| 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: 13
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi: Something changed - I can't get it to ask for CPU tasks - it's only asking for GPU ones (& none exist) I was running fine before - CPU tasks (WCG) & GPU tasks (Collatz)
2/10/2011 3:19:01 AM World Community Grid Sending scheduler request: Requested by user. 2/10/2011 3:19:01 AM World Community Grid Requesting new tasks for GPU 2/10/2011 3:19:02 AM World Community Grid Scheduler request completed: got 0 new tasks Can anybody help me? |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
If there presently is no WCG work on that system, then do a WCG project reset so it will put the debt back to zero. If there is work in progress, then the client is happy with what it's got momentarily. Note that work on multi-project clients uses a debit / credit debt system. When a project has had it's alloted time, it will stop asking work for such a project until the other projects have caught up.
----------------------------------------I'm suspecting that in the project combo you run, the debt system will always get out of whack. With an added options flag <zero_debts>n</zero_debts> to the cc_config.xml **, you can reset the debts for all projects [restart client]. Change n it to 1 to reset and change n to 0 to stop resetting on client restart. Don't know if reading in the config from the advanced menu will do it too, but doubt it. Many config <options> require a client restart. edit: ** See FAQ index for a link to the wiki that provides instructions how to edit/create the cc_config.xml [Edit 1 times, last edit by Former Member at Feb 10, 2011 8:55:27 AM] |
||
|
|
sk..
Master Cruncher http://s17.rimg.info/ccb5d62bd3e856cc0d1df9b0ee2f7f6a.gif Joined: Mar 22, 2007 Post Count: 2324 Status: Offline Project Badges:
|
<zero_debts>1</zero_debts> results in a Config: zero long-term debts on startup message, suggesting this requires a system restart.
To confirm I upped the cache to 9days, tried this without restarting Boinc (ie just re-reading the cc_config file) and it made no difference. Then I closed Boinc, waited 1min, opened Boinc and I still got, 18/03/2011 09:18:12 World Community Grid Sending scheduler request: To fetch work. 18/03/2011 09:18:12 World Community Grid Requesting new tasks for GPU 18/03/2011 09:18:15 World Community Grid Scheduler request completed: got 0 new tasks So a system restart would be required. If you don't want to restart the system, or reset the project, there is another way: Suspended all other projects (I had won't get new tasks already in place). Boinc will immediately ask WCG for CPU work, if your cache is high enough. Certainly easier than restarting the system or resetting the project just to get a few tasks This is a usefule technique for anyone who keeps their cache low for GPU work, or wants to just run WCG tasks for a while (when Betas are about). Just remember to decrease the cache and then enable projects again, in that order ![]() |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
snip
...if your cache is high enough. AND not yet filled to the brim already. Then upping the cache by a day and immediately upon backfilling with WCG work reducing that cache again to normal operating levels will do it too... BUT will only blow the debts further out of whack, so the consequence is that the user will have to continue manhandling the client... so recommendation remains to just reset the debts ;O Remember, suspended tasks also counts for the "total" cache, sum of all projects. Used to be an old trick with 5.2 - 5.4 or thereabouts which did not count suspended work, so tasks could be cached way out the capacity of the host. --//-- |
||
|
|
Crystal Pellet
Veteran Cruncher Joined: May 21, 2008 Post Count: 1413 Status: Offline Project Badges:
|
Obviously you had enough CPU-work for WCG (or for other projects), but not enough GPU-work.
Although WCG doesn't have GPU-work available, the (stupid) client requests GPU-work from WCG, because of the WCG's long term GPU-debt. (another way to report RtR CPU-tasks ;)) If you want to do more WCG work, increase the resource share. If you want to do more WCG work temporarily, you could use boinccmd from a command line and increase the debts: Example: "Path\boinccmd" --set_debts www.worldcommunitygrid.org 10000 86400 The numbers are seconds. In Boinc Manager you can see the actual debts when clicking properties of a project. [ot] @Mark Reiss: Because of database problems with BOINCstats about a week ago all signature-id's were renumbered. That's why your BOINCstats-signature shows an invalid id. Others got the signature of another cruncher. Your new id and signature is 4005. [/ot] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
WCG tells clients to back-off asking for GPU work for 7 or 14 days, as in: We Aint got it! That is until you hit the Update button which resets.
I never see GPU work requests, though my sched_debug flag is set... more tricks from the Real Thoroughly Friendly Manual ;P --//-- |
||
|
|
kateiacy
Veteran Cruncher USA Joined: Jan 23, 2010 Post Count: 1027 Status: Offline Project Badges:
|
If you want to do more WCG work temporarily, you could use boinccmd from a command line and increase the debts: Example: "Path\boinccmd" --set_debts www.worldcommunitygrid.org 10000 86400 The numbers are seconds. In Boinc Manager you can see the actual debts when clicking properties of a project. Thanks for the tip, Crystal Pellet! I hadn't stumbled on that particular capability of boinccmd before, and it will come in handy. ![]() |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
An evil trick ;>)... who'd have thought of facilitating a command that allows upping/downing debts to push work requesting. From the manual:
--set_debts URL1 STD1 LTD1 {URL2 STD2 LTD2 ...} Set the short- and long-term debts of one or more projects. Note: if you adjust the debts of a project, the debts of other projects are changed, so if you want to set the debts of multiple projects, do it in a single command. Image:list-add.pngNew in 5.10.14 Describes what is seen when resetting a project... all debts are recalculated and will keep it in mind next time manipulating work requesting --//-- |
||
|
|
Crystal Pellet
Veteran Cruncher Joined: May 21, 2008 Post Count: 1413 Status: Offline Project Badges:
|
An evil trick ;>)... . . Describes what is seen when resetting a project... all debts are recalculated and will keep it in mind next time manipulating work requesting ![]() Don't be anxious, Sek The calculation of the debts was always a subject for David A. for reprogramming BOINC. Your right about the fact it's not that simple. In version 6.10.58: The short term debt (STD) can be positive or negative, but the sum of all projects is zero. When I set the STD for WCG to 3000, the debts of all other projects will be lowered so that their sum together (+ or -) will be -3000. The behaviour for the long term debt (LTD) is a bit different. Only active projects with tasks in your list are recalculated. Also different is that the maximum of the LTD is zero, so in my example in the former post +86400 for WCG will be changed to 0 and the LTD of the other active projects will be lowered in a proportion depending of their resource shares. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
There's more stuff the good doctor has been talking about for years, and stuck in status immobiliaria... DCF per science app (no, not the apple short;) seems to be around in a private compile of 6.10.58, the one most sought after feature I'd think for anyone participating at ''umbrella'' grids. It would in a major way take out the horrendous caching fluxes... recently for a 1.75 days cache from 1.5 days to almost 8 days and back.
--//-- |
||
|
|
|