[Live-demo] disc space

Angelos Tzotsos gcpp.kalxas at gmail.com
Thu Aug 1 16:37:40 PDT 2013


Update: from 10580 on there is a tmp disk space log, which can be used 
to check the tmp disk space before and after an installer. This space 
should be subtracted from the disk usage.

For example Kosmo:
disk usage (after - before): 8569 - 7945 = 624MB
but tmp log (after - before): 685 - 460 = 225MB Those are included in 
the above disk usage and the disk space is cleaned at the end of the 
build process.
Kosmo disk usage without /tmp = 624-225= 399MB (still too high)

Rasdaman:
disk usage (after - before): 13949 - 13180 = 769MB
tmp usage (after - before): 2328-1728=600MB
Rasdaman disk usage without /tmp: 169MB

GeoNetwork:
disk usage (after - before): 7945 - 7371= 574MB
tmp usage (after - before): 460 - 138=322 MB
GeoNetwork disk usage without /tmp: 252MB

Udig:
disk usage (after - before): 9041 - 8569=472MB
tmp usage (after - before): 925 - 685=240MB
Udig disk usage without /tmp: 232MB

I will try to confirm within the iso if possible.

Angelos

On 08/01/2013 08:36 PM, Angelos Tzotsos wrote:
> I think that the problem is that tmp disk space is also included in each
> step.
> So if an application downloads much data in /tmp and does not clean up
> (because the setdown.sh will clean /tmp at  the end), the size goes up.
> Remember we do a df in chroot so we get all the host filesystem info (the
> build server does not do anything else during build)
>
> Perhaps a du -sk /tmp before and after each package installation would help
> determine how much data are actually installed (/tmp excluded)
>
>
>
>
>
> On Thu, Aug 1, 2013 at 8:11 PM, <maplabs at light42.com> wrote:
>
>>
>> On 1 Aug 2013 13:04:18 -0400, Stefan Steiniger <sstein at geo.uzh.ch> wrote:
>> Hi,
>>
>>> so I see that Kosmo takes a lot of space (600+MB).
>>>
>> I am looking at this chart, too. I see Kosmo taking 143M on disk, in
>> /usr/lib/Kosmo-3.0
>>
>> I think this chart has some kind of basic problem.. this makes no sense to
>> me.. more when I get the 560 build
>>
>>    -Brian
>>
>>
>>   > Wonder why - if the core of Kosmo was actually taken from JUMP GIS
>>
>>> (12MB) (look at the size of OpenJUMP: < 100MB). So maybe they use lots of
>>> libraries for being able to read certain file types (gdal? mrsid/ecw-raster
>>> drivers? ), and each different per OS (win vs Linux vs MacOsx). Or do they
>>> have included lots of data.
>>> So, it is somehow surprising when seeing disk space use if Komso in
>>> comparison to gvSIG and OpenJUMP.
>>> cheers,
>>> stefan
>>>
>>> Am 01.08.13 08:09, schrieb Hamish:
>>>> Hi,
>>>>
>>>> Hamish wrote:
>>>>
>>>>>    I will work up a new usage chart for tomorrow's meeting. >
>>>> caveats/notes as these disk space results are not definitive:
>>>>
>>>>    - OSSIM is missing in this build
>>>>    - the first of e.g. grass or qgis to run will install many of the
>>> other's dependency packages, so will unjustly seem larger/slower. >   -
>>> /tmp usage is included here, but not on the disc. so some may be up to
>>> doubled. >   - probably some other misleading/confounding variables in it
>>>> but it's good for a general idea of what's going on. So to the plots---
>>>>
>>>>
>>>> Disk space used by each install script as a time-series, sorted
>>>> by biggest hog first, and a whisker plot of the distribution:
>>>>
>>>> http://muck.otago.ac.nz/~**hamish/osgeo/disk_usage_**timeseries.png<http://muck.otago.ac.nz/~hamish/osgeo/disk_usage_timeseries.png>
>>>> http://muck.otago.ac.nz/~**hamish/osgeo/disk_usage_**sorted.png<http://muck.otago.ac.nz/~hamish/osgeo/disk_usage_sorted.png>
>>>> http://muck.otago.ac.nz/~**hamish/osgeo/disk_usage_**whisker.png<http://muck.otago.ac.nz/~hamish/osgeo/disk_usage_whisker.png>
>>>>
>>>>
>>>> Time used by each install script as a time-series, sorted
>>>> by longest running first, and a whisker plot of the distribution:
>>>>
>>>> http://muck.otago.ac.nz/~**hamish/osgeo/time_to_install_**
>>> timeseries.png<http://muck.otago.ac.nz/~hamish/osgeo/time_to_install_timeseries.png>
>>>> http://muck.otago.ac.nz/~**hamish/osgeo/time_to_install_**sorted.png<http://muck.otago.ac.nz/~hamish/osgeo/time_to_install_sorted.png>
>>>> http://muck.otago.ac.nz/~**hamish/osgeo/install_time_**whisker.png<http://muck.otago.ac.nz/~hamish/osgeo/install_time_whisker.png>
>>>>
>>>> (suggest 'make -j 4' or so might be help for the compiles)
>>>>
>>>>
>>>> have at it,
>>>> Hamish
>>>>
>>>> ______________________________**_________________
>>>> Live-demo mailing list
>>>> Live-demo at lists.osgeo.org
>>>> http://lists.osgeo.org/**mailman/listinfo/live-demo<http://lists.osgeo.org/mailman/listinfo/live-demo>
>>>> http://live.osgeo.org
>>>> http://wiki.osgeo.org/wiki/**Live_GIS_Disc<http://wiki.osgeo.org/wiki/Live_GIS_Disc>
>>>>
>>> ______________________________**_________________
>>> Live-demo mailing list
>>> Live-demo at lists.osgeo.org
>>> http://lists.osgeo.org/**mailman/listinfo/live-demo<http://lists.osgeo.org/mailman/listinfo/live-demo>
>>> http://live.osgeo.org
>>> http://wiki.osgeo.org/wiki/**Live_GIS_Disc<http://wiki.osgeo.org/wiki/Live_GIS_Disc>
>>>
>>>
>>>
>> ______________________________**_________________
>> Live-demo mailing list
>> Live-demo at lists.osgeo.org
>> http://lists.osgeo.org/**mailman/listinfo/live-demo<http://lists.osgeo.org/mailman/listinfo/live-demo>
>> http://live.osgeo.org
>> http://wiki.osgeo.org/wiki/**Live_GIS_Disc<http://wiki.osgeo.org/wiki/Live_GIS_Disc>
>>
>
>


-- 
Angelos Tzotsos
Remote Sensing Laboratory
National Technical University of Athens
http://users.ntua.gr/tzotsos



More information about the Live-demo mailing list