Message boards :
Number crunching :
Hadam3p no work available?
Message board moderation
Author | Message |
---|---|
Send message Joined: 9 Jan 05 Posts: 30 Credit: 434,469 RAC: 0 |
Sorry if this is a known thing -- been away for a while, tried a search but didn't see anything quite relevant. My account is currently set for hadam3p models only. Whenever the project is resumed it tries to get more work and then tells me none is available for the projects I've selected, etc., despite the status page which shows plenty hadam3p available -- and the number has changed over time so the page does seem functional. This has been true for weeks. I ignored it during the recent server issues and kept the project mostly suspended, but apparently servers are fixed and status page are fixed and it still says there are over 100K of these tasks but I still get this message. It'll be a while yet before my existing model completes but wondering what's causing the message / what I'm missing. I haven't changed my settings since I got the hadam3p I'm crunching now. Clues appreciated. (Windows XP, Boinc 6.6.20) |
Send message Joined: 9 Jan 07 Posts: 467 Credit: 14,549,176 RAC: 317 |
Nothing helpful to suggest, other than 'me too'. BOINC Manager is normally set to 'no new tasks' so that a crash can be restored from a backup rather than getting a new task, so the first few times the 'no work available' message appeared I just set 'no new tasks' again and waited a day or so. However, a few like the following have happened: [size=10]04-Jun-2009 10:37:00 [climateprediction.net] Sending scheduler request: To fetch work. Requesting 274401 seconds of work, reporting 0 completed tasks 04-Jun-2009 10:37:04 [climateprediction.net] Scheduler request succeeded: got 0 new tasks [color=red]04-Jun-2009 10:37:04 [climateprediction.net] Message from server: No work sent 04-Jun-2009 10:37:04 [climateprediction.net] Message from server: No work available for the applications you have selected. Please check your settings on the web site.[/color] 04-Jun-2009 10:37:12 [climateprediction.net] [task_debug] result hadsm3mh_kx3y_006032178_2 checkpointed 04-Jun-2009 10:38:06 [climateprediction.net] Sending scheduler request: To fetch work. Requesting 274516 seconds of work, reporting 0 completed tasks 04-Jun-2009 10:38:10 [climateprediction.net] Scheduler request succeeded: got 1 new tasks 04-Jun-2009 10:38:10 [climateprediction.net] [task_debug] result state=NEW for hadsm3mh_kx0m_006032058_8 from handle_scheduler_reply 04-Jun-2009 10:38:11 [climateprediction.net] [task_debug] result state=FILES_DOWNLOADING for hadsm3mh_kx0m_006032058_8 from CS::update_results 04-Jun-2009 10:38:12 [climateprediction.net] Started download of hadsm3mh_kx0m_006032058.zip 04-Jun-2009 10:38:13 [climateprediction.net] Finished download of hadsm3mh_kx0m_006032058.zip[/size] ... so I now turn 'no new tasks' off for a while and it now gets tasks eventually. |
Send message Joined: 9 Jan 05 Posts: 30 Credit: 434,469 RAC: 0 |
Hmmm, well it helps to know I'm not nuts, or at least not the only one! I wondered if it was due to inactivity or something I can't see; it's been hours now. I'll go look up the debug flags; I don't even see the amount of work its requesting with current settings. My messages look like: 6/29/2009 22:42:30 climateprediction.net Sending scheduler request: To fetch work. 6/29/2009 22:42:30 climateprediction.net Requesting new tasks 6/29/2009 22:42:35 climateprediction.net Scheduler request completed: got 0 new tasks 6/29/2009 22:42:35 climateprediction.net Message from server: No work sent 6/29/2009 22:42:35 climateprediction.net Message from server: No work available for the applications you have selected. Please check your settings on the web site. 6/29/2009 22:47:30 climateprediction.net Sending scheduler request: To send trickle-up message. 6/29/2009 22:47:30 climateprediction.net Requesting new tasks 6/29/2009 22:47:36 climateprediction.net Scheduler request completed: got 0 new tasks 6/29/2009 22:47:36 climateprediction.net Message from server: No work sent 6/29/2009 22:47:36 climateprediction.net Message from server: No work available for the applications you have selected. Please check your settings on the web site. etc. |
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
From my recent messages list: 30/06/2009 12:33:55 PM|climateprediction.net|Sending scheduler request: To send trickle-up message. Requesting 25920 seconds of work, reporting 1 completed tasks It's been running for nearly 3 hours now. :) |
Send message Joined: 9 Jan 05 Posts: 30 Credit: 434,469 RAC: 0 |
well according to sched_op_debug it now makes sense to me: 6/29/2009 23:26:54 climateprediction.net [sched_op_debug] Starting scheduler request 6/29/2009 23:26:54 climateprediction.net Sending scheduler request: To fetch work. 6/29/2009 23:26:54 climateprediction.net Requesting new tasks 6/29/2009 23:26:54 climateprediction.net [sched_op_debug] CPU work request: 0.00 seconds; 0 idle CPUs 6/29/2009 23:26:54 climateprediction.net [sched_op_debug] CUDA work request: 30240.00 seconds; 1 idle GPUs 6/29/2009 23:26:59 climateprediction.net Scheduler request completed: got 0 new tasks 6/29/2009 23:26:59 climateprediction.net [sched_op_debug] Server version 607 6/29/2009 23:26:59 climateprediction.net Message from server: No work sent 6/29/2009 23:26:59 climateprediction.net Message from server: No work available for the applications you have selected. Please check your settings on the web site. Except, how do I turn the GPU preference off so it will stop asking for work for it? I thought it was by project and I'm not seeing it in CPDN. I've ignored this new aspect of BOINC and am therefore,um, ignorant about it... p.s. from a brief scan of boinc message boards I gather this is a known issue and no way to turn it off by project at the moment. |
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
|
Send message Joined: 9 Jan 07 Posts: 467 Credit: 14,549,176 RAC: 317 |
I'm on 6.2.19, which is pre-CUDA. However, I'll turn on that debug flag and see what happens next time. |
Send message Joined: 29 Sep 08 Posts: 5 Credit: 4,330,352 RAC: 0 |
I re-set the project and it got some more work, before that it just was stuck in a loop of asking for GPU work. According to here: http://www.dunadd.co.uk/seti/forum/viewtopic.php?t=1338&sid=9d22a5431b3621e86870a62b45afcff1 Suspending, waiting 10mins then trying again works, but not in my hands. |
Send message Joined: 9 Jan 05 Posts: 30 Credit: 434,469 RAC: 0 |
Les Bayliss wrote: Thanks Les, that's actually the thread I was referring to when saying it looked like it was a known issue. I'm guessing climate will give me more work when I've completed more of my existing model. If not, I'll be back scratching my head :) |
Send message Joined: 31 Dec 07 Posts: 1152 Credit: 22,053,321 RAC: 4,417 |
Les Bayliss wrote: I have found that in order for the server to give me a new WU the time to completion on one of my existing WU’s has to be less than 12 hours. |
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
The time until more work is sent, is the value against Computer is connected to the Internet about every in prefs. The default is .1 days, which is 2.4 hours. However, I found years ago that it could be as long as 5 hours, or shorter than 2.4 hours. So BOINC was a bit 'iffy'. And now there's also the Maintain enough work for an additional value to muddy the water. With the long climate models, and a bit of experimenting, it should still be possible to set it to a time that suits the user. i.e. Not in the middle of the night, if this is important. Unless other projects are also being run, it which case, good luck with getting one when you're watching. :) |
Send message Joined: 29 Sep 04 Posts: 2363 Credit: 14,611,758 RAC: 0 |
In the BOINC Manager Advanced menu there's a Preferences option. In its Network usage tab one can change the additional work buffer figure to up to 10.00 days, then click OK. This affects only the compter that it's on and for that computer only overrides the preferences selected on the website in one's account. As far as I can see the additional work buffer time is calculated by BOINC project-by-project. For example, mine's set to 10 days on the C2D. Until my models will take less than 10 days to complete I can't fetch new climate models. But even when the models will take more than 10 days to complete I can still fetch new work from other projects. If you have any tasks for a project suspended, BOINC refuses to fetch new work for that project. Cpdn news |
©2024 climateprediction.net