climateprediction.net home page
Disk Space Problems

Disk Space Problems

Questions and Answers : Preferences : Disk Space Problems
Message board moderation

To post messages, you must log in.

AuthorMessage
Vaipen

Send message
Joined: 30 Dec 06
Posts: 13
Credit: 1,642,756
RAC: 0
Message 43391 - Posted: 8 Nov 2011, 12:09:30 UTC

CP.net constantly wants more disk space. I don't understand why. I have set it to use
- 12 GB
- leave 0,2 GB free.
- 25% of my disk space.
- 56% of total disk size and
- 50% of total disk space, when active
- 95% of total disk space when inactive.

HADAM3P needs 1907 MB and it says I have 1313 MB free, so it needs 549 MB.
Coupled Model Full Res Ocean needs 486 MB more.

With these settings how can it still be needed more space? And what setting should I change?

ID: 43391 · Report as offensive     Reply Quote
Profile Dave Jackson
Volunteer moderator

Send message
Joined: 15 May 09
Posts: 4314
Credit: 16,380,160
RAC: 3,563
Message 43392 - Posted: 8 Nov 2011, 13:34:33 UTC - in response to Message 43391.  

My guess is it will take which ever of those option is the most restrictive.
My settings are,
Use at most 10GB disk space
Leave at least 5GB free
Use at most 80% of total disk space.
Tasks checkpoint to disk at most every 60s
Use at most 50% of page file (swap space)
The version of BOINC I use only has different settings for memory, not disk space depending on whether the computer is active or not. I don't worry too much about that, having expanded my ram to 4GB. If this ever becomes an issue I will increase it further. What is your total disk space? If you are running on a small partition the 25% of disk space may be the issue?

Dave
ID: 43392 · Report as offensive     Reply Quote
David Motz

Send message
Joined: 6 Mar 05
Posts: 2
Credit: 144,912
RAC: 0
Message 43512 - Posted: 4 Dec 2011, 16:30:32 UTC

I'm running 3 projects, climateprediction.net, seti@home and World Community Grid. I have 10 gig allocated for BOINC and Climateprediction is using 9.87 gig, so World Community Grid is left out in the cold. I've looked at Climateprediction settings and I don't see any way to throttle it back on the disk.

How can I make it play nice with my other project? Thanks
ID: 43512 · Report as offensive     Reply Quote
Profile geophi
Volunteer moderator

Send message
Joined: 7 Aug 04
Posts: 2167
Credit: 64,403,322
RAC: 5,085
Message 43513 - Posted: 4 Dec 2011, 17:34:50 UTC - in response to Message 43512.  

You have two hadam3p SAF tasks currently allocated to that PC.

hadam3p_saf_7kia_2001_1_007583515

and

hadam3p_saf_7lfx_2002_1_007584726

and you definitely don't want to delete files associated with those model runs.

Take a look in the boinc data directory, projects, climateprediction.net subdirectory and see if there are any other folders that start with had or famous. If so, you could delete those other folders. When a model crashes, it's supposed to delete the folder on the disk for that model run. However, that doesn't always happen.
ID: 43513 · Report as offensive     Reply Quote
David Motz

Send message
Joined: 6 Mar 05
Posts: 2
Credit: 144,912
RAC: 0
Message 43516 - Posted: 6 Dec 2011, 13:13:04 UTC - in response to Message 43512.  

I finally just detached from CP.net. If someone can tell me how to run it without it hogging my entire BOINC disk allocation, I will reattach, but I won't allow it to take over the way it has.
ID: 43516 · Report as offensive     Reply Quote
Profile Iain Inglis
Volunteer moderator

Send message
Joined: 16 Jan 10
Posts: 1079
Credit: 6,907,363
RAC: 6,402
Message 43520 - Posted: 6 Dec 2011, 16:01:14 UTC - in response to Message 43516.  

I finally just detached from CP.net. If someone can tell me how to run it without it hogging my entire BOINC disk allocation, I will reattach, but I won't allow it to take over the way it has.

Disk resources used by CPDN include the common project files, running models and crashed models. Crashed models can be removed as described earlier by geophi. The common files can be reduced somewhat by a project reset (when no models are in progress, otherwise they'll be aborted); new common files will be re-downloaded but old applications, for example, will not - so a small amount of disk space can be recovered. There is nothing that can be done about running models: they use as much space as they need and if there isn't enough space then they won't run.

So, assuming you have already taken geophi's advice, the bottom line is that you can run one model at a time or if that is still too much then none at all. These models are large by distributed computing standards and they won't suit every BOINC user.
ID: 43520 · Report as offensive     Reply Quote
Vaipen

Send message
Joined: 30 Dec 06
Posts: 13
Credit: 1,642,756
RAC: 0
Message 43554 - Posted: 12 Dec 2011, 20:02:31 UTC

Which is fairly annoying and not bright considering the importance of this project.

I got it too work now, running one model but don't ask me how and what I did. I increased most space assignments I think. The difference between local and settings online I still can't see clearly. Even when I upped local settings it didn't seem to have enough allocations.

So I can delete had and famous? Or the others. Geophi isn't particularly clear on that.
ID: 43554 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 43555 - Posted: 12 Dec 2011, 20:31:01 UTC - in response to Message 43554.  

Delete the folders of ALL failed models that don't currently appear in the Tasks tab.


Backups: Here
ID: 43555 · Report as offensive     Reply Quote

Questions and Answers : Preferences : Disk Space Problems

©2024 climateprediction.net