Hi Nuno, On 10 October 2013 22:18, Nuno Miguel Lourenço < nuno-miguel-loure...@telecom.pt> wrote:
> Hi,**** > > ** ** > > I’m updating a OpenStreetMaps based DB, but we are doing incremental steps > to the DB data.**** > > We are importing just a few countries and then proceed to continents and > at the end the planet file.**** > > ** ** > > We are already using the country data! NOTE: The countries are not > neighbors!**** > > We are importing some Europe countries from > http://download.geofabrik.de/europe.html getting the several *.osm.pbf *files > for the ones we want!**** > > ** ** > > We are coming across an issue…**** > > We are getting a lot of **** > > *Detail: Key (id)=(xxxxx) is duplicated*.; nested exception is > org.postgresql.util.PSQLException: ERROR: could not create unique index "* > pk_relations*"**** > > We are getting this for the relations table, for the nodes, ways tables > also!!! > It sounds like your .osm.pbf files contain duplicate data. The countries may not be neighbours, but there may still be data relationships that connect them. I assume (providing your exact command line and exception messages would be very helpful) you're trying to use a task like --write-pgsql which assumes that the database is empty. In your case you're trying to run it multiple times against a non-empty database which will fail if the input files contain duplicate data. > **** > > ** ** > > Is there any way we can avoid this? > Perhaps. Combining all .osm.pdf files together into a single file and remove duplicate data before importing should fix it. > **** > > We tried to remove the duplicates on the *osmosisUpdate* function but it > did not work due to this being called after all the pushes of data to the > DB. > The osmosisUpdate function is only called by the --write-pgsql-change task, but I don't believe you're using that task ... > **** > > ** ** > > Is there any other way to do this with osmosis? > Merge all the .osm.pbf files together before importing. > **** > > Merging the several .*osm.pbf* files to generate a unique one is not a > solution. > Okay. Can you please explain why this is not an option? Brett
_______________________________________________ osmosis-dev mailing list osmosis-dev@openstreetmap.org https://lists.openstreetmap.org/listinfo/osmosis-dev
