deployment

A repository with scripts, docker-compose, etc to guide deployment.

How does it work

This deployment repo contains a single docker-compose file deployed with docker swarm. The secrets are injected using:

printf "strava-client-secret-here" | docker secret create strava_client_secret -

There are currently 8 secrets defined:

• strava_client_id: The strava client id.
• strava_client_secret: The associated secret.
• db_connection_string: The connections string to the postgres db.
• db_host: The host of the database (endpoint)
• db_port: The port on which the database is listening
• db_user: User that has access to the database
• db_password: The password for the database
• db_name: The name of the used database

Continuous deployment is setup using a simple cronjob:

*/1 * * * * cd /var/deployment && git pull && docker stack deploy -c docker-compose.yml api --with-registry-auth

Architecture

The bikedata project consists of various servers:

The frontend server is hosted at https://www.bikedataproject.org/ - a relatively lightweight server. Another server, the data-server (aka the bikedataprojectdata-server) does the heavy lifting (generating the tiles, keeping track of all the data, ...) The webserver acts as a proxy for the data server, the webserver's NGINX forwards the necessary requests.