| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: Question about Tasks Running as High Priority [Answered in Spades] |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 8
|
|
| Author |
|
|
wplachy
Senior Cruncher Joined: Sep 4, 2007 Post Count: 423 Status: Offline |
My question, and it is my curiosity rather then perception of a problem, is that I have two PCs that were Windows devices that I brought up under Linux. I ran them as Linux for a few weeks and then brought them back up under Windows. On both devices after switching back to Windows the first two jobs per core ran as High Priority even though they were not due for 10 days and the estimated WU run times were 3 and 6 hours. After the first job per core reached about 50% it switched back to normal. When the second WU started they returned to HP, processed about 50% then returned to normal. All subsequent jobs ran as normal. Both Windows boxes run BOINC 6.2.28.
----------------------------------------The work queue was 0.10 days and each device only pulled 2 jobs per core. Before shutting down the devices to switch OS I stopped pulling new tasks, allowed the queued WUs to finish and made sure they were reported by using manual update. Both the Windows and Linux installs were stand-alone with separate system disks, BOINC installs, versions (Windows 6.2.28 and Linux 6.10.17) and folders. That is, nothing was shared or reused. Can anyone point me to an explanation of why the manager would determine WUs processed after the machine was unused for a while should be run as High Priority? Thank you in advance for any answers. Bill P
Bill P
---------------------------------------- [Edit 1 times, last edit by wplachy at Aug 27, 2010 2:13:24 AM] |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
In the most simplistic explanation, the client keeps stats of the connection and computing times. It thinks that it will not get a chance to complete and upload the result before the deadline from past recent history so goes into panic state, which is why you will not be receiving new tasks either till the HP state clears.
----------------------------------------In 6.10 there is now a condition to stop inflating that connection value when the client has been off for longer than I think 10 days. If there were overdue tasks still in the queue not yet started, the client will self-abort these in 6.10. No point in starting them when it knows that the servers already send a new copy for the No Reply tasks.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Aug 27, 2010 1:35:10 AM] |
||
|
|
anhhai
Veteran Cruncher Joined: Mar 22, 2005 Post Count: 839 Status: Offline Project Badges: 
|
the reason is that bionic keeps track of what % of the time it is actively running. What I mean is say you have your computer one 1 day running bionic full time, then you turn your computer off for 9 days, then bionic will think that it will be only running 10% of the time. So when you start up again, and get a task that takes say 1 hr, well bionic will budget 10 for it.
---------------------------------------- |
||
|
|
anhhai
Veteran Cruncher Joined: Mar 22, 2005 Post Count: 839 Status: Offline Project Badges:
|
hehe, Sekerob, you are too fast.
---------------------------------------- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
the client does keep track of how much time per day boinc is crunching, it saw that in the last 10 days you did less than 3 hrs so by its logic needs hp to finish on time if at all, as you run 24/7 this corrects back to normal.
this is also used on boxes that get turned off for part of the day, I have one that runs 10 hrs of 24, my cache selected is 2 days, it adjusts work requests to keep approx 20 hrs of work, ie what I can finish in 2 days ;) |
||
|
|
BobCat13
Senior Cruncher Joined: Oct 29, 2005 Post Count: 295 Status: Offline Project Badges: 
|
Boinc keeps track of how often the computer is on (on_frac), how often boinc is running when the computer is on, and how often the network access is available while boinc is running. These values are saved in the client_state file under the time_stats section. Here is an example:
----------------------------------------<time_stats> <on_frac>0.998098</on_frac> <connected_frac>0.142578</connected_frac> <active_frac>0.999853</active_frac> <last_update>1282872594.925999</last_update> </time_stats> By switching from Windows to Linux for a couple of weeks, then switching back to Windows, the on_frac for Windows is now low according to boinc, which causes the tasks to run in high priority since boinc isn't sure the computer will be on enough to finish each task before the deadline. By leaving the computer on running the same OS for a few days, boinc will keep increasing the setting for on_frac and high priority mode will go away. [Edit 1 times, last edit by BobCat13 at Aug 27, 2010 1:41:03 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
hehe, Sekerob, you are too fast. yeah, what he said ;)  |
||
|
|
wplachy
Senior Cruncher Joined: Sep 4, 2007 Post Count: 423 Status: Offline |
Sekerob, anhhai, fredski and BobCat13, thank you for the prompt and very comprehensive answers. It now makes sense to even me.
----------------------------------------Thank you   Bill P
Bill P
|
|||
|
||||
|
|