[OSM-dev] database growth on updated partial import

Frederik Ramm frederik at remote.org
Tue Aug 24 00:02:54 BST 2010


Stephan Knauss wrote:
> Autovacuum is on

Unlikely, but check if you have any messages about max_fsm_pages being 
too small in your Postgres log. If that happens it means that the normal 
vacuum process is not enough (try a vacuum full and see if size goes 
down; if yes, increase your max_fsm_pages).

> It seams to be planet_osm_ways growing. It also contains lots of ways 
> that are completely outside the bounding box.

Unsure why that can happen. The bbox filter is only applied to nodes 
(because only they have a location). Maybe all created/modified ways are 
imported even if they refer to non-existing nodes entirely?

> Would it be safe to delete rows with geometry outside the bounding box? 

I think so, but the planet_osm_way table does not have a geometry. You 
would have to write a query that eliminates those ways which have 
dangling references to nodes I think. Sounds slow ;)


Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00'09" E008°23'33"

More information about the dev mailing list