Using NAV with Docker for development

Docker is a lightweight “virtualization” framework for creating isolated environments, useful both in development and production. For more information on Docker visit their homepage or read the documentation.

Installing Docker and docker-compose

Docker provides up-to-date documentation on how to install it for most popular operating systems [*]. NAV’s Docker definitions should work smoothly on Linux, but may have some rough edges on Docker Desktop for Mac.


To avoid having to use sudo with docker commands, it is recommended to add your user to the docker group. You may need to relogin for it to take effect.

Getting started with Docker Compose

After installing Docker, you will need to obtain the NAV source code.

The source code contains a docker-compose.yml configuration file for Docker Compose. This configuration defines a fully integrated NAV runtime environment, with all its dependencies. This environment is designed to run NAV directly from the checked out source code, and as such it defines an environment for developers, not for production use of NAV. The alternative is to manage all the dependencies and integrations on your own host machine.

The quickest way to build the container images and start all the services for the first time is by running these commands:

make .env
docker compose up


The first time you run this would be the perfect time to grab some coffee (and maybe redecorate your living room), as the initial build may take a while.


The container images for NAV development are designed to bind-mount your source code directory inside the running containers. In order to avoid leaving files owned by strange user IDs in your source code directory, the images will create a non-privileged nav user with a specific user-id and group-id. These IDs should match that of your user account on the host system, so therefore the Docker Compose build process needs to know your UID and GID.


This UID/GID mapping is not really relevant if you are running Docker Desktop on a Mac, since it uses an entirely different mechanism for bind-mounted volumes. You still will need to set the UID and GID arguments for the build to work, though.

The quickest way to go about this is the make .env command. This will attempt to generate a .env file in your top-level source code directory, which will set the UID and GID variables from your running environment. Docker Compose will implicitly read the environment variables in this file when it builds or runs the services defined in docker-compose.yml. If, for some reason, the make .env command does not work for you, you can create the .env file by hand (but supply real values if you’re on Linux):

.env example

Using the container(s)

The Docker Compose specificiation creates several containers (called “services” in Docker Compose lingo). Several of them will mount the checked out source code directory internally on the /source directory, allowing them to always be up-to-date with the latest changes you are making in your favorite editor.

These are the defined services:


This container runs the NAV backend processes and cron jobs. It also runs the sass-watcher job, which will watch *.scss files for modifications and recompile NAV’s CSS when changes do occur.


This container runs the Django development server to serve NAV’s web-based user interface. By default, Docker Compose will expose this web service on port 80 on the host system, i.e. at http://localhost/


This runs a bog standard Postgres image from the Docker Hub, to serve as NAV’s main data store.


This runs both carbon-cache backend and a graphite-web frontend, for NAV’s storage and retrieval of time-series data. By default, Docker Compose will expose the web service on port 8000 on the host system, i.e. http://localhost:8000/


This container will watch the doc/ directory for changes and initiate a rebuild of the NAV documentation whenever the documentation source files are modified. The built documentation should normally be browseable via the web service at http://localhost/doc/

Accessing internals of running containers

If need be, you can access the internals of the running containers (to control NAV daemons using the nav command, adjust the running config, or whatever) by running a bash shell inside the container, like so (for the nav container):

docker compose exec nav /bin/bash

Manually restarting the web server

To manually restart the web server, all you need is:

docker compose restart web

Rebuilding the NAV code from scratch

A complete rebuild of the NAV code can be initiated by:

docker compose restart nav

Rebuilding the containers

Running :kdb:`docker compose up` will normally build the container images, before starting them, if they don’t exist already. However, if the image definitions have changed (e.g. when you are switching between development branches or changed the Dockerfile definitions, or any of the files used as part of the image definitions), you may need to rebuild the images. To initiate a full build (which will still utilize Docker’s build cache), run this:

docker compose build

Another valid method is to use the --build option when starting the containers. This will ensure the images are always rebuilt if necessary as part of the startup process:

docker compose up --build

Sometimes, you may find that a rebuild isn’t enough to clear out all the cruft after switching development branches or adding or changing NAV’s default configuration file examples. The Docker Compose environment defines two persistent volumes that will retain their data between restarts and rebuilds: nav_cache and nav_config. The former exist just to share some caching data between the various service containers. The second ensures the set of NAV config files remain persistent between restarts or rebuilds, and also that all service containers can share the same set of files. When you really want to start from scratch, you can fully nuke the Docker Compose environment and the persistent volumes using this command (before initiating a new up or build command):

docker compose down --volumes

Controlling processes inside the nav container

The main nav container uses supervisord to control multiple processes. While the nav command can be used to control individual NAV services, supervisorctl can be used to control other processes used within the development environment:


This is the regular system cron daemon, responsible for running recurring NAV tasks.


This is a one-time supervisor task to start all of NAV when the container starts.


This is a process that monitors the python/nav/web/sass/ subdirectory for changes, and re-runs python build_sass (i.e. rebuilding all the SASS-based stylesheets) on changes.

The individual logs of these program are typically found inside the nav container in the /var/log/supervisor/ directory. The NAV process logs themselves are placed inside the /tmp/ directory inside the nav container.

Controlling log levels and configuration

The log levels of various parts of NAV are controlled through the config file /etc/nav/logging.conf inside the containers.

The nav and web containers share a common configuration volume named nav_config. This volume should persist even between rebuilds of the containers themselves. If you want NAV to install a completely new set of config files from scratch, you may need to manually trash this volume using the -v option to the docker-compose down command.

Overriding the compose services

If you need to override certain aspects of the Docker Compose service definitions for your own purposes during development, you can usually do so without patching the docker-compose.yml file. You can “patch” the definitions via Docker Compose’s override mechanism: Simply add a docker-compose.override.yml to the top-level source directory.

Preventing NAV backend services from starting at container startup

You can add the environment variable NONAVSTART=1 to prevent the backend daemons from being started at the nav container startup time (allowing for complete manual control of daemons, by entering the container using exec, as documented above). This can be done by adding something akin to this:

version: '2'
      - NONAVSTART=1

The same technique can be used to insert your own environment into the web container.

Happy hacking!