| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: FightAIDS@Home Phase 2
Thread: Suspending tasks and starting new ones |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 14
|
|
| Author |
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges: 
|
Can one of the techs explain what criteria is used when pausing tasks that are already running high priority and starting new ones. The reason I ask is more than once I've had tasks that have run for hours stopped and new ones started that all have the same due date which is more than 20 hours away. Seems like a huge waste of resources to stop tasks that would have easily finished and send the unfinished part out to someone else. When I try to pause the new ones and restart the old ones more times than not the restarted ones are server aborted. I don't get it.
----------------------------------------
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
  |
||
|
|
Crystal Pellet
Veteran Cruncher Joined: May 21, 2008 Post Count: 1405 Status: Offline Project Badges:
|
Which BOINC version are you using?
|
||
|
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
Not a server influencable function. Old clients in all cores HP processing state would test large numbers of tasks that look to go overdue, in search of the shorter [self defeating]. This was changed in newer clients which test up to the number of active cores, then continue just in EDF, order of receipt, which is 6 decimal exact [we just see the whole seconds seemingly having multiple with same time]. In such cases LAIM is a good idea for those already running longer. You don't need LAIM for newly tested tasks as those are held in memory until reaching 1st checkpoint. They when paused sit there in pre-empted state. The unlimited testing in old client could lead to free memory problems, one reason to set the core count limit.
|
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges:
|
Which BOINC version are you using? 7.4.42
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
|
||
|
|
Crystal Pellet
Veteran Cruncher Joined: May 21, 2008 Post Count: 1405 Status: Offline Project Badges:
|
Which BOINC version are you using? 7.4.42 In your case all concerning tasks seem to have the same deadline, but in fact there was a very short difference not visible for the user. That's why the client switched to do other tasks first. You can avoid that by adjusting BOINC preferences. Reduce the minimum buffer a bit and you may increase the additional buffer with the same value. [Edit 1 times, last edit by Crystal Pellet at Oct 12, 2015 11:06:20 AM] |
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges:
|
Which BOINC version are you using? 7.4.42 In your case all concerning tasks seem to have the same deadline, but in fact there was a very short difference not visible for the user. That's why the client switched to do other tasks first. You can avoid that by adjusting BOINC preferences. Reduce the minimum buffer a bit and you may increase the additional buffer with the same value. Downloading an exorbitant amount of tasks with a 1/2 day buffer has been the issue and it's still happening. (min set to 0.25 additional set to 0.50) I'm done trying to micro-manage this. If the servers want to send me 100+tasks with the cache set at 0.50 and abort partially finished tasks that would have easily completed if left alone so be it. It's still happening even with the current tasks running at high priority. The ones that don't finish or even start can be resent to someone else. In the mean time I'll just reset cache to 0.
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
----------------------------------------[Edit 1 times, last edit by nanoprobe at Oct 12, 2015 12:17:58 PM] |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7846 Status: Offline Project Badges: 
|
You should not have to micromanage anything. I experienced a similar problem when the length of the tasks of a project varied wildly. There was a time lag in BOINC being able to adjust to the widely variying times. I lowered the additional cache to zero and the minimum cache to .25. It did not cure the problem, but minimized it. If I could not crunch the WU's I let the system do its thing.
----------------------------------------Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges:
|
You should not have to micromanage anything. I experienced a similar problem when the length of the tasks of a project varied wildly. There was a time lag in BOINC being able to adjust to the widely variying times. I lowered the additional cache to zero and the minimum cache to .25. It did not cure the problem, but minimized it. If I could not crunch the WU's I let the system do its thing. Cheers My runtimes are not an issue. They are all within 2 hours of each other. Something else is going on here but at the moment I no longer care since it seems that very few if any other people are experiencing the issue I have. I'll reach sapphire in 4 days and then I'll be dropping off FAHB and moving to something else. Any tasks not finished or started will be returned.
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
|
||
|
|
keithhenry
Ace Cruncher Senile old farts of the world ....uh.....uh..... nevermind Joined: Nov 18, 2004 Post Count: 18667 Status: Offline Project Badges:
|
You should not have to micromanage anything. I experienced a similar problem when the length of the tasks of a project varied wildly. There was a time lag in BOINC being able to adjust to the widely variying times. I lowered the additional cache to zero and the minimum cache to .25. It did not cure the problem, but minimized it. If I could not crunch the WU's I let the system do its thing. Cheers My runtimes are not an issue. They are all within 2 hours of each other. Something else is going on here but at the moment I no longer care since it seems that very few if any other people are experiencing the issue I have. I'll reach sapphire in 4 days and then I'll be dropping off FAHB and moving to something else. Any tasks not finished or started will be returned. Wierd, what you're descibing Nano is what I'm used to seeing with BOINC. Lately though, I'm seeing just the opposite. Both my machines are BOINC 7.6.9. One crunches FAH2 and I'm not seeing anything odd there. The other machine runs OET and three times so far, I've gotten resends (_2 & _3) that have a deadline 4-5 DAYS sooner that any of the other OETs in my queue yet BOINC happily keeps on crunching on the currently active WUs and doesn't switch to the resends, at least not immediately or within an hour or two (half the time needed to finish an OET WU). What I'm used to seeing is an almost immediate switch with the resend showing "High priority". If I force BOINC to switch, it does not switch back but completes the resend and then finishes the suspended/waiting to run WU. |
||
|
|
cjslman
Master Cruncher Mexico Joined: Nov 23, 2004 Post Count: 2082 Status: Offline Project Badges: 
|
...since it seems that very few if any other people are experiencing the issue I have. I have seen this phenomena with other projects, specially when I load the queue to the max trying to reach a badge at project EOL. I'm not crunching FAH2/FAHB right now, but what you describe does sound familiar. From a physiological point of view, yes, it is stressing and drives you nuts. When it happened to me, I was assured by various fellow crunchers (who had many a crunching scar to show off) to not to worry about it: the BOINC manager would figure it out and in the end, life would be beautiful. I really don't remember exactly what did happen (not sure if it's the age or was the liquor), but I do remember that I did make to the corresponding badges. Wish you luck !CJSL Crunching for a better world... |
||
|
|
|