Page 1 of 1

The size of it!

Posted: Fri Mar 24, 2006 7:23 am
by UBT - Mikee
I think I'm gonna get a problem with this soon.

Just had a look at the size of the Climate predition directory and at the moment it's at 1.10 GB (1,182,774,784 bytes).

453 MB (475,641,344 bytes) was created over 2 days (3232 files over March 14 and March 15) in directory c/program files/boinc/projects/climate..../sulphur...../tmp

The problem is that I'm only on 9.8% through a unit and if this happens 30 more times then the hard drive is going to be full. The back up drive is all ready at 92% so I won't be able to back up that directory much more anyway.

I know tmp directories can usually be deleted but I want to be sure this will be OK. So, anyone know if I can delete this huge tmp dir and/or ignore it in backups?

Re: The size of it!

Posted: Fri Mar 24, 2006 9:28 am
by UBT - Timbo
mikeejones1 wrote:I think I'm gonna get a problem with this soon.(snip) I know tmp directories can usually be deleted but I want to be sure this will be OK. So, anyone know if I can delete this huge tmp dir and/or ignore it in backups?
The last time I crunched a CPDN WU, (some time ago), I had a similar problem.

And IIRC, the project was suggesting at the time, that you retain the stuff the WU generates, in case later on during the crunch, your WU crashes. Then you can zip up the work done to date and send to it them. This is "just in case" the model you were working on was returning a "doomsday" scenario. (IIRC, there's only one WU of each type - so there's no "back-up", like there is with other projects, where multiple members get the same WU).

As times gone on, I think CPDN has become so successful that the chances are that, if you have a model crash, somewhere out there, then another model will be similar so it's less of a issue.


HOWEVER. I don't know if CPDN checks itself during each trickle as to whether all these files are "intact", so by deleting them, you might be causing the WU to fail.

Might be an idea to check on the timestamps for each set of files - if not "accessed" recently, then you are probably safe to delete it.

regards,
Tim

Posted: Fri Mar 24, 2006 8:48 pm
by Robert Boon
I think I'm gonna get a problem with this soon.

Just had a look at the size of the Climate predition directory and at the moment it's at 1.10 GB (1,182,774,784 bytes).

453 MB (475,641,344 bytes) was created over 2 days (3232 files over March 14 and March 15) in directory c/program files/boinc/projects/climate..../sulphur...../tmp

The problem is that I'm only on 9.8% through a unit and if this happens 30 more times then the hard drive is going to be full. The back up drive is all ready at 92% so I won't be able to back up that directory much more anyway.
I suspect the rate of increase of the Climate Prediction directory will slow down as you get further. I am at 18.7% on a Sulphur model run and the directory only stands at 1.83 GB.

Posted: Wed Apr 05, 2006 6:29 pm
by UBT - John
Got to 26.6% on a sulphur model at 2.08GB.

(I went on my hols for a week and it disappeared! Now 6.74% into the new model)

Posted: Thu Apr 06, 2006 12:42 am
by RodEllery
Climate prediction keeps it's finished WUs in case they need them. Search their forums and find out how to delete completed WUs.

That said each active WU takes a lot of diskspace.

--
Rod

Posted: Thu Apr 06, 2006 9:18 am
by UBT - Halifax-lad
It doesn't do that anymore they disabled that feature a long time ago, have just confirmed it by checking the BOINC files and there are no CPDN files there except the current one