Most things in life are a journey and the destination of this particular journey was to try and create a custom map style that represented the unique features and challenges of Tandale.
Which meant I needed to download and install TileMill, an interactive map design tool.
Which meant I needed to learn Carto, the CSS-like language for map styling.
Which meant I looked for a template project so I didn’t have to start from scratch.
Which meant I found OSM Bright.
Which meant I needed to start small and find a map extract of Tanzania to work with.
Which brings me to the starting point of the journey and the reason for this post in the first place.
When I normally need to install UNIX-y command line and server tools I turn to Homebrew, the tool set that “installs the stuff you need that Apple didn’t”. Homebrew supports installing both PostgreSQL and PostGIS but a bit of background research showed that installing these on Lion and on Mountain Lion could be problematic. A bit of further research soon turned up Postgres.app, which claims to be “the easiest way to run PostgreSQL on the Mac”. Postgres.app is a single shot installer which wraps PostgreSQL and PostGIS into an easy to install and run self contained environment.
I’m a big fan of this approach to a software development environment. All of the stuff I’ve put up on GitHub and on WordPress.org has been written using MAMP, the single shot installer which wraps up Apache, MySQL and PHP on the Mac so Postgres.app gave instant appeal to me. So, download, install, start.
Next I found an OSM map extract of Tanzania courtesy of GeoFabrik, which I also downloaded. Now to load the map into PostgreSQL. I made sure my shell’s
PATH pointed to the command line tools provided by Postgres.app by prepending
/Applications/Postgres.app/Contents/MacOS/bin to the
PATH defined in my
psql and created a database called
tanzania. So far so good.
$ psql psql (9.2.2) Type "help" for help. gary=# CREATE DATABASE tanzania; CREATE DATABASE gary=# \q
To load the map into the database I had a choice of two command line tools; Imposm or osm2pgsql. The latter of the two seemed to work out of the box according to the documentation so I used Homebrew to install this tool.
$ brew install osm2pgsql
Now to load the map …
$ osm2pgsql -c -G -U gary -d tanzania ~/Projects/maps/data/tanzania.osm.pbf osm2pgsql SVN version 0.81.0 (64bit id space) Using projection SRS 900913 (Spherical Mercator) Setting up table: planet_osm_point NOTICE: table "planet_osm_point" does not exist, skipping NOTICE: table "planet_osm_point_tmp" does not exist, skipping SELECT AddGeometryColumn('planet_osm_point', 'way', 900913, 'POINT', 2 ); failed: ERROR: function addgeometrycolumn(unknown, unknown, integer, unknown, integer) does not exist LINE 1: SELECT AddGeometryColumn('planet_osm_point', 'way', 900913, ... ^ HINT: No function matches the given name and argument types. You might need to add explicit type casts. Error occurred, cleaning up
The lack of the
AddGeometryColumn function was the clue here. Whilst Postgres.app may come with PostGIS, my custom database was lacking all the PostGIS functionality. So I deleted my initial database and tried to recreate it with the
template_postgis template, which also failed.
$ psql psql (9.2.2) Type "help" for help. gary=# DROP DATABASE tanzania; DROP DATABASE gary=# CREATE DATABASE tanzania TEMPLATE=template_postgis; ERROR: template database "template_postgis" does not exist gary=# \q
As Regina correctly pointed out in the comments, I didn’t really need to go through the manual process of loading the PostGIS template, the
create extension postgis command in
psql would have done this for me much quicker and elegantly, reducing the commands to setup my database to just two statements …
$ psql psql (9.2.2) Type "help" for help. gary=# CREATE DATABASE tanzania; CREATE DATABASE gary=# \connect tanzania; You are now connected to database "tanzania" as user "gary". tanzania=# CREATE EXTENSION postgis; CREATE EXTENSION gary=# \q
… simple when you know how.
So I needed to create the
template_postgis database from scratch, loading in the
spatial_ref_sys.sql SQL files and then recreate my custom database, based on the template contained in the
template_postgis database. The PostGIS SQL files are supplied as part of Postgres.app, if you know where to look for them; you’ll find them inside the app’s container in
$ createdb template_postgis $ createlang plpgsql template_postgis createlang: language "plpgsql" is already installed in database "template_postgis" $ psql -d template_postgis -f /Applications/Postgres.app/Contents/MacOS/share/contrib/postgis-2.0/postgis.sql SET BEGIN CREATE FUNCTION CREATE FUNCTION CREATE TYPE ... COMMIT $ psql -d template_postgis -f /Applications/Postgres.app/Contents/MacOS/share/contrib/postgis-2.0/spatial_ref_sys.sql BEGIN INSERT 0 1 ... COMMIT ANALYZE $ psql psql (9.2.2) Type "help" for help. gary=# CREATE DATABASE tanzania TEMPLATE=template_postgis; CREATE DATABASE gary=# \q
Now, at last, I was able to load my Tanzanian map.
$ osm2pgsql -c -G -U gary -d tanzania ~/Projects/maps/data/tanzania.osm.pbf osm2pgsql SVN version 0.81.0 (64bit id space) Using projection SRS 900913 (Spherical Mercator) Setting up table: planet_osm_point NOTICE: table "planet_osm_point" does not exist, skipping NOTICE: table "planet_osm_point_tmp" does not exist, skipping Setting up table: planet_osm_line NOTICE: table "planet_osm_line" does not exist, skipping NOTICE: table "planet_osm_line_tmp" does not exist, skipping Setting up table: planet_osm_polygon NOTICE: table "planet_osm_polygon" does not exist, skipping NOTICE: table "planet_osm_polygon_tmp" does not exist, skipping Setting up table: planet_osm_roads NOTICE: table "planet_osm_roads" does not exist, skipping NOTICE: table "planet_osm_roads_tmp" does not exist, skipping Allocating memory for dense node cache Allocating dense node cache in one big chunk Allocating memory for sparse node cache Sharing dense sparse Node-cache: cache=800MB, maxblocks=102401*8192, allocation method=3 Mid: Ram, scale=100 Reading in file: /Users/gary/Projects/maps/data/tanzania.osm.pbf Processing: Node(6820k 682.0k/s) Way(980k 16.90k/s) Relation(23580 1122.86/s) parse time: 89s Node stats: total(6820388), max(1910954191) in 10s Way stats: total(980191), max(180648305) in 58s Relation stats: total(23580), max(2409445) in 21s Committing transaction for planet_osm_point Committing transaction for planet_osm_line Committing transaction for planet_osm_polygon Committing transaction for planet_osm_roads Writing way (980k) Committing transaction for planet_osm_point Committing transaction for planet_osm_line Committing transaction for planet_osm_polygon Committing transaction for planet_osm_roads Writing relation (23569) Sorting data and creating indexes for planet_osm_point Sorting data and creating indexes for planet_osm_line Sorting data and creating indexes for planet_osm_polygon node cache: stored: 6820388(100.00%), storage efficiency: 50.68% (dense blocks: 637, sparse nodes: 6403164), hit rate: 99.45% Sorting data and creating indexes for planet_osm_roads Analyzing planet_osm_point finished Analyzing planet_osm_polygon finished Analyzing planet_osm_roads finished Analyzing planet_osm_line finished Copying planet_osm_point to cluster by geometry finished Copying planet_osm_roads to cluster by geometry finished Creating indexes on planet_osm_roads finished All indexes on planet_osm_roads created in 12s Completed planet_osm_roads Copying planet_osm_polygon to cluster by geometry finished Copying planet_osm_line to cluster by geometry finished Creating indexes on planet_osm_point finished All indexes on planet_osm_point created in 21s Completed planet_osm_point Creating indexes on planet_osm_polygon finished All indexes on planet_osm_polygon created in 28s Completed planet_osm_polygon Creating indexes on planet_osm_line finished All indexes on planet_osm_line created in 30s Completed planet_osm_line Osm2pgsql took 218s overall
One final gotcha awaited though. Restarting Postgres.app later that day made psql fail with an error.
$ psql psql: could not connect to server: No such file or directory Is the server running locally and accepting connections on Unix domain socket "/tmp/.s.PGSQL.5432"?
Although Postgres.app was running, it looked like the server wasn’t. Checking the system error logs via Console.app showed me that my newly populated database was running out of shared memory.
22/12/2012 11:05:44.319 com.heroku.postgres-service: FATAL: could not create shared memory segment: Cannot allocate memory 22/12/2012 11:05:44.319 com.heroku.postgres-service: DETAIL: Failed system call was shmget(key=5432001, size=3809280, 03600). 22/12/2012 11:05:44.319 com.heroku.postgres-service: HINT: This error usually means that PostgreSQL's request for a shared memory segment exceeded available memory or swap space, or exceeded your kernel's SHMALL parameter. You can either reduce the request size or reconfigure the kernel with larger SHMALL. To reduce the request size (currently 3809280 bytes), reduce PostgreSQL's shared memory usage, perhaps by reducing shared_buffers or max_connections. 22/12/2012 11:05:44.319 com.heroku.postgres-service: The PostgreSQL documentation contains more information about shared memory configuration. 22/12/2012 11:20:40.584 com.heroku.postgres-service: server starting
Thankfully this is a known problem; PostgreSQL is really a server application, not a laptop application. The default Mac configuration isn’t enough to support a medium sized PostgreSQL database, but adding the following configuration settings to
/etc/sysctl.conf, creating it via
sudo if it doesn’t already exist and rebooting solved that final problem.
kern.sysv.shmmax=1610612736 kern.sysv.shmall=393216 kern.sysv.shmmin=1 kern.sysv.shmmni=32 kern.sysv.shmseg=8 kern.maxprocperuid=512 kern.maxproc=2048
I now have a working PostgreSQL and PostGIS install, with a map loaded, which TileMill can access. Now all I need to do is learn Carto and actually make the map I originally set out to do … another learning journey has started.