I have created my first real karaf-based docker image: https://github.com/steinarb/sonar-collector#run-the-sonar-metrics-collector-in-docker
The sonar-collector is intended as a webhook called by sonar on completing a build, it will store key sonar measurements as a row in a database table. Sonar has these numbers internally but doesn't have a way to get at them. At first we typed these numbers manually into a spreadsheet to get statistics on a set of analyzed programs. But that was tedious, so we created sonar-collector to capture the numbers instead. The docker image is built with maven against a local docker and then after testing, the image was pushed to docker hub: https://github.com/steinarb/sonar-collector#building-the-docker-image The docker image is a very thin layer on top of the official karaf 4.2.8 docker image https://github.com/steinarb/sonar-collector/blob/master/docker/docker/Dockerfile#L1 It copies in two files into the image's karaf etc directory: 1. org.apache.karaf.features.cfg a. Adds a feature repository for sonar-collector https://github.com/steinarb/sonar-collector/blob/master/docker/docker/org.apache.karaf.features.cfg#L28 b. Adds two features to the boot features i. An enviroment variable overloadable DataSourceFactory feature that defaults to postgresql https://github.com/steinarb/sonar-collector/blob/master/docker/docker/org.apache.karaf.features.cfg#L52 ii. The sonar-collector-webhook-with-postgresql feature https://github.com/steinarb/sonar-collector/blob/master/docker/docker/org.apache.karaf.features.cfg#L53 2. org.ops4j.datasource-sonar-collector.cfg which is a modified version of the file generated by pax-jdbc on first start, and the JDBC driver name and JDBC connection info overridable by system environment variables https://github.com/steinarb/sonar-collector/blob/master/docker/docker/org.ops4j.datasource-sonar-collector.cfg#L2 This image will provision itself from maven central (sonar-collector is released to maven-central) and will always use the latest released version, so there is no need to build a new docker image if a new version of sonar-collector is released to maven central. Since the database schema is set up by liquibase and the SQL used is trivial (inserts rows into a single table), it should in theory be possible to use a different database than PostgreSQL, and I have provided the possibility to override the database used when doing "docker run" of the image. But using other databases than PostgreSQL is something I haven't extensively tested.