climateprediction.net home page
No Work?!

No Work?!

Message boards : Number crunching : No Work?!
Message board moderation

To post messages, you must log in.

AuthorMessage
SuperSluether

Send message
Joined: 6 Jul 14
Posts: 11
Credit: 367,660
RAC: 0
Message 50335 - Posted: 29 Sep 2014, 1:09:58 UTC

I know CPDN had an issue a while ago where everything was down for a while, but after they recovered they had work to send out. Just a few weeks (days?) ago, they had about 50,000 tasks from one application, and everything else was empty. Now there are 0 tasks ready to send out! What's going on?!
ID: 50335 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 50336 - Posted: 29 Sep 2014, 3:42:09 UTC - in response to Message 50335.  

As can be seen from reading other threads, there were numerous problems with the "short" models. The remainder have been removed.

More work will become available when the researchers from the various external climate centres provide more, and possibly after some more testing.

ID: 50336 · Report as offensive     Reply Quote
Profile Dave Jackson
Volunteer moderator

Send message
Joined: 15 May 09
Posts: 4314
Credit: 16,377,675
RAC: 3,657
Message 50337 - Posted: 29 Sep 2014, 7:16:43 UTC - in response to Message 50336.  

As can be seen from reading other threads, there were numerous problems with the "short" models. The remainder have been removed.


Shame, they all ran to completion on my linux system.

To reduce the chances of running out of work or to be realistic how often it happens, make sure that under <tools> <network usage> you have both minimum and maximum work buffer to 10 days which is the maximum allowed.
ID: 50337 · Report as offensive     Reply Quote
SuperSluether

Send message
Joined: 6 Jul 14
Posts: 11
Credit: 367,660
RAC: 0
Message 50343 - Posted: 30 Sep 2014, 0:03:50 UTC - in response to Message 50337.  


To reduce the chances of running out of work or to be realistic how often it happens, make sure that under <tools> <network usage> you have both minimum and maximum work buffer to 10 days which is the maximum allowed.


I just re-attached the project when I found they had no work. I set my network buffer to 0.1 max additional 0.5 because I run many projects, and I don't want BOINC using too much disk space. It's already using almost 20 GB when certain projects run.
ID: 50343 · Report as offensive     Reply Quote
SuperSluether

Send message
Joined: 6 Jul 14
Posts: 11
Credit: 367,660
RAC: 0
Message 50344 - Posted: 30 Sep 2014, 0:04:59 UTC - in response to Message 50343.  

I also just noticed that on the Server Status page, there is no work generator server. Does this mean that work packets are created by actual scientists instead of generated by computers?
ID: 50344 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 50345 - Posted: 30 Sep 2014, 2:28:18 UTC - in response to Message 50344.  

There are a lot of BOINC functions that aren't used by cpdn.

Work is created by the researchers, probably using computers, and supplied to cpdn for testing/placing on the download server.

The data in the "zip 13" file is used by a program to automatically create the next part of a long sequence, and this dataaset is also placed onto the download server.
And it's why zip 13's are returned to their own server, where the regen program resides.

ID: 50345 · Report as offensive     Reply Quote
Profile Dave Jackson
Volunteer moderator

Send message
Joined: 15 May 09
Posts: 4314
Credit: 16,377,675
RAC: 3,657
Message 50346 - Posted: 30 Sep 2014, 7:11:55 UTC

I don't want BOINC using too much disk space. It's already using almost 20 GB when certain projects run.


It is worth checking whether you have any model folders left in your BOINC/Projects/CPDN folder. This often occurs when tasks crash but at least on Linux boxes it was also happening for the Short models even when they completed successfully. When I noticed this, I was able to save 4GB of space by deleting the folders for reported tasks. By the time the ones I have in the queue are finished there will be another 3GB to clear.

When not started CPDN tasks take up very little space - between 140 - 180MB currently on this computer.
ID: 50346 · Report as offensive     Reply Quote
Profile Blurf

Send message
Joined: 13 Jun 08
Posts: 6
Credit: 1,372,493
RAC: 0
Message 50349 - Posted: 1 Oct 2014, 1:46:21 UTC

No work and my BOINC is working just dandy.
ID: 50349 · Report as offensive     Reply Quote
SuperSluether

Send message
Joined: 6 Jul 14
Posts: 11
Credit: 367,660
RAC: 0
Message 50350 - Posted: 1 Oct 2014, 4:19:56 UTC - in response to Message 50346.  

I don't want BOINC using too much disk space. It's already using almost 20 GB when certain projects run.


It is worth checking whether you have any model folders left in your BOINC/Projects/CPDN folder. This often occurs when tasks crash but at least on Linux boxes it was also happening for the Short models even when they completed successfully. When I noticed this, I was able to save 4GB of space by deleting the folders for reported tasks. By the time the ones I have in the queue are finished there will be another 3GB to clear.

When not started CPDN tasks take up very little space - between 140 - 180MB currently on this computer.


No, the ATLAS and vLHC use a lot of memory, and I had about 6GB for ATLAS at one point. It went back down now that BOINC switched to other projects, but I know that this project can take up space when it's running.
ID: 50350 · Report as offensive     Reply Quote
Eirik Redd

Send message
Joined: 31 Aug 04
Posts: 391
Credit: 219,888,554
RAC: 1,481,373
Message 50369 - Posted: 2 Oct 2014, 20:28:28 UTC

Today (whatever that means globally)

A small batch of PNW happened.

Got a few of the small batch.

Keep my boxes open to CPDN -- if there's work, I'll take it.

Look at the "science" thread.

Keep on crunchin
ID: 50369 · Report as offensive     Reply Quote

Message boards : Number crunching : No Work?!

©2024 climateprediction.net