climateprediction.net home page
Task at 100% but not finishing

Task at 100% but not finishing

Message boards : Number crunching : Task at 100% but not finishing
Message board moderation

To post messages, you must log in.

AuthorMessage
Trotador

Send message
Joined: 21 Aug 11
Posts: 10
Credit: 24,612,313
RAC: 1,296
Message 48792 - Posted: 14 Apr 2014, 12:05:38 UTC

Hi,

This task

http://climateapps2.oerc.ox.ac.uk/cpdnboinc/result.php?resultid=16386669

is about to be 24 hours at 100% but boincmanager says it is still running. However, it is actually not using any CPU resource. I have shutdown the boincmanager and check that it does not kept stuck in memory. Suspending and restarting the task did not help either.

It seems that the computation has finished but for whatever reason it does not manage to close itself. I think that all crunching results should be already at cpdn server. Could you confirm and advise?. If I abort it will be reissued and maybe it is not necessary.

thanks


ID: 48792 · Report as offensive     Reply Quote
Profile geophi
Volunteer moderator

Send message
Joined: 7 Aug 04
Posts: 2167
Credit: 64,526,915
RAC: 6,447
Message 48793 - Posted: 14 Apr 2014, 12:25:56 UTC - in response to Message 48792.  

What are the messages in the Event Log/Messages tab in Boinc Manager? Are there any files in the Transfers tab of Boinc Manager?
ID: 48793 · Report as offensive     Reply Quote
Profile Dave Jackson
Volunteer moderator

Send message
Joined: 15 May 09
Posts: 4345
Credit: 16,523,697
RAC: 5,963
Message 48794 - Posted: 14 Apr 2014, 12:28:43 UTC

All 20 trickles have been uploaded which means you will have all the credit due for the model. Has the final zip file been uploaded? You can see this in the event log. If not I would be tempted to try an update project to see if it disappears. If it has gone which I suspect is the case, you can safely abort as all the scientific information has been sent to the project.
ID: 48794 · Report as offensive     Reply Quote
Trotador

Send message
Joined: 21 Aug 11
Posts: 10
Credit: 24,612,313
RAC: 1,296
Message 48795 - Posted: 14 Apr 2014, 12:42:36 UTC

When I click update there is no relevant message, just "sending scheduler reques requested by user" and "scheduler request completed". there is no files in the transfer tab.

I've looked in the cpdn directory and there is a folder for this unit. Two text files and several folders are inside. The files, stderr_um.txt is empty and stdout_um.txt contains what seems to be all the development of the task processing ending with "Model finished with xxxx CPU time... Closing model.." which looks like the task was properly finished.

The folders are full of files.

I can't state whether the zip file has been sent or not.

thanks
ID: 48795 · Report as offensive     Reply Quote
Profile Dave Jackson
Volunteer moderator

Send message
Joined: 15 May 09
Posts: 4345
Credit: 16,523,697
RAC: 5,963
Message 48798 - Posted: 14 Apr 2014, 16:10:08 UTC - in response to Message 48795.  
Last modified: 14 Apr 2014, 16:21:42 UTC

If the final zip file has been uploaded there would have been two lines similar to these in the event log.

Mon 14 Apr 2014 14:25:42 BST | climateprediction.net | Started upload of hadam3p_eu_a8vq_2013_1_008572548_0_8.zip
Mon 14 Apr 2014 14:31:30 BST | climateprediction.net | Finished upload of hadam3p_eu_a8vq_2013_1_008572548_0_8.zip

Except for that model type there are only 4 of them as I remember. Unfortunately, if you shut down BOINC Manager before checking the message log will have been wiped.
ID: 48798 · Report as offensive     Reply Quote
Profile geophi
Volunteer moderator

Send message
Joined: 7 Aug 04
Posts: 2167
Credit: 64,526,915
RAC: 6,447
Message 48799 - Posted: 14 Apr 2014, 17:20:05 UTC - in response to Message 48798.  

Unfortunately, if you shut down BOINC Manager before checking the message log will have been wiped.


There is a file in the BOINC directory that logs those messages. I just can't remember the name of it right now.
ID: 48799 · Report as offensive     Reply Quote
Profile astroWX
Volunteer moderator

Send message
Joined: 5 Aug 04
Posts: 1496
Credit: 95,522,203
RAC: 0
Message 48800 - Posted: 14 Apr 2014, 19:18:58 UTC - in response to Message 48799.  

" stdoutdae " -- it can be opened with a text editor.

Trotador:
You've been at this awhile so you might also see a file named stdoutdae.old. That file is as type says -- old. It is created some time after stdoutdae exceeds two MB.

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.
ID: 48800 · Report as offensive     Reply Quote
Trotador

Send message
Joined: 21 Aug 11
Posts: 10
Credit: 24,612,313
RAC: 1,296
Message 48801 - Posted: 14 Apr 2014, 20:19:58 UTC
Last modified: 14 Apr 2014, 20:23:04 UTC

No trace of upload in the stdout file. I can see the last trickle message but I can't find any later finished/upload message.

In the "dataout" folder of the task, there are two files with the name of the unit and ending in .nc that I guess should be the output data but no zip file.

Edit: these two files update their modification date every time I click boinc manager update
ID: 48801 · Report as offensive     Reply Quote

Message boards : Number crunching : Task at 100% but not finishing

©2024 climateprediction.net