climateprediction.net home page
Confused Results

Confused Results

Questions and Answers : Unix/Linux : Confused Results
Message board moderation

To post messages, you must log in.

AuthorMessage
squaregoldfish

Send message
Joined: 22 Aug 06
Posts: 6
Credit: 2,836,837
RAC: 0
Message 26147 - Posted: 18 Jan 2007, 19:14:18 UTC

Hopefully someone in charge can sort this out for me...

I\'ve been crunching a work unit since mid December (ID 5906707). On the 15th, the process accidentally got killed, so BOINC thought the processing had finished and returned an error result. I then received a brand new work unit (5888601).

Knowing that the killed process was actually still OK, I restored from backup and have returned to crunching it as before. The thing is, my Results page reports the work unit as In Error. If I continue processing it, will the server sort itself out, or should I just get a new work unit?

As a side note, the other units on my account are all ones that I have not computed at all. If these need to be put back in the pool of work to be done, please do whatever is necessary:

5392510 - Aborted on my previous machine - it was too slow
5899290 - No idea what this is
5888601 - Sent to me after the incident above.

Thanks in advance for any help,
Steve.
ID: 26147 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 26149 - Posted: 18 Jan 2007, 20:27:35 UTC

BOINC, which is the software on the servers as well as user\'s computers, doesn\'t allow for a second go at processing a work unit, so the message that gets to the server first, stays there. (This \"no redoing\" is a requirement of the other projects. We\'re different.)
So you\'ll always have that error message. And when/if the model completes, you\'ll get another message saying that the final results have been rejected because of the previous error.
Ignore all of these messages. The project doesn\'t use them.
As long as the data files get to the server OK, then the results will be OK.

All models that fail to contact the server for about 6 weeks are considered lost, and the data sets for them are automatically marked for possible re-issue.

ID: 26149 · Report as offensive     Reply Quote
squaregoldfish

Send message
Joined: 22 Aug 06
Posts: 6
Credit: 2,836,837
RAC: 0
Message 26153 - Posted: 18 Jan 2007, 21:55:44 UTC - in response to Message 26149.  

Thanks Les - I\'ll keep going with what I\'ve got.

Incidentally, does the BOINC error state mean that I won\'t get any more credit for the data I send in? It won\'t affect my ego too badly if I don\'t - I\'m just curious.

Steve.
ID: 26153 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 26158 - Posted: 18 Jan 2007, 22:27:37 UTC
Last modified: 18 Jan 2007, 22:28:03 UTC

The error messages are ignored by the project.

Credit is based on trickles.
Trickles are returned on about the 4th of December each year.
The credit program runs once per day, because of the load on the servers.
So you\'ll keep getting credits until the model completes or fails permanently.

The only time you won\'t get any credits for trickles returned, is when you run from a backup. Then you have to wait until the model has reach new territory. No \"double dipping\" allowed.

ID: 26158 · Report as offensive     Reply Quote
squaregoldfish

Send message
Joined: 22 Aug 06
Posts: 6
Credit: 2,836,837
RAC: 0
Message 26162 - Posted: 18 Jan 2007, 23:33:39 UTC - in response to Message 26158.  

Thanks for your quick and useful answers - much appreciated. I\'m also pleased to see that your software can cope with idiots like me who keeping messing things up...
ID: 26162 · Report as offensive     Reply Quote

Questions and Answers : Unix/Linux : Confused Results

©2024 climateprediction.net