Docker Image for Apache Jena Fuseki

This repository hosts a Docker recipe for distributing Apache Jena Fuseki 2 server with SPARQL endpoint and web interface.

The Docker image is currently available from the Docker Hub as:

dasch-swiss/apache-jena-fuseki

Note that although these Docker images are based on the official Apache Jena Fuseki releases and do not alter them in any way, they do not constitute official releases from the Apache Software Foundation.

As of version 5.5.0-1, the database comes with a pre-configured TDB2 dataset called dsp-repo and a Lucene text index. For more details, see the dsp-repo.ttl dataset configuration file.

Dockerfile overview

The Dockerfile uses the official openjdk:11-jre-slim-buster base image, which is based on the debian:buster-slim image; this clocks in at about 69 MB.

The ENV variable FUSEKIVERSION determines which version of Fuseki is downloaded. Updating the version also requires updating the FUSEKISHA512 variable, which values should match the official Jena download .tar.gz.sha512 hashes.

The ASFMIRROR uses http://www.apache.org/dyn/mirrors/mirrors.cgi that redirect to a local mirror, with a fallback to the ASFARCHIVE http://archive.apache.org/dist/ for older versions. Note that due to subsequent sha512 checking these accessed with http rather than https.

To minimize layer size, there's a single RUN with curl, sha512sum, tar zxf and mv - thus the temporary files during download and extraction are not part of the final image.

Some files from the Apache Jena distributions are stripped, e.g., fuseki.war file.

The Fuseki image includes some helper scripts to do tdb loading using fuseki-server.jar. In addition Fuseki has a docker-entrypoint.sh that populates shiro.ini with the password provided as -e ADMIN_PASSWORD to Docker, or with a new randomly generated password that is printed the first time. The dsp-repo.ttl dataset configuration is used to set up the default dsp-repo dataset through the $FUSEKI_BASE/configuration/ directory mechanism.
There is also a REBUILD_INDEX_OF_DATASET environment variable that makes the container rebuild Fuseki's built-in Lucene index before starting the database normally. To enable this, the value of this variable must be set to the name of the dataset whose index should be rebuilt. Alternatively the REBUILD_INDEX_MARKER_FILE variable can be set to the path of a marker file inside the container containing the aforementioned dataset name. The marker file is deleted after a successful rebuild. The index data is assumed to be located at $INDEX_BASE/ (default: /fuseki/lucene/).

Releasing

Releasing should be done through the Github release process, which will kick-off a Github-CI job that will release the Docker Image under the release tag.