Merge remote-tracking branch 'upstream/pull/2893'

[rails.git] / CONFIGURE.md

diff --git a/CONFIGURE.md b/CONFIGURE.md
index 91a64119ed33cb5d7ccf057ba266a6684f323bce..598ea3f65d9207c25a3d7f6380c9b5319c9f2da3 100644 (file)
--- a/CONFIGURE.md
+++ b/CONFIGURE.md
@@ -128,6 +128,7 @@ If you want to deploy The Rails Port for production use, you'll need to make a f
 * Passenger will, by design, use the Production environment and therefore the production database - make sure it contains the appropriate data and user accounts.
 * Your production database will also need the extensions and functions installed - see [INSTALL.md](INSTALL.md)
 * The included version of the map call is quite slow and eats a lot of memory. You should consider using [CGIMap](https://github.com/zerebubuth/openstreetmap-cgimap) instead.
 * Passenger will, by design, use the Production environment and therefore the production database - make sure it contains the appropriate data and user accounts.
 * Your production database will also need the extensions and functions installed - see [INSTALL.md](INSTALL.md)
 * The included version of the map call is quite slow and eats a lot of memory. You should consider using [CGIMap](https://github.com/zerebubuth/openstreetmap-cgimap) instead.

-* The included version of the GPX importer is slow and/or completely inoperable. You should consider using [the high-speed GPX importer](https://git.openstreetmap.org/gpx-import.git/).

 * Make sure you generate the i18n files and precompile the production assets: `RAILS_ENV=production rake i18n:js:export assets:precompile`
 * Make sure the web server user as well as the rails user can read, write and create directories in `tmp/`.
 * Make sure you generate the i18n files and precompile the production assets: `RAILS_ENV=production rake i18n:js:export assets:precompile`
 * Make sure the web server user as well as the rails user can read, write and create directories in `tmp/`.

+* If you want to use diff replication then you might want to consider installing the shared library special SQL functions for the `xid_to_int4` function. A pure SQL version is available, but may become a performance issue on large databases with a high rate of changes. Note that you will need a version of PostgreSQL < 9.6 (yes, _less than_) to use `xid` indexing, whether pure SQL or shared library.
+* If you expect to serve a lot of `/changes` API calls, then you might also want to install the shared library versions of the SQL functions.