| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: Waiting to run [RESOLVED] |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 30
|
|
| Author |
|
|
Gandalf_PCHF
Advanced Cruncher Joined: Dec 21, 2008 Post Count: 81 Status: Offline |
My BOINC Manager - Tasks are showing tasks with the Status column showing Waiting to run. I have about 32 tasks with as much as 99% of the task completed now halted with a status of Waiting to run.
----------------------------------------I also have tasks with a Status remark of Running, high priority. Can someone please explain what has happened to my WCG software. I've been running WCG for about 3 years and these conditions are new to me. I have not had any changes in hardware or software. I did turn on the GPU function, but a new member of our team just started running the WCG with a new install of the software without any modifications to the options that I know of and he is showing the Waiting to run status as well. All help is mightally appreciated.   O/S -LinuxMint 64-bit, Release 13 (maya). DE: Cinnamon [Edit 1 times, last edit by Gandalf_PCHF at Jan 16, 2013 1:12:03 PM] |
||
|
|
branjo
Master Cruncher Slovakia Joined: Jun 29, 2012 Post Count: 1892 Status: Offline Project Badges: 
|
It is common also for me (running W7 64, Boinc Manager v. 7.0.36). When Boinc Manager download some task with shorter deadline (e.g. 3 days for HCC), the tasks with normal deadline (e.g. 7 days for HCC) are put on hold regardless on percentage of the task completed and Boinc Manager start to crunch that "high priority" task with shortened deadline. After BM finish it, continue on normal priority tasks.
----------------------------------------Cheers   Crunching@Home since January 13 2000. Shrubbing@Home since January 5 2006  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi Gandalf,
32 waiting [of WCG only?] because others are running in high priority suggests you've got too high a work buffer, or lately did not have enough up-time to complete all tasks in normal First In First Out [FIFO] order, which different BOINC clients handle in varying, sometimes confusing ways. The remedy is defenitely to lower the "Connect about every xx days " or "Minimum Work Buffer" depending on client version AND the "Maximum additional work buffer" if that is set high too. Given WCG uptime history and duration of outages, generally 1.5 days is more than enough to bridge any connection problem. |
||
|
|
Gandalf_PCHF
Advanced Cruncher Joined: Dec 21, 2008 Post Count: 81 Status: Offline |
This sounds good to a point, and thanks for the timely reply. However, I am running W7 64 BM v. 6.10.58, taking your info into account why do I have 32 tasks on hold? When it finished with a priority task, should it not either return to the task it had put on hold or at least start running another priority task? It seem as though it forgets about the task it put on hold and just starts up a new task then if necessary puts it on hold and goes off to do a priority task. branjo how many tasks do you have on hold? If you are running a quad processor, then you could have 8 tasks running at the same time. In my thinking you should only have 8 tasks on hold at most or is my thinking wrong?
----------------------------------------PS. my computer is up 24/7. O/S -LinuxMint 64-bit, Release 13 (maya). DE: Cinnamon [Edit 1 times, last edit by Gandalf_PCHF at Nov 29, 2012 11:31:34 AM] |
||
|
|
Gandalf_PCHF
Advanced Cruncher Joined: Dec 21, 2008 Post Count: 81 Status: Offline |
SekeRoy - thanks for the info. You may have hit the nail on the head. My ... Connect about every ... was set at 0.30 days. Probably my own fault. The problem with trying to make things work faster is sometime you crash. Have reset it to your recommend 1.5. We will see if the hold count goes up or down. I'll post the results in a couple of days when I'm should your fix fixed the problem.
----------------------------------------This may be branjo's fix as well. O/S -LinuxMint 64-bit, Release 13 (maya). DE: Cinnamon |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Why? Well I said " which different BOINC clients handle in varying, sometimes confusing ways. ''. It's known to happen and the known fix is lowering the cache with 6.10.58 (the version you now tell us to have). Eventually sanity will return, but considering 24/7 the permanent solution is cache lowering. As of now wont advice on upgrading, simply because 7.xx is still in flux, but 7.0.38 does fine, and not seeing this ever greater "waiting to run" building. Running at times with 1.75 days cache which mostly will prioritize the repair jobs with short deadline, but never it's having more than 8 waiting [that'I've seen].
----------------------------------------edit: to clarify, 8 waiting to run on an octo. [Edit 1 times, last edit by Former Member at Nov 29, 2012 6:36:14 PM] |
||
|
|
branjo
Master Cruncher Slovakia Joined: Jun 29, 2012 Post Count: 1892 Status: Offline Project Badges:
|
This sounds good to a point, and thanks for the timely reply. However, I am running W7 64 BM v. 6.10.58, taking your info into account why do I have 32 tasks on hold? When it finished with a priority task, should it not either return to the task it had put on hold or at least start running another priority task? It seem as though it forgets about the task it put on hold and just starts up a new task then if necessary puts it on hold and goes off to do a priority task. branjo how many tasks do you have on hold? If you are running a quad processor, then you could have 8 tasks running at the same time. In my thinking you should only have 8 tasks on hold at most or is my thinking wrong? PS. my computer is up 24/7. Just a few - as far as I observed, max 3. But as soon as "prioritized" are finished, BM is going back to those put on hold (i.e. it is not starting new). In the past few days, BM is even waiting while normal tasks with over 90% of the task completed finish and then start "High priority" ones. Try SekeRob's advices and post, please, if you succeeded  Crunching@Home since January 13 2000. Shrubbing@Home since January 5 2006 |
||
|
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges:
|
I have a question...Why stop a GPU WU at 99.707% with 1 sec remaining?
---------------------------------------- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Ask the developer as it has been asked dozens of times. If it is pre-empting time, it is pre-empting time, which is programmed to happen at a checkpoint, even if it's the last one [but who's to tell that's really true, and who's to tell the next checkpoint is not hours away... the core argument]. In case of an EDF state, not even any waiting... the decision is taken and immediately effectuated. You could set the switch time to a ludricrous high value, and most tasks will run uninterrupted [if you have multiple active projects attached to a client]
----------------------------------------edit: Whilst, in some cases if there are more tasks on a client than cores that are in deadline thread, that would be kind of the behavior on v 6. On 7 it's much better. [Edit 1 times, last edit by Former Member at Nov 29, 2012 7:58:52 PM] |
||
|
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges:
|
I'm seeing it on v7.
---------------------------------------- |
||
|
|
|