ml-docker

This repo contains a set of Docker containers for building and running MarkLogic server.

Images

There are three images

coreos is the Fedora 23 base system
runner is a minimal container for running MarkLogic Server
builder is configured to allow you to build the server

coreos

The coreos container installs Fedora 23 and sets up Java. It creates one non-root user that you can use when docker execing into the container.

There are three build arguments:

java a path to a Java RPM to install, you must supply this
user the name of the non-root user, defaults to devuser
uid the UID of the non-root user, defaults to 1000

By setting the user and uid arguments appropriately, you can allow the user logged into the container to update volumes mounted on it.

runner

The runner container installs and runs MarkLogic server. You must supply a MarkLogic RPM to the marklogic build argument. You can download them from http://developer.marklogic.com/

This container also installs the MarkLogic Python API.

There are two build arguments:

marklogic a path to a MarkLogic RPM to install, you must supply this
user the name of the non-root user, defaults to devuser and must be the same as the user passed to the coreos image

builder

The builder container extends the coreos container with build tools sufficient to compilee MarkLogic Server. (Assuming, of course, that you have access to the sources.)

There’s a single build argument:

user the name of the non-root user, defaults to devuser and must be the same as the user passed to the coreos image

Ancillary scripts

Optional: two scripts exists to 'pull' these files during the build. They are meant to be augmented and are only called if apprprate rpms are not found. The getjava.sh script will pull a java 1.8_60 JDK from oracle. The getml.sh script will simply remind you to obtain a MarkLogic rpm

Building the images

A build script is provided in bin.

Usage:

./build.sh [coreos|runner|builder] {--tag tag} {--user devuser} {--uid 1000}

This builds one of 3 variants. You must build coreos first. If you specify a tag for the coreos image, then you’ll have to modify the Dockerfiles for the other images.

sebastiankalinowski / ml-docker