climateprediction.net home page
What happened to this wu?

What happened to this wu?

Message boards : Number crunching : What happened to this wu?
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile ThePhantom86
Avatar

Send message
Joined: 6 Aug 04
Posts: 42
Credit: 3,445,139
RAC: 950
Message 25126 - Posted: 16 Nov 2006, 2:28:58 UTC

ID: 25126 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 25128 - Posted: 16 Nov 2006, 3:28:05 UTC

It was crashed by computer ID = 101052, and then the dataset was re-issued to you.
This is normal. As well as the original issue up to 4 re-issues can be made.

Interestingly, the failing computer should not be receiving models, and I\'m going to email Carl about it.

ID: 25128 · Report as offensive     Reply Quote
Profile ThePhantom86
Avatar

Send message
Joined: 6 Aug 04
Posts: 42
Credit: 3,445,139
RAC: 950
Message 25148 - Posted: 17 Nov 2006, 0:40:56 UTC

But I\'m not crunching it either. It says the state is \"over\" for it.
ID: 25148 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 25149 - Posted: 17 Nov 2006, 3:09:58 UTC

The state only says \"over\" for Tim Walter\'s computer.

For your computer, ID = 500980, Intel Celeron CPU 1.70GHz, it\'s still running, and last trickled on: 16 Nov 2006 3:12:07 UTC, which was it\'s second trickle.

And this computer only has half the recommended minimum amount of memory, so I hope that you\'re making regular backups.


Backups: Here
ID: 25149 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 25154 - Posted: 17 Nov 2006, 21:46:46 UTC
Last modified: 17 Nov 2006, 21:50:16 UTC

Just as an addendum, a lot of the software fields aren\'t used here for the same things as in the other projects.

For instance, on the Work unit pages, the field errors has: Too many total results
This is because, here, the field records the number of trickles, thinking that they\'re results for the same wu returned by too many people.

And the field Server state is used to say if the dataset should be re-issued, (up to 4 times is possible). If it says over, then it can mean that the processing is over for a model, (but only if it also has certain words in both of the next 2 columns),
or it can mean that that dataset is not to be re-issued.

And in the case of your copy of the dataset, it means the latter.
So If your computer crashes the model, then the dataset is dumped.

And on the server status page, the field Workunits waiting for validation is just the number of WUs returned for one reason or another.
There is no validation on this project.

ID: 25154 · Report as offensive     Reply Quote
Profile ThePhantom86
Avatar

Send message
Joined: 6 Aug 04
Posts: 42
Credit: 3,445,139
RAC: 950
Message 25162 - Posted: 18 Nov 2006, 4:23:37 UTC

I don\'t mean to argue Les, but I\'m not working on 5733482. My computer is working on 5749260.
ID: 25162 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 25163 - Posted: 18 Nov 2006, 5:23:47 UTC

No worries. I find that chasing back and forth through all of these numbers on other people\'s records becomes confusing after a while. Like about 5 seconds. :)

OK.
I don\'t think that \'we\' can help with this. You\'ll need to look through the messages for the time period of when 5749260 showed up, and see what they say about the other one. (Perhaps in stdoutdae.txt if you need to go back further than those currently in the Messages tab.)

For some reason 5733482 may have been abandoned, without this being reported back to the server.
If you never received this model, there is another possibility:

The 13th of October is the day that there was a server problem after a software upgrade, which caused thousands of models to be issued in minutes to any computer where BOINC requested a new model.
Except that the models were never actually sent; their records were just marked as having been sent, and they were removed from the data pool, which quickly ended up empty. And the affected people had their Account pages filled with the numbers of models which they never received. Hundreds of them.

So you may have been involved at the start or end of this, getting one real model and one phantom model.

The 13th was also a Friday, and Carl had to spend almosty all of his two days off chasing down the problem, fixing it, and then generating new models. Definitely not happy about it.

ID: 25163 · Report as offensive     Reply Quote

Message boards : Number crunching : What happened to this wu?

©2024 climateprediction.net