This is a work in progress document.
The EHRI REST backend is deployed as a Neo4j unmanaged extension. In practice, this means:
- it compiles to an “uberjar” (all dependencies in one jar) which live
pluginsdirectory of a Neo4j installation
- the following setting in
neo4j.confmaps our Jersey web URIs to a path (/ehri):
In practice, a set of symlinks are used to allow easier versioning of releases:
$NEO4J_HOME/plugins/ehri-rest.jar points to
/opt/webapps/ehri-rest/current, which is itself a symlink to
When new versions of the EHRI code is released the uberjar is uploaded
/opt/webapps/ehri-rest/deploys directory, named with the
timestamp and the code’s git revision. Then the
current symlink is
updated to point to the new deployment and the Neo4j service restarted.
Using the Fabfile for automated deployment tasks¶
fabfile.py script provides a set of tasks that can be used to
release new versions of the code. Once Fabric (version 1.8) has been
installed (typically via
sudo pip install fabric) you can view the
available tasks like so:
$> fab --list Fabric deployment script for EHRI rest backend. Available commands: clean_deploy Build a clean version and deploy. copy_db Copy a (not running) DB from the remote server. current_version Show the current date/revision current_version_log Output git log between HEAD and the current deployed version. deploy Deploy the latest version of the site to the servers, install any latest Point symlink at latest version. online_backup Do an online backup to a particular directory on the server. online_clone_db Copy a Neo4j DB from a server using the backup tool. prod Use the remote virtual server. reindex_all Run a full reindex of Neo4j -> Solr data reindex_repository Reindex items held by a repository. restart Restart neo4j-service. rollback Rollback to the last versioned dir and restart. stage Use the remote staging server. start Start neo4j-service. stop Stop neo4j-service. test Use the remote testing server. update_db Update a Neo4j DB on a server.
More detailed info for tasks are available with the
$> fab -d online_clone_db Displaying detailed information for task 'online_clone_db': Copy a Neo4j DB from a server using the backup tool. This creates a copy of the running DB in /tmp, zips it, downloads the zip, extracts it to the specified DB, and cleans up. clone_db:/local/path/to/graph.db Arguments: local_dir
Three environments are available (which assume you have the hostnames defined in your .ssh/config file):
- test (sets hostname to
- stage (sets hostname to
- prod (sets hostname to
You can activate these environments by running them as commands prior to another task, i.e.
$> fab stage deploy # runs the deploy task with the 'ehristage' server