Instructions for creating docker images

First make sure you have a docker daemon running on a reachable host that exposes the REST-interface over HTTP. On ubuntu this can be accomplished by editing the file /etc/default/docker and adding the following values to the variable DOCKER_OPTS:

DOCKER_OPTS="-H tcp://0.0.0.0:2375 -H unix:///var/run/docker.sock"

Then, restart the docker daemon using: service docker restart.

For a more secure connection, the same can be accomplished using an SSH-tunnel without exposing the docker REST-interface to the outside world (which is a bad thing).

The docker images are built when the docker-build profile is active. This profile is automatically activated when the docker.host variable is provided, for example:

mvn -Ddocker.host=http://localhost:2375 package

Images are built during the package phase.

Maven creates the following docker images:

• cds-base: base image for the other CDS images.
• cds-config: image containing CDS configuration (the configdir).
• cds-postgresql: image containing a PostgreSQL/PostGis installation containing the CDS database.
• cds-ldap: image containing an OpenDS installation containing several entries that are used by the CDS.
• cds-tomcat: base image for CDS tomcat instances.
• cds-admin: tomcat containing admin war.
• cds-webservices: tomcat containing wars for the webservices.
• cds-jobexecutor: contains and executes the job-executor jar. Queries the database for new jobs to execute.
• cds-apache: contains Apache vhost-configurations for the CDS.
• cds-cron: cronjob to update datasets.

Starting CDS in Docker

Create data volume containers:

docker run -d -v /var/lib/postgresql --name cds-master-dbdata cds-postgresql true
docker run -d -v /opt/OpenDS-2.2.1/db --name cds-master-ldapdata cds-ldap true
docker run -d -v /etc/cds/workspaces --name cds-master-workspaces cds-webservices true
docker run -d -v /var/lib/cds --name cds-master-metadata cds-admin true

Note: the data volume containers contain the data that is mutated by the CDS components. With the exception of the workspaces container this data cannot be reproduced by repeating the installation instructions. Keep this in mind when destroying these containers!

Copy the deegree workspaces into the cds-master-workspaces volume.

docker run --rm --volumes-from cds-master-workspaces -v /path/to/local/workspaces:/etc/cds/workspaces-src cds-base sh -c 'cp -r /etc/cds/workspaces-src/* /etc/cds/workspaces'

Create data volume container containing the CDS configdir:

docker run --name cds-master-config -e CDS_INSPIRE_HOST=http://vrn-test-services.idgis.nl cds-config

Create service containers:

TODO: setup e-mail
docker run --name cds-master-postgresql -P -d --volumes-from cds-master-dbdata --restart=always cds-postgresql
docker run --name cds-master-ldap -P -d --volumes-from cds-master-ldapdata --restart=always cds-ldap
docker run --name cds-master-admin -P -d --volumes-from cds-master-config \
	--volumes-from cds-master-metadata --link cds-master-postgresql:db \
	--link cds-master-ldap:ldap --restart=always cds-admin
docker run --name cds-master-jobexecutor -P -d --volumes-from cds-master-config \
	--volumes-from cds-master-metadata --link cds-master-postgresql:db \ 
	--link cds-master-ldap:ldap --restart=always cds-job-executor
docker run --name cds-master-webservices -P -d --volumes-from cds-master-config \
	--volumes-from cds-master-workspaces --link cds-master-postgresql:db \
	--volumes-from cds-master-metadata \
	--link cds-master-ldap:ldap --restart=always cds-webservices 
docker run --name cds-master-apache -p 80:80 -d --link cds-master-admin:admin \
	--link cds-master-webservices:webservices \
	--volumes-from cds-master-metadata \
	-e CDS_SERVER_NAME=vrn-test.idgis.nl \
	-e CDS_WEBSERVICES_SERVER_NAME=vrn-test-services.idgis.nl \
	-e [email protected] \
	--restart=always \
	cds-apache
docker run --name cds-master-cron -d --link cds-master-postgresql:db cds-cron

Configuration (using environment variables):

• Apache -> CDS_SERVER_NAME: hostname for the admin vhost.
• Apache -> CDS_WEBSERVICES_SERVER_NAME: hostname for the services vhost.
• Apache -> CDS_SERVER_ADMIN: e-mail address for the CDS server administrator.

Restoring database backups

Stop all services that use the database in order to ensure that all connections are closed:

docker stop cds-master-cron
docker stop cds-master-jobexecutor
docker stop cds-master-webservices
docker stop cds-master-admin

Import both the manager and the metadata schema by dropping it and recreating it from a backup:

docker exec -i cds-master-postgresql psql -d cds -c "drop schema manager cascade;"
docker exec -i cds-master-postgresql pg_restore --dbname=cds --format=custom --single-transaction < manager.backup
docker exec -i cds-master-postgresql psql -d cds -c "drop schema metadata cascade;"
docker exec -i cds-master-postgresql pg_restore --dbname=cds --format=custom --single-transaction < metadata.backup

Start the stopped services again:

docker start cds-master-admin
docker start cds-master-webservices
docker start cds-master-jobexecutor
docker start cds-master-cron

Force import of all jobs

The cds-master-cron container schedules new import jobs each night. These jobs, however, first check the metadata and only import datasets when the modification date has changed. To force an update of all datasets regardless of the modification date run the following command:

docker exec -i cds-master-postgresql psql -d cds -c "select manager.schedule_jobs_force();"

This is useful after importing a backup since the last modification date is part of the backup, while the imported data is not.