Ansible Vagrant setup for a local triplestore instance of SNAP:DRGN.
This does not handle live server provisioning!
Vagrant and VirtualBox (or some other VM provider) can be used to quickly build or rebuild virtual servers. This Vagrant profile installs Apache, and the 4Store triplestore using the Ansible provisioner.
To use the vagrant file, you will need to have done the following:
- Download and Install VirtualBox
- Download and Install Vagrant
- Install Ansible
- Checkout this repository
- Open a shell prompt (Terminal app on a Mac) and cd into the repository folder (which contains the
Vagrantfile) - Run the following command to install the necessary Ansible roles for this profile:
$ ansible-galaxy install --force -r requirements.yml - Once all of that is done, type
vagrant up, and Vagrant will create a new VM, install the base box, and configure it.
Once the new VM is up and running (after vagrant up is complete and you're back at the command prompt), you can log into it via SSH if you'd like by typing in vagrant ssh. Otherwise, the next steps are below.
The VM checks out the snap code and data from github into a local shared docroot directory, but does not import data into the 4store triplestore. To do this there is a separate playbook called import-rdf.yml. To import the RDF data into the 4store service run this playbook separately by typing:
ansible-playbook provisioning/import-rdf.yml -i provisioning/inventory
The import-rdf.yml playbook is written so you don't need to worry about manually performing these steps to import any data youself. You can simply run the ansible-playbook command above to import/re-import the data into your local development SNAP virtual machine. These notes are here for documentation purposes because it took ruddy ages to figure out the first time!
Documentation about the 4store triplestore is currently only available (as of 2018-11) via a mirror site at https://4store.danielknoell.de/trac/wiki/. 4store is installed via the provisioning playbook (which uses apt-get install 4store in the VM. The package manager install configures a 'fourstore' user in Ubunutu, as well as the init.d config files, etc. This means 4store can be started and stopped via the systemctl commands:
sudo systemctl start 4store
sudo systemctl stop 4store
The provisioning playbook sets the port to 9000 and creates and sets the default triplestore to 'snap'. The data directory for 'snap' is:
/var/lib/4store/snap
and this directory needs to be writable by the 'fourstore' user so importing data/reading metadata files work properly. This directory is internal to the client VM and not a shared folder with the host machine (like docroot is).
To import the SNAP RDF data the 4store httpd daemon cannot be running. You must stop all 4store processes using the systemctl stop command and run only the backend using sudo -H -u fourstore 4s-backend snap to ensure the daemon is running under the 'fourstore' user. If you run 4s-backend without specifying the user the process will be owned by, probably 'vagrant' or 'root' and you will have problems importing data because of the permissions dissonance between the filesystem and the daemon process.
With the VM running the SNAP:DRGN triplestore should be accessible via:
192.168.33.33:9000 - base port
192.168.33.33:9000/status - info
192.168.33.33:9000/query - web query
192.168.33.33:9000/sparql - endpoint
These provisioning playbooks are not currently configured to set the web app up. TODO for someone else. ;-)
Original Vagrant/Ansible playbooks created in 2014 by Jeff Geerling, author of Ansible for DevOps.
Modified for the 2018 SNAP November hackathon: https://github.com/Prosopographia