[Tile-serving] [osm2pgsql] large postgres shared_buffers slows down import (#163)

[jeffjanes]

Yes, FlushRelationBuffers can be somewhat slow with very large shared_buffers.  However, it should also be called so rarely as to make no difference in most cases.  What is the import doing that causes this to happen in a loop?  Is it inserting a row into a table, and then truncating the table, over and over again once per relation?  

Also, is 9.3.2 the version of PostgreSQL you are using?  There have been changes in this area over recent releases, so it is important to know the version.

I wonder if this is new (either new to PostgreSQL or new to osm2pgsql) because I haven't noticed it before and this is the type of thing I look for.   Can you give the exact command line you used for the import?

---
Reply to this email directly or view it on GitHub:
https://github.com/openstreetmap/osm2pgsql/issues/163#issuecomment-47119887
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tile-serving/attachments/20140625/7793d62f/attachment.html>