| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Help Stop TB
Thread: Limited workunits for Help Stop TB |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 752
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I got 1 new WU early this morning and it has only been 11 days since my last one.
 WOW, at this rate I should have my BRONZE badge by CHRISTMAS. lol  On a serious note, this WU is a _4. _0 was a no reply _1, _2, and _3 all are PVal. Their status reports all appear to have successfully completed. My _4 is running and should finish in ~9 hours. |
||
|
|
KWSN Sir Clark
Cruncher Joined: Jan 26, 2006 Post Count: 31 Status: Offline Project Badges: 
|
I give up. Too much hassle having to be around to hit Update.
----------------------------------------Seems most of the WUs seem to be snapped up by those with farms or just plain lucky. Haven't had a single WU since the project started and that's with leaving my PC only set to receive HSTB and no other BOINC projects for four weeks Back to FAH2 - good luck with this folks. [Edit 1 times, last edit by KWSN Sir Clark at May 11, 2016 12:26:04 PM] |
||
|
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
Only setting HST1 or FAH2 as second project walks into 2 issues: 1, incremental back-off if work request fails [from 2 minutes to 3+ hours], 2, if you run FAH2, the work request interval is very long. On my 4 core they take 26-28 hours, meaning I have 3.7 work requests per day on that host at a random time. The chance there being a HST in the feeder at that time is very slim when 214 thousand active devices are trying to get 1 of the 10K tasks or so in the feeder each day.
|
||
|
|
pvh513
Senior Cruncher Joined: Feb 26, 2011 Post Count: 260 Status: Offline Project Badges: 
|
I use crontab to do the 3/33 updates for me. Googling should point you to something similar for Windows. I combine HST with OET which has a nice quick turnaround time so that every time crontab does the update, the machine actually needs work. You also need to look at Tools -> Computing preferences -> Max additional work buffer. I have that set at 0.2 days. A non-zero value is important there. At some point I noted that one of my computers never got any HST work. It had Max additional work buffer set to zero.
|
||
|
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
The thing MaxAWB promotes is calling for more work in one go, then when it does, it takes equally long as the depth of the MaxAWB setting before again a workfetch is done, so 0.2 days is at least a 4.8 hour interval before MinWB is hit again.
|
||
|
|
pvh513
Senior Cruncher Joined: Feb 26, 2011 Post Count: 260 Status: Offline Project Badges:
|
I have MInimum work buffer set to 0.5. My impression was that this instructs boinc to load 0.5+0.2 = 0.7 days of work onto my computer, then allows it to drop to 0.5 before requesting more work. But if you force an update, it will try to top up to 0.7 again, even if it hasn't fallen below 0.5 yet. That is the crucial part. This avoids the situation where you force an update, but boinc says "don't need work" and doesn't even try to get HST work. As I said, with MaxAWB set to zero, I got no HST work whatsoever. At least that was the case a few weeks ago.
|
||
|
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
What works for you works for you, but what exactly is the difference between dropping below MinWB+MaxAWB or dropping below MinWB without MaxAWB? Full is full, so a fetch if successful lifts you over MaxAWB+MinWB, and there's still no fetching [don't need]... when partially successful, yes there's another fetch request and on and on.
Other than that, got my reservations about crontabbing the updating and continuously hitting the scheduler, whether work is needed or not... BOINC may not ask for more, but it still makes the scheduler roundtrip (for instance it gets updated host/account stats information and web profile changes, checks alignment of CPID and probably more). Anyway, I just counted May1-May10 and hands-off, MinWB only set and have 14 HST, mostly repairs, occasionally originals... at this rate, it's Ruby by May 24. :D |
||
|
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
Looked briefly for where the 'scheduling priority' is stored in a file apart from it being in memory [see it by visiting project properties in the manager]. If the value is negative, and could be parsed in by the script, skip hitting update, if positive, proceed to execute a request. Did find the <sched_rpc_pending>0</sched_rpc_pending> in client_state, but that does not seem to fit the bill. Anyway, if the scheduling priority value could be parsed and evaluated, it would reduce the update hitting considerably.
|
||
|
|
deltavee
Ace Cruncher Texas Hill Country Joined: Nov 17, 2004 Post Count: 4894 Status: Offline Project Badges: 
|
How about boinccmd --get_project_status which gives Scheduler RPC Pending:
|
||
|
|
pvh513
Senior Cruncher Joined: Feb 26, 2011 Post Count: 260 Status: Offline Project Badges:
|
What works for you works for you, but what exactly is the difference between dropping below MinWB+MaxAWB or dropping below MinWB without MaxAWB? The difference is that with MaxAWB set to zero, boinc will immediately fetch work when a WU is finished and it drops below MinWB. This will happen at a random time, so likely no HST work is available. When 3/33 arrives and you force an update, boinc will conclude that it is over MinWB (as it already requested work before) so it won't request new work and cannot get HST by definition. The end result is you get no HST work (or virtually never). If MaxAWB is non-zero, boinc will wait when a WU is finished since it doesn't drop below MinWB. When 3/33 arrives and you force an update, boinc will see that it has finished WUs and is below MinWB+MaxAWB, so it will request new work and has a decent chance of getting HST work. That is the difference. |
||
|
|
|