climateprediction.net home page
Notifying the project of a crashed model?

Notifying the project of a crashed model?

Questions and Answers : Preferences : Notifying the project of a crashed model?
Message board moderation

To post messages, you must log in.

AuthorMessage
John Perko

Send message
Joined: 3 Sep 04
Posts: 9
Credit: 582,919
RAC: 0
Message 17064 - Posted: 8 Nov 2005, 21:01:40 UTC

The CP project, in its license agreement, asks to be notified when a WU crashes, so they can assign it to someone else, but they don\'t say who to notify.

I got a FORTRAN end-of-file read error using BOINC 5.2.6 during WU 1ktv_100094407, so I reset the project. Perhaps its recoverable?

Thanks for any advice.

John
ID: 17064 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 17065 - Posted: 8 Nov 2005, 21:20:21 UTC

I think the \"notify\" requirement is a leftover from the early pre-BOINC days of cp.
All failed models now will be automatically flagged by the server software for possible re-issue, up to a maximum of 5 times.
You can see the \"issue\" number of your models by looking at the last digit. If it\'s zero, it\'s the first time it\'s been run. This can also be seen by going to your result page for a model, and clicking on Workunit.

The only way to restart a failed model, is if you have a backup from before the failure.

ID: 17065 · Report as offensive     Reply Quote
John Perko

Send message
Joined: 3 Sep 04
Posts: 9
Credit: 582,919
RAC: 0
Message 17067 - Posted: 8 Nov 2005, 21:48:39 UTC - in response to Message 17065.  

Thanks. How does the server know it has failed? Does it flag it as one to be re-issued after a certain interval in which there is no communication from the client about that particular WU?
ID: 17067 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 17068 - Posted: 8 Nov 2005, 22:25:47 UTC

Look at your list of models. Any with \"Client error\" have failed and have been noted as such by the server.
After about 6 weeks without a trickle, a model is considered abandoned.

ID: 17068 · Report as offensive     Reply Quote

Questions and Answers : Preferences : Notifying the project of a crashed model?

©2024 climateprediction.net