As I mentioned in my previous post I’ve been playing with Docker lately trying to build an OpenDJ image that I could use for integration testing. Before digging deeper into my testing strategy I decided to publish this short post to explain how such image can be created using a simple Dockerfile.
You can go ahead and download the project from Github if you want to play with it. This is what the Dockerfile looks like:
FROM java:8 WORKDIR /opt EXPOSE 1389 1636 4444 ADD opendj.zip /var/tmp/ RUN unzip /var/tmp/opendj.zip -d /opt/ && rm -fr /var/tmp/opendj.zip ENV JAVA\_HOME /usr/lib/jvm/java-8-openjdk-amd64/ RUN /opt/opendj/setup --cli -p 1389 --ldapsPort 1636 --enableStartTLS --generateSelfSignedCertificate --baseDN "dc=example,dc=com" -h localhost --rootUserPassword password --acceptLicense --no-prompt --doNotStart ADD 99-custom-schema.ldif /opt/opendj/config/schema/ ADD custom-dit.ldif /var/tmp/ RUN /opt/opendj/bin/import-ldif --backendID userRoot --ldifFile /var/tmp/custom-dit.ldif && rm -fr /var/tmp/custom-dit.ldif CMD ["/opt/opendj/bin/start-ds", "--nodetach"]
Let’s look at this file line by line.
The base image is _java:8 _which is available from the public Docker Hub and, as the name implies, provides us with an image that has Java 8 already installed.
EXPOSE 1389 1636 4444
We need to expose the OpenDJ ports so that we can connect to them once the container is running:
- 1389: LDAP port
- 1636: LDAPS port
- 4444: Admin port
ADD opendj.zip /var/tmp/
By default the Dockerfile expect to find the OpenDJ binaries in a file called opendj.zip. You can download any OpenDJ version from the Fogerock Backstage page, rename it to opendj.zip and move it to the project folder.
If you prefer to use the nightly build you can comment out the previous line and uncomment the following one:
RUN echo `curl -s https://forgerock.org/djs/opendjrel.js | grep -o "http://.*\.zip" | tail -1` | xargs curl -o /var/tmp/opendj.zip
Once the OpenDJ zip file has been added we need to extract the contents to the installation directory:
RUN unzip /var/tmp/opendj.zip -d /opt/ && rm -fr /var/tmp/opendj.zip
And then run the installation script:
RUN /opt/opendj/setup --cli \ -p 1389 \ --ldapsPort 1636 \ --enableStartTLS \ --generateSelfSignedCertificate \ --baseDN "dc=example,dc=com" \ -h localhost \ --rootUserPassword password \ --acceptLicense \ --no-prompt \ --doNotStart
Besides specifying ports and hostname we also have to choose a base DN, password for the “cn=Directory Manager” user, and tell the script not to start OpenDJ just yet.
I decided to add the following lines at the end to be able to load a custom schema file and some sample data to initialize our directory:
ADD 99-custom-schema.ldif /opt/opendj/config/schema/ ADD custom-dit.ldif /var/tmp/ RUN /opt/opendj/bin/import-ldif \ --backendID userRoot \ --ldifFile /var/tmp/custom-dit.ldif \ && rm -fr /var/tmp/custom-dit.ldif
And finally, this line tells Docker which command to run when the container starts:
CMD ["/opt/opendj/bin/start-ds", "--nodetach"]
We need to start OpenDJ in no-detach mode to keep the container running afterwards.
And that’s it! You can build and spin up a container by running the following commands (feel free to use your own image name and port mappings):
docker build -t groman/opendj-test:1.0 . docker run -d -p 1389:1389 1636:1636 444:4444 groman/opendj-test:1.0
Be advised that any changes made after the container was started (via ldapmodify for example) will be lost once the container is stopped. This was put in place intentionally to make sure our integration tests always work with the same pre-define data.