Docker Compose setup for CKAN

• Overview
• Installing Docker
• docker compose vs docker-compose
• Install CKAN plus dependencies
• Development mode
  • Create an extension
• CKAN images
  • Extending the base images
  • Applying patches
• Debugging with pdb
• Datastore and Datapusher
• NGINX
• The ckanext-envvars extension

Overview

This is a set of configuration and setup files to run a CKAN site.

The CKAN images used are from the official CKAN ckan-docker repo

The non-CKAN images are as follows:

• DataPusher: CKAN's pre-configured DataPusher image.
• PostgreSQL: Official PostgreSQL image. Database files are stored in a named volume.
• Solr: CKAN's pre-configured Solr image. Index data is stored in a named volume.
• Redis: standard Redis image
• NGINX: latest stable nginx image that includes SSL and Non-SSL endpoints

The site is configured using environment variables that you can set in the .env file.

Installing Docker

Install Docker by following the following instructions: Install Docker Engine on Ubuntu

To verify a successful Docker installation, run docker run hello-world and docker version. These commands should output versions for client and server.

docker compose vs docker-compose

All Docker Compose commands in this README will use the V2 version of Compose ie: docker compose. The older version (V1) used the docker-compose command. Please see Docker Compose for more information.

Install CKAN plus dependencies

Copy the included .env.example and rename it to .env. Modify it depending on your own needs.

Using the default values on the .env.example file will get you a working CKAN instance. There is a sysadmin user created by default with the values defined in CKAN_SYSADMIN_NAME and CKAN_SYSADMIN_PASSWORD(ckan_admin and test1234 by default). This should be obviously changed before running this setup as a public CKAN instance.

To build the images:

docker compose build

To start the containers:

docker compose up

This will start up the containers in the current window. By default the containers will log direct to this window with each container using a different colour. You could also use the -d "detach mode" option ie: docker compose up -d if you wished to use the current window for something else.

At the end of the container start sequence there should be 6 containers running

After this step, CKAN should be running at CKAN_SITE_URL.

Development mode

To develop local extensions use the docker-compose.dev.yml file:

To build the images:

docker compose -f docker-compose.dev.yml build

To start the containers:

docker compose -f docker-compose.dev.yml up

See CKAN Images for more details of what happens when using development mode.

Create an extension

You can use the ckan extension instructions to create a CKAN extension, only executing the command inside the CKAN container and setting the mounted src/ folder as output:

docker compose -f docker-compose.dev.yml exec ckan-dev /bin/bash -c "ckan generate extension --output-dir /srv/app/src_extensions"

The new extension files and directories will be created in the src/ folder. You might need to change the owner of its folder to have the appropiate permissions.

CKAN images

The Docker image config files used to build your CKAN project are located in the ckan/ folder. There are two Docker files:

• Dockerfile: this is based on ckan/ckan-base:<version>, a base image located in the DockerHub repository, that has CKAN installed along with all its dependencies, properly configured and running on uWSGI (production setup)

• Dockerfile.dev: this is based on ckan/ckan-base:<version>-dev also located located in the DockerHub repository, and extends ckan/ckan-base:<version> to include:

  • Any extension cloned on the src folder will be installed in the CKAN container when booting up Docker Compose (docker compose up). This includes installing any requirements listed in a requirements.txt (or pip-requirements.txt) file and running python setup.py develop.
  • CKAN is started running this: /usr/bin/ckan -c /srv/app/ckan.ini run -H 0.0.0.0.
  • Make sure to add the local plugins to the CKAN__PLUGINS env var in the .env file.

Extending the base images

You can modify the docker files to build your own customized image tailored to your project, installing any extensions and extra requirements needed. For example here is where you would update to use a different CKAN base image ie: ckan/ckan-base:<new version>

To perform extra initialization steps you can add scripts to your custom images and copy them to the /docker-entrypoint.d folder (The folder should be created for you when you build the image). Any *.sh and *.py file in that folder will be executed just after the main initialization script (prerun.py) is executed and just before the web server and supervisor processes are started.

For instance, consider the following custom image:

ckan
├── docker-entrypoint.d
│   └── setup_validation.sh
├── Dockerfile
└── Dockerfile.dev

We want to install an extension like ckanext-validation that needs to create database tables on startup time. We create a setup_validation.sh script in a docker-entrypoint.d folder with the necessary commands:

#!/bin/bash

# Create DB tables if not there
ckan -c /srv/app/ckan.ini validation init-db 

And then in our Dockerfile.dev file we install the extension and copy the initialization scripts:

FROM ckan/ckan-base:2.9.7-dev

RUN pip install -e git+https://github.com/frictionlessdata/ckanext-validation.git#egg=ckanext-validation && \
    pip install -r https://raw.githubusercontent.com/frictionlessdata/ckanext-validation/master/requirements.txt

COPY docker-entrypoint.d/* /docker-entrypoint.d/

NB: There are a number of extension examples commented out in the Dockerfile.dev file

Applying patches

When building your project specific CKAN images (the ones defined in the ckan/ folder), you can apply patches to CKAN core or any of the built extensions. To do so create a folder inside ckan/patches with the name of the package to patch (ie ckan or ckanext-??). Inside you can place patch files that will be applied when building the images. The patches will be applied in alphabetical order, so you can prefix them sequentially if necessary.

For instance, check the following example image folder:

ckan
├── patches
│   ├── ckan
│   │   ├── 01_datasets_per_page.patch
│   │   ├── 02_groups_per_page.patch
│   │   ├── 03_or_filters.patch
│   └── ckanext-harvest
│       └── 01_resubmit_objects.patch
├── setup
├── Dockerfile
└── Dockerfile.dev

pdb

Add these lines to the ckan-dev service in the docker-compose.dev.yml file

Debug with pdb (example) - Interact with docker attach $(docker container ls -qf name=ckan)

command: python -m pdb /usr/lib/ckan/venv/bin/ckan --config /srv/app/ckan.ini run --host 0.0.0.0 --passthrough-errors

Datastore and datapusher

The Datastore database and user is created as part of the entrypoint scripts for the db container. There is also a Datapusher container running the latest version of Datapusher.

NGINX

The base Docker Compose configuration uses an NGINX image as the front-end (ie: reverse proxy). It includes HTTPS running on port number 8443 and an HTTP port (81). A "self-signed" SSL certificate is generated beforehand and the server certificate and key files are included. The NGINX server_name directive and the CN field in the SSL certificate have been both set to 'localhost'. This should obviously not be used for production.

Creating the SSL cert and key files as follows: openssl req -new -newkey rsa:4096 -days 365 -nodes -x509 -subj "/C=DE/ST=Berlin/L=Berlin/O=None/CN=localhost" -keyout ckan-local.key -out ckan-local.crt The ckan-local.* files will then need to be moved into the nginx/setup/ directory

envvars

The ckanext-envvars extension is used in the CKAN Docker base repo to build the base images. This extension checks for environmental variables conforming to an expected format and updates the corresponding CKAN config settings with its value.

For the extension to correctly identify which env var keys map to the format used for the config object, env var keys should be formatted in the following way:

All uppercase Replace periods ('.') with two underscores ('__') Keys must begin with 'CKAN' or 'CKANEXT'

For example:

• CKAN__PLUGINS="envvars image_view text_view recline_view datastore datapusher"
• CKAN__DATAPUSHER__CALLBACK_URL_BASE=http://ckan:5000

These parameters can be added to the .env file

For more information please see ckanext-envvars