hours before deadline", "On a request for new work", ...) Will this be accepted as needed/feasible I don't know. Â - ALF - "Find out what you don't do well ..... then don't do it!" :) Â |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.