climateprediction.net home page
Computation error

Computation error

Questions and Answers : Windows : Computation error
Message board moderation

To post messages, you must log in.

AuthorMessage
old_user64735

Send message
Joined: 19 Mar 05
Posts: 2
Credit: 74,295
RAC: 0
Message 23560 - Posted: 10 Jul 2006, 18:00:16 UTC

I\'ve just finished my first module for climateprediction, over 700 hours. It seemed to go a bit strange when reporting back. No error message. After downloading a lot of files I got \"ready to run\", then \"computation error\". I\'ve tried updating and resetting and all I get is \"computation error\".

Am I doimg something wrong?
ID: 23560 · Report as offensive     Reply Quote
Profile MikeMarsUK
Volunteer moderator
Avatar

Send message
Joined: 13 Jan 06
Posts: 1498
Credit: 15,613,038
RAC: 0
Message 23568 - Posted: 10 Jul 2006, 19:59:37 UTC

Hi Les,

It looks like your model crashed, the main thing I would recommend is taking a backup at intervals (I do it weekly). This is fairly quick and simple:

* Right-click on the icon, \'exit\'
* Navigate to c:\\program files\\ using My Computer or Explorer
* Right-click on \\Boinc\\, \'copy\', \'paste\'
* (If you get a \'files in use\' error, then reboot, and repeat the above steps)
* Now you can restart the model, by running the manager.

There\'s a brief summary of various crashes and ideas on how to avoid them, in the following thread.

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_thread.php?id=4231
I'm a volunteer and my views are my own.
News and Announcements and FAQ
ID: 23568 · Report as offensive     Reply Quote
old_user64735

Send message
Joined: 19 Mar 05
Posts: 2
Credit: 74,295
RAC: 0
Message 23601 - Posted: 13 Jul 2006, 20:33:10 UTC - in response to Message 23568.  

Hi Les,

It looks like your model crashed, the main thing I would recommend is taking a backup at intervals (I do it weekly). This is fairly quick and simple:

* Right-click on the icon, \'exit\'
* Navigate to c:\\program files\\ using My Computer or Explorer
* Right-click on \\Boinc\\, \'copy\', \'paste\'
* (If you get a \'files in use\' error, then reboot, and repeat the above steps)
* Now you can restart the model, by running the manager.

There\'s a brief summary of various crashes and ideas on how to avoid them, in the following thread.

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_thread.php?id=4231


Thanks for the reply. The problem started when the processing of my first module ended. It seemed to upload and download OK. Now when I try to update the project I get files downloaded, then I get \"ready to run\". As soon as it starts to run I get \"computation error\".

I don\'t follow how you said to do the reset. What do you mean by

Right-click on \\Boinc\\, \'copy\', \'paste\'

Copy what? That looks like a folder to me. I\'d be grateful for any further advice.
ID: 23601 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 23602 - Posted: 13 Jul 2006, 21:25:29 UTC

It IS a folder. BOINC
ALL of this folder, (and it\'s sub-folders), MUST be copied and then pasted somewhere else. While the program is NOT running.

To restore in the event of a crash:
Exit
Delete the contents of the BOINC folder
Copy the backup folder
Paste it into the BOINC folder
Reboot to remove anything left in memory from the crash.

I don\'t follow how you said to do the reset. ...

Mike was describing how to make a backup, not how to reset.

ID: 23602 · Report as offensive     Reply Quote
KWSN Sir Clark

Send message
Joined: 8 Jul 05
Posts: 33
Credit: 1,274,211
RAC: 0
Message 23605 - Posted: 14 Jul 2006, 2:02:15 UTC
Last modified: 14 Jul 2006, 2:04:01 UTC

My backup method:

Suspended the project in the Tasks tab.
Exit BOINC.
Navigate to the folder containing the BOINC folder.
Delete any existing backup.
Right-click on the BOINC folder and create a ZIP file of it (I use WinRAR).
(Creating a RAR file takes a lot longer on my system)
Start up BOINC again and resume crunching.
Only takes a couple of minutes tops......I do it daily (when I remember)

To restore the backup I just delete the BOINC folder and unzip the zip file.
Back crunching again within a minute.

I\'m only crunching CPDN so I don\'t have other project WUs to worry about.
ID: 23605 · Report as offensive     Reply Quote
old_user35767

Send message
Joined: 11 Jan 05
Posts: 1
Credit: 22,415
RAC: 0
Message 23986 - Posted: 17 Aug 2006, 14:32:02 UTC

Hello all,

I run BOINC.
I run climate prediction.
Or rather I used to.

It ran for about 8 months and I got half way through a model
The total run time required was about 3000 hours or more!

Then about a month ago I got a Computation Error.
It really does not tell the user very much
except that there is a problem.

Then BOINC dowloaded a fresh set of files
and again there was a Computation Error.
And again..
And again..

It really does not bother me that much
but I am curious.

John
ID: 23986 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 23987 - Posted: 17 Aug 2006, 14:43:12 UTC

John

You\'re trying to run the new Coupled Ocean models with an obsolete version of BOINC. They need a 5.* version, and you still have 4.19

And make sure that you uninstall 4.19 before trying to upgrade, or you\'ll probably end up with split versions. The 5.* versions are completely different to the 4.* versions.

ID: 23987 · Report as offensive     Reply Quote
old_user132880

Send message
Joined: 8 Dec 05
Posts: 21
Credit: 215,749
RAC: 0
Message 24220 - Posted: 6 Sep 2006, 14:37:05 UTC - in response to Message 23987.  

John

You\'re trying to run the new Coupled Ocean models with an obsolete version of BOINC. They need a 5.* version, and you still have 4.19

And make sure that you uninstall 4.19 before trying to upgrade, or you\'ll probably end up with split versions. The 5.* versions are completely different to the 4.* versions.



I do daily backups, and my most recent model crashed in the same exact spot as it did before. So....not the best of models then?

ID: 24220 · Report as offensive     Reply Quote
Profile MikeMarsUK
Volunteer moderator
Avatar

Send message
Joined: 13 Jan 06
Posts: 1498
Credit: 15,613,038
RAC: 0
Message 24224 - Posted: 6 Sep 2006, 19:21:29 UTC

Some combinations of start parameters are inherently unstable, finding out which ones these are is one of the goals of the project. If you wanted to experiment further, you could try running from an older backup, but the chances are that the model will stop at the same point again. I would instead recommend getting a new model instead.
I'm a volunteer and my views are my own.
News and Announcements and FAQ
ID: 24224 · Report as offensive     Reply Quote
Profile Jonathan Jacobs

Send message
Joined: 26 Nov 05
Posts: 3
Credit: 356,351
RAC: 0
Message 24239 - Posted: 9 Sep 2006, 2:31:31 UTC

I do not know if someone else has the answer to this already on this site. But, here is goes. My CP-WU had comp. errors in it(why I\'m here), now it is not downloading any wu\'s. I check my account and I can see the wu there and there is more then half listed with errors. Can I get them (the server) to redownload them? That would also be apart of my wish list.

Some of you say to do a backup. Why??? Wouldn\'t the error still be there? In the data or in the file.

I have ver. 5.4.9
ID: 24239 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 24240 - Posted: 9 Sep 2006, 3:12:09 UTC

There are various errors on the pages for each of your models, and you may get some clues from Mike\'s sticky post here.

The server will only allocate the next dataset on it\'s list. While crashed models DO cause their datasets to get flagged for re-issue, this will be to a \'random computer\'.

Backups.
You should make them on a regular basis, and restore one made BEFORE a crash.

One of the problems with your computer, is that it\'s short on memory. While it may have the minimum recommended, the value of 478.42 MB indicates that you don\'t have a separate graphics card, and some of the system memory is being used by the onboard graphics chips. And even more will get used by other graphics intense programs when they start to display something. Which means a sudden drop in memory available to the climate program. If this happens at a critical moment, CRASH.

ID: 24240 · Report as offensive     Reply Quote
Profile old_user17289

Send message
Joined: 13 Sep 04
Posts: 228
Credit: 354,979
RAC: 0
Message 24271 - Posted: 11 Sep 2006, 4:10:32 UTC - in response to Message 24239.  

Some of you say to do a backup. Why??? Wouldn\'t the error still be there? In the data or in the file.

It depends on what caused the crash. A hardware or power failure may have caused it, so a rerun from the last backup will certainly succeed.

But even internal crashes may successfully resume. I had a WU that crashed when around 99% complete. Reran in 3 times from the backup (at 98%), and the 3rd time it succeeded and completed the WU.
ID: 24271 · Report as offensive     Reply Quote

Questions and Answers : Windows : Computation error

©2024 climateprediction.net