[OSM-dev] world wide PBF exports corrupt

Peter K peathal at yahoo.de
Mon Oct 20 06:31:42 UTC 2014


Hi Paul, Hi Simon,

this clarifies the daily vs. weekly and thanks for the osmupdate tool,
should I add this to the wiki? And is there a tool where I can verify
the new PBF or is my only guess the size?

Regarding the corruption indeed osm looks good, but the mirrors don't:
ftp://ftp.spline.de/pub/openstreetmap/pbf/ (only 17gb)
http://ftp.snt.utwente.nl/pub/misc/openstreetmap/ (only 16gb)
http://ftp.heanet.ie/mirrors/openstreetmap.org/pbf/ (only 6gb)

over 1 week old:
http://ftp.osuosl.org/pub/openstreetmap/pbf/
http://ftp5.gwdg.de/pub/misc/openstreetmap/planet.openstreetmap.org/pbf/

So I think there is something wrong in the common toolchain (?). I would
prefer to use mirros to take off load from the main servers and have
more local ones.

Kind Regards,
Peter.


-- 
GraphHopper.com - Fast & Flexible Road Routing



On 20.10.2014 01:55, Paul Norman wrote:
>
> On 10/19/2014 3:04 PM, Peter K wrote:
>> it looks like all pbf export servers I know have no recent or corrupt
>> pbf exports (small file size of 17gb or 6gb instead of 27gb):
>> http://wiki.openstreetmap.org/wiki/Planet.osm
>>
>> Is this a known issue of some export tool? Or do you know an
>> alternative?
> http://planet.openstreetmap.org/pbf/planet-141015.osm.pbf is 26GB,
> which looks normal to me.
>
> The mirrors are that - mirrors, not export servers.
>
> Before using the planet file, you can run Osmupdate[1] on it, which
> takes under an hour on most hardware and will give you an up to the
> hour version of the OpenStreetMap database.
>> (Also it looks like there are no daily export servers anymore?)
> We've never had planet exports generated daily. If you need a complete
> planet file every day, the standard means is to download one and then
> update it every day, which uses far less bandwidth than downloading a
> new one. This is what is done for taginfo, and what external services
> like OSRM do.
>
> [1]: http://wiki.openstreetmap.org/wiki/Osmupdate
>





More information about the dev mailing list