This repository is a collection of Ansible roles for the configuration & provisioning of our tile server @ maps.zoondka.com.

Our tile server is a single robust machine composed of various tools & applications for all things vector tiles. It acts as a database for OSM & other map data, a vector tile generator, a database for generated tiles, and a server for both tiles and our zoondka-maps app which consumes them.

Here is an overview of our software stack:

• Data import & storage: osm2pgsql & PostgreSQL
• Tile generation and serving: Tilerator and our own fork of Kartotherian, Kartozoa
• Tile storage: Cassandra, as a single node cluster

For more details of the various components & their configurations, you can take a look through the roles directory.

Ansible
Vagrant (for dev VM)

We use this same Ansible playbook in production @ Zoondka, so it includes two different sets of variables - dev & production. For testing, feel free to use the default dev variables. I've included a Vagrantfile & you can get up & running with a test VM easily.

With Vagrant installed, you can easily setup a dev VM that mirrors our production environment @ maps.zoondka.com (excluding the actual hardware) . This is useful for testing the software stack:

git clone https://github.com/zoondka/maps.zoondka.com
cd maps.zoondka.com
vagrant up

With the VM up & running, you can re-provision it anytime with vagrant provision or by linking to the vagrant inventory file & using the ansible-playbook command directly:

ln -s .vagrant/provisioners/ansible/inventory/vagrant_ansible_inventory dev
ansible-playbook -i dev playbook.yml 

Now, you can vagrant ssh into the machine to start generating & serving tiles.

Included in this playbook is some OSM test data on Lichtenstein...we can use that as an example:

vagrant ssh
sudo su postgres
cd /srv/data
osm2pgsql --create --slim --flat-nodes nodes.bin -C 2000 --number-processes 1 --hstore liechtenstein-latest.osm.pbf
. post-import-steps.sh # add indexes and custom functions according to osm-bright.tm2source @ https://github.com/kartotherian/osm-bright.tm2source
exit

We use Tilerator for generating our vector tiles. For detailed usage, take a look @ their README.

Once you've imported some data, you can vagrant ssh into your machine & fire up tilerator (workers) and tilerator-ui (ui only) services:

sudo systemctl enable tilerator tilerator-ui
sudo systemctl start tilerator tilerator-ui

To talk to Tilerator from your host machine, you can forward the tilerator-ui port via SSH:

vagrant ssh -- -L 6533:localhost:16533

Now, on your host, you can post to localhost:6533 to add a job to Tilerator:

curl -X POST 'http://localhost:6533/add?generatorId=gen&storageId=v2&zoom=0&fromZoom=0&beforeZoom=15&biggerThan=0&parts=12'

Or, in a web browser, navigate to localhost:6533/jobs to take a look at job progress.

Once you've generated some tiles, you can serve them with Kartozoa. Simply enable & start the Kartozoa service:

sudo systemctl enable kartozoa
sudo systemctl start kartozoa

Tiles should now be available internally @ localhost:16532/osm-intl/{zoom}/{x}/{y}.pbf and externally @ https://maps.zoondka.com/kartozoa/osm-intl/{zoom}/{x}/{y}.pbf. Take a look @ Kartozoa for more info.

As with Kartozoa, simply enable & start the zoondka-maps service to host our mapping app:

sudo systemctl enable zoondka-maps
sudo systemctl start zoondka-maps

The app should now be available @ localhost:16531 or https://maps.zoondka.com.

If you'd like to use parts of this playbook as a starting point for another machine, feel free to fork the repository & enjoy. You will want to create your own production variables and then groom the Ansible roles as you desire. For example, if you'd like to set up a machine just for generating and serving vector tiles using Kartotherian, then this a great repository to start with.

In production, depending on the machine you are using & the dataset you'd like to import, you will probably want to tune your osm2pgsql command, for this take a look at osm2pgsql usage