Rosetta Server Thinks WU Won't Finish in Time

Message boards : Number crunching : Rosetta Server Thinks WU Won't Finish in Time

To post messages, you must log in.

AuthorMessage
Richard Turner

Send message
Joined: 5 May 09
Posts: 3
Credit: 425,865
RAC: 215
Message 61807 - Posted: 17 Jun 2009, 13:56:22 UTC

I keep getting '0 new tasks' messages in BOINC 6.6.36:

17/06/2009 13:16:43	rosetta@home	Sending scheduler request: Requested by user.
17/06/2009 13:16:43	rosetta@home	Requesting new tasks for GPU
17/06/2009 13:16:48	rosetta@home	Scheduler request completed: got 0 new tasks
<snip>
17/06/2009 13:20:53	rosetta@home	Sending scheduler request: To fetch work.
17/06/2009 13:20:53	rosetta@home	Requesting new tasks for GPU
17/06/2009 13:20:58	rosetta@home	Scheduler request completed: got 0 new tasks
<snip>
17/06/2009 13:25:03	rosetta@home	Sending scheduler request: To fetch work.
17/06/2009 13:25:03	rosetta@home	Requesting new tasks for GPU
17/06/2009 13:25:08	rosetta@home	Scheduler request completed: got 0 new tasks
<snip>
17/06/2009 13:29:14	rosetta@home	Sending scheduler request: To fetch work.
17/06/2009 13:29:14	rosetta@home	Requesting new tasks for GPU
17/06/2009 13:29:19	rosetta@home	Scheduler request completed: got 0 new tasks
17/06/2009 13:33:24	rosetta@home	Sending scheduler request: To fetch work.
17/06/2009 13:33:24	rosetta@home	Requesting new tasks for CPU
17/06/2009 13:33:29	rosetta@home	Scheduler request completed: got 0 new tasks
17/06/2009 13:33:29	rosetta@home	Message from server: No work sent
17/06/2009 13:33:29	rosetta@home	Message from server: (won't finish in time) Computer on 95.0% of time, BOINC on 99.2% of that
17/06/2009 13:37:34	rosetta@home	Sending scheduler request: To fetch work.
17/06/2009 13:37:34	rosetta@home	Requesting new tasks for GPU
17/06/2009 13:37:39	rosetta@home	Scheduler <snip>
17/06/2009 13:41:44	rosetta@home	Sending scheduler request: To fetch work.
17/06/2009 13:41:44	rosetta@home	Requesting new tasks for GPU
17/06/2009 13:41:49	rosetta@home	Scheduler request completed: got 0 new tasks
17/06/2009 13:45:54	rosetta@home	Sending scheduler request: To fetch work.
17/06/2009 13:45:54	rosetta@home	Requesting new tasks for CPU
17/06/2009 13:45:59	rosetta@home	Scheduler request completed: got 0 new tasks
17/06/2009 13:45:59	rosetta@home	Message from server: No work sent
17/06/2009 13:45:59	rosetta@home	Message from server: (won't finish in time) Computer on 95.0% of time, BOINC on 99.2% of that
17/06/2009 13:50:04	rosetta@home	Sending scheduler request: To fetch work.
17/06/2009 13:50:04	rosetta@home	Requesting new tasks for CPU
17/06/2009 13:50:09	rosetta@home	Scheduler request completed: got 0 new tasks
17/06/2009 13:50:09	rosetta@home	Message from server: No work sent
<snip>
17/06/2009 13:54:29	rosetta@home	Sending scheduler request: To fetch work.
17/06/2009 13:54:29	rosetta@home	Requesting new tasks for CPU
17/06/2009 13:54:34	rosetta@home	Scheduler request completed: got 0 new tasks
17/06/2009 14:01:40	rosetta@home	Sending scheduler request: To fetch work.
17/06/2009 14:01:40	rosetta@home	Requesting new tasks for CPU
17/06/2009 14:01:45	rosetta@home	Scheduler request completed: got 0 new tasks
17/06/2009 14:01:45	rosetta@home	Message from server: No work sent
17/06/2009 14:01:45	rosetta@home	Message from server: (won't finish in time) Computer on 95.0% of time, BOINC on 99.2% of that
17/06/2009 14:05:50	rosetta@home	Sending scheduler request: To fetch work.
17/06/2009 14:05:50	rosetta@home	Requesting new tasks for CPU
17/06/2009 14:05:55	rosetta@home	Scheduler request completed: got 0 new tasks
17/06/2009 14:12:00	rosetta@home	Sending scheduler request: To fetch work.
17/06/2009 14:12:00	rosetta@home	Requesting new tasks for GPU
17/06/2009 14:12:05	rosetta@home	Scheduler request completed: got 0 new tasks
<snip>
17/06/2009 14:36:41	rosetta@home	Sending scheduler request: To fetch work.
17/06/2009 14:36:41	rosetta@home	Requesting new tasks for CPU
<snip>
17/06/2009 14:36:46	rosetta@home	Scheduler request completed: got 0 new tasks
17/06/2009 14:36:46	rosetta@home	Message from server: No work sent
17/06/2009 14:36:46	rosetta@home	Message from server: (won't finish in time) Computer on 95.0% of time, BOINC on 99.2% of that


I have snipped attempts to get work from SETI (nothing is available there). The difference is I am getting Rosetta work on other computers, and I have not been getting it on this computer for well over a day. The Rosetta server does not seem to think I'll finish a 3 hour WU before the deadline. This is not correct - weighting of Rosetta (on this machine) is low: 10 out of a total of 170, or currently because of the SETI situation, out of 120. Because of this problem I zeroed debts yesterday morning - to see if it made any difference - but with no effect on this issue. The computer is a dual core, and is one about 16 hours a day (probably more). This does not make sense. An equally rated project (ie. 10) has been running on one of the cores for the majority of the time (even though the deadline for that is May 2010) and there is absolutely no danger of it not finishing within a few weeks.

Can anyone advise, please?
ID: 61807 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Paul D. Buck

Send message
Joined: 17 Sep 05
Posts: 815
Credit: 1,812,737
RAC: 0
Message 61809 - Posted: 17 Jun 2009, 15:05:15 UTC - in response to Message 61807.  

I keep getting '0 new tasks' messages in BOINC 6.6.36:

...
I have snipped attempts to get work from SETI (nothing is available there). The difference is I am getting Rosetta work on other computers, and I have not been getting it on this computer for well over a day. The Rosetta server does not seem to think I'll finish a 3 hour WU before the deadline. This is not correct - weighting of Rosetta (on this machine) is low: 10 out of a total of 170, or currently because of the SETI situation, out of 120. Because of this problem I zeroed debts yesterday morning - to see if it made any difference - but with no effect on this issue. The computer is a dual core, and is one about 16 hours a day (probably more). This does not make sense. An equally rated project (ie. 10) has been running on one of the cores for the majority of the time (even though the deadline for that is May 2010) and there is absolutely no danger of it not finishing within a few weeks.

Can anyone advise, please?

I am guessing that you have one CPDN task running from the deadline mentioned. Sadly, the later versions of BOINC have a large number of issues in the work fetch and work scheduling areas that the developers are not interested in addressing ... this is one of them ... you will have to wait till the CPDN task gets farther along or suspend it ... if you suspend it you will likely get an immediate download of work from one or more projects ... you can then unsuspend the CPDN task and let it get its share ...
ID: 61809 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Richard Turner

Send message
Joined: 5 May 09
Posts: 3
Credit: 425,865
RAC: 215
Message 61811 - Posted: 17 Jun 2009, 16:31:21 UTC

Thanks for the reply. Is it BOINC that decides whether the Rosetta server sends work? Or the Rosetta server? A genuine question.
ID: 61811 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Hammeh

Send message
Joined: 11 Nov 08
Posts: 63
Credit: 211,283
RAC: 0
Message 61833 - Posted: 18 Jun 2009, 13:21:41 UTC

I'm guessing that the BOINC client does some calculations on CPU time which is available on your computer and reports the data to the rosetta server during the scheduler request for new work. It is then the rosetta server that decides the tasks will not be finished in time and gives the error.
ID: 61833 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile dcdc

Send message
Joined: 3 Nov 05
Posts: 1832
Credit: 119,680,204
RAC: 9,884
Message 61835 - Posted: 18 Jun 2009, 19:11:41 UTC - in response to Message 61833.  
Last modified: 18 Jun 2009, 19:13:12 UTC

I'm guessing that the BOINC client does some calculations on CPU time which is available on your computer and reports the data to the rosetta server during the scheduler request for new work. It is then the rosetta server that decides the tasks will not be finished in time and gives the error.


I believe you're right - my understanding is that the server decides based on the info BOINC provides as below:

17/06/2009 14:36:46 rosetta@home Message from server: (won't finish in time) Computer on 95.0% of time, BOINC on 99.2% of that

I presume it must also take into account either the average run-time for reported work, or the run-time setting on your rosetta settings page.
ID: 61835 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Hammeh

Send message
Joined: 11 Nov 08
Posts: 63
Credit: 211,283
RAC: 0
Message 61836 - Posted: 18 Jun 2009, 21:01:50 UTC
Last modified: 18 Jun 2009, 21:03:04 UTC

I would also assume that the project time duration correction factor plays a part in this error. Also, please remember that on the requests which are asking for GPU work, they will always recieve 0 new tasks because rosetta has no GPU applications or tasks to send.
ID: 61836 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Richard Turner

Send message
Joined: 5 May 09
Posts: 3
Credit: 425,865
RAC: 215
Message 61845 - Posted: 19 Jun 2009, 7:03:45 UTC

What you guys are saying sort of agrees with my understanding. There is something not working properly somewhere, for sure. I now have a second machine displaying the same message. It is another dual core running Rosetta and ClimatePrediction with equal weighting. BOINC has now downloaded two CP WUs for completion middle of next year, which it is chugging away processing (one on each of the processors), but I am getting the same message w.r.t. Rosetta ("won't finish in time"). Neither CP WUs are running high priority.
ID: 61845 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Hammeh

Send message
Joined: 11 Nov 08
Posts: 63
Credit: 211,283
RAC: 0
Message 61848 - Posted: 19 Jun 2009, 10:25:31 UTC

There is not fix that I can see, other than suspending the CPDN project tasks and requesting new rosetta ones. This is a bug with the scheduler that has been pointed out to the developers many times.
ID: 61848 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
DJStarfox

Send message
Joined: 19 Jul 07
Posts: 145
Credit: 1,250,162
RAC: 0
Message 61864 - Posted: 20 Jun 2009, 14:39:50 UTC

Your other option is to downgrade BOINC to version 6.4.7 until the 6.6 series is stable enough and addresses this issue.
ID: 61864 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Hammeh

Send message
Joined: 11 Nov 08
Posts: 63
Credit: 211,283
RAC: 0
Message 61866 - Posted: 20 Jun 2009, 16:57:19 UTC

Yes - agreed with above. If you are not running any GPU (CUDA) applications then I would downgrade to 6.4.7 if you are having problems. All of the 6.6.x clients seems to have major problems still.
ID: 61866 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
gerlik

Send message
Joined: 19 Nov 06
Posts: 10
Credit: 122,074
RAC: 0
Message 62500 - Posted: 27 Jul 2009, 13:40:42 UTC - in response to Message 61866.  

Yes - agreed with above. If you are not running any GPU (CUDA) applications then I would downgrade to 6.4.7 if you are having problems. All of the 6.6.x clients seems to have major problems still.



Work with 6.45 & cant download/upload: error --> server probably down (since saturday)
ID: 62500 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : Rosetta Server Thinks WU Won't Finish in Time



©2024 University of Washington
https://www.bakerlab.org