GIS Docker

Purpose for this repository is to act as a playground for testing different GIS related web technologies. Therefore, it cannot be considered as fully working application that can be automatically deployed into a server. However, this repository contains some useful examples that can be utilized in other projecs.

This repository contains containerized (Docker) parts of a GIS application e.g. different map layer protocols, routing, geocoding. Following can be considered as vital part of any GIS application. Additionally, this repository contains some utility containers (osm2pgrouting & digiroad2pgrouting) that are used for importing network data (osm & digiroad) into database for pgrouting.

Docker containers

• Client 8082:8082
• Server 8085:8085
• Database 5435:5432
• Geoserver 8080:8080
• Mapproxy 8083:8083
• Nginx 80:80
• Nominatim 7070:8100

Utility containers

• osm2pgrouting (run on demand)
• digiroad2pgrouting (run on demand)

Overview of repository elements

• Nginx -> Proxy for proxying traffic for client & server containers
• Client -> Main application for different GIS services (map layers, routing, geocoding)
• Server -> Currently only used for authentication
• Database -> Contains pgrouting and geocoding (nominatim) related data (osm, digiroad). Additionally OIDC related tables
• Geoserver -> Serves map layers in different protocols (WMS, WMTS, WFS, WFS-T, VectorTile, WPS)
• Mapproxy -> Serves background map for client (MML Taustakartta)
• Nominatim -> Geocoder for client application

Proxying (Nginx & Server)

Nginx is used for proxying network traffic between containers.

/ -> client:8082
/api -> server:8085/api
/proxy -> server:8085/proxy
/oidc -> server:8085/oidc
/interaction -> server:8085/interaction

Server has another proxy that handles integration between other containers that are being called from client. Purpose is to check client session on each request to prevent unauthorized requests. Additionally, it adds basic authentication header for requests. These credentials are stored on env variables that are given to each container. Exception is Geoserver where credentials are manually set for "gis_infra"-user

/proxy/geoserver -> http://geoserver:8080/geoserver (basic auth header added for request)
/proxy/mapproxy -> http://mapproxy:8083/mapproxy (basic auth header added for request)
/proxy/nominatim -> http://nominatim:7070 (basic auth header added for request)

Database

Database container consists of two databases: gis & nominatim.

gis-database contains three schemas: public, routing_digiroad & routing_osm. Helsinki.osm.pbf & Uusimaa Digiroad subset data "gis"-database consumes around 1.35 gigabytes.

• public -> contains OIDC related tables
• routing_digiroad -> contains converted digiroad data for pgrouting
• routing_osm -> contains converted osm data for pgrouting

nominatim-database contains public schema that contains converted OSM data for nominatim geocoder. Helsinki.osm.pbf data consumes around 900 megabytes.

Database and table sizes can be checked with following sql queries.

SELECT pg_size_pretty(pg_database_size('gis'));
SELECT pg_size_pretty(pg_database_size('nominatim'));
SELECT pg_size_pretty(pg_total_relation_size('routing_osm.ways'));

Installing and running containers

1. Install docker
2. Build and run Docker containers with:

$ docker-compose -f docker-compose.dev.yaml up --build

In order to pgrouting & nominatim to work properly we need to import some OSM & Digiroad data for the project.

NOTE: Currently OSM filename is set on nominatim & pgrouting docker-compose files.

1. Load OSM by cities or countries and other regions
2. Load Digiroad data from here
3. Place loaded data under root data folder. This folder that premade folders for each data type.
4. Import OSM & Digiroad data into database for pgrouting by running pgrouting-database-setup.sh script.
5. Import OSM data into database for nominatim by running nomatim-setup.sh script.

Additionally client provides by default to use Kapsi hosted Taustakartta as background map. In order to get background map working, we need to seed these tiles by using Mapproxy. Seeding can be started by running mapproxy-seed.sh script. This will take some time (approx. 1-2 hours). Leave it running on background.

After starting containers the main application should be available from http://localhost:80 address. Test user is [email protected] -> testi.

Useful docker commands

Build individual container

$ docker-compose -f <compose-file> build <container-name>

Run individual container

$ docker-compose -f <compose-file> up <container-name>

Start containers without build

$ docker-compose -f docker-compose.dev.yaml up

Start containers with build

$ docker-compose -f docker-compose.dev.yaml up --build

Shutdown containers

$ docker-compose -f docker-compose.dev.yaml down

Installing and running client & server applications (for development purposes)

In order to run client & server applications node and yarn needs to be installed (or yarn can be seen as optional ;)).

Server

Install server dependencies

$ cd server
$ yarn install

Run sequelize migrations and seeds (db container should be running by now)

$ npx sequelize-cli db:migrate
$ npx sequelize-cli db:seed:all

Start server with nodemon

$ yarn run start-dev

Client

Install client dependencies

$ cd client
$ yarn install

Start client

$ yarn run serve