Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
![]() |
World Community Grid Forums
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
No member browsing this thread |
Thread Status: Active Total posts in this thread: 6
|
![]() |
Author |
|
insert username here
Cruncher Joined: Feb 2, 2011 Post Count: 2 Status: Offline |
I just recently joined WCC, yesterday in fact, and I've noticed that I have some strange "running, high priority" messages in my BOINC tasks view. Yes, I do realize that this normally means that BOINC has downloaded work and believes that it has to prioritize tasks to meet deadlines....
![]() I find it strange that BOINC has started and paused so many work units and even stranger that it is prioritizing units that have a later due date than previously started ones. |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
In the first line, the size of your task cache is too big. With variable run times and BOINC always assuming that if a task runs e.g. 50% longer, that will apply to all following tasks of the sameproject speak all of WCG. Even at 7 days cache that would push it beyond the 10 days standard deadline. Fix: Reduce your "Connect Every..." and "Additional Buffer" settings so that the sum is not exceeding for instance 5 days. We recommend not to exceed 1 day as work is general available, and WCG unscheduled downtime is never longer. That is of course, there are always individual reasons why one caches that much work, but then consider the variability of runtimes :D
----------------------------------------When the HP condition develops, the client works through a number of test scenarios to find out if the "newest" (latest deadline) and recent work might in fact have different run times, so the client starts a number of cached tasks to find out, which is what you see. After, BOINC will reconsider if to test more from the end or finish off those "waiting to run" first, but certainly, long as all jobs run in HP, no more work will be fetched (should not). --//-- [OT] See you got Kubuntu. Is it the KDE SC 4.6.0?. I put it on top of Maverick 10.10, but cant get it to stabilize. There's a constantly error cycling window that prevents to start even the system monitor to see what is causing this.[/OT] edit: I see your quad or is it a duo with hyperthreading is projecting the HFCC jobs at 11+ hours. The current science mean is 6.6 hours, so if a regular quad, something caused BOINC to think they need atm lots more time, bloating the buffered work total. [Edit 1 times, last edit by Former Member at Feb 3, 2011 9:15:31 AM] |
||
|
gb009761
Master Cruncher Scotland Joined: Apr 6, 2005 Post Count: 2983 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Just to follow on from SekeRob's observations - I'm guessing it's the completed CEP2 task (finishing with 11H 39M) that's bloated all the other HFCC WU's up over the 11Hr mark - so once a few HFCC WU's have finished (and brought down the remaining estimates), the High Priority issue may go away.
----------------------------------------As SekeRob has suggested though, you may like to reconsider your cache size. ![]() |
||
|
insert username here
Cruncher Joined: Feb 2, 2011 Post Count: 2 Status: Offline |
Ah, ok. It didn't occur to me about the cache (I have it currently set at 5 days.) I didn't know that the runtimes for HFCC would vary so greatly (I should have researched the project more.) I've done most of my crunching for Einstein in which runtimes do not vary greatly amongst similar workunits. I see after running a few that runtimes for HFCC, on this computer, vary from 4 to 12 hours.
I'll do as suggested and make adjustments to the cache and connect times. Thanks for the help and clarifications. @SekeRob: I have KDE 4.5.1, thanks for the heads up on 4.6.0. |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
As an off-topic follow up, KDE 4.6.1 is the SP1 equivalent of windows. Read here: http://aseigo.blogspot.com/2011/02/if-you-thought-460-was-good.html . Will try it again then.
|
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Interesting blog, I've bookmarked it for further perusal. Thanks for that. Hope 4.6.1 does it for you. I'm not usually that adventurous... I'll bid my time until April for the distro upgrade/KDE 4.6.X.
-----------------------------------------edit- I "inserted" the "username" finally. ![]() [Edit 2 times, last edit by Former Member at Feb 10, 2011 5:24:53 AM] |
||
|
|
![]() |