OVERVIEW

This Ansible project was created by Data Curation Experts based on work for the Chemical Heritage Foundation, UC Santa Barbara, and Washington University in St. Louis. It builds a production-style Hydra application on CentOS 7 - on Amazon EC2, a Vagrant virtual machine, or a bare-bones server.

Setup

Before you can use this project, you must:

1. Clone this repo to your local workstation using: git clone https://github.com/acozine/sufia-centos.git * NOTE: this script runs on your local system and connects to the server you want to configure, do not try to run it on the server directly*
2. Install Ansible on your local system.
3. Continue with the Amazon EC2, Vagrant, or SSH section depending on your server type

EC2

To create an Amazon EC2 instance:

1. Create new vars/main.yml files in the launch_ec2 and ec2 roles
2. Add your organization's AWS credentials there
3. Create a new vars/main.yml file in the services role
4. Override any default variables you wish to change there (we recommend overriding the postgresql database, user, and password settings)
5. Run ansible-playbook --private-key /path/to/your/keypair.pem ec2.yml (if you encrypt your variables with ansible-vault, add --ask-vault-pass; if you are not using passwordless sudo, add --ask-sudo-pass or -K)
6. If the playbook fails, you can restart it at a particular task with ansible-playbook --private-key /path/to/your/keypair.pem ec2.yml --start-at-task='rolename | taskname'

Vagrant

To use this project with Vagrant:

1. Create a Vagrant project
2. Modify the Vagrantfile to use a Centos 7 box (config.vm.box = )
3. Modify the Vagrantfile to use Ansible (see sample Vagrantfile for ideas)
4. Be sure to point to the vagrant.yml file, which skips the launch_ec2 and ec2 roles
5. Clone this project as the provisioning sub-directory of your Vagrant project
6. Run vagrant up
7. If the playbook fails, you can restart it at a particular tasks by uncommenting the line in your Vagrantfile that begins ansible.start_at_task and changing the value to the task name

Generic server running SSH

To run the Ansible provisioning scripts against a minimal Centos 7 server via ssh:

1. Create an installation user with passwordless sudo access on your server. We called ours centos. If you use a different name, edit the remote_user entry in vanilla.yml
2. Add the public key for your install user on the server in the users's .ssh/authorized_keys file. For help setting up public key authentication, see the Centos Wiki
3. these scripts assume that you have a separate virtual drive available for the application installation at /dev/sdc (assuming /dev/sda is your main system drive and /dev/sdb is your swap drive).
4. Edit the hosts file in this directory and replace the ip address for your Hydra application with your server's IP
5. Edit the vars: section of the vanilla.yml file and enter data for your system - see the commented lines
6. Run ansible-playbook -i hosts vanilla.yml from the root directory of this repo on your local machine
7. if the playbook fails, you can restart it at a particular task with ansible-playbook -i hosts vanilla.yml --start-at-task='rolename | taskname'

Next Steps

This project deploys your code with Capistrano. If your project doesn't already use Capistrano. Look to the Capistrano getting started documentation.

You'll need to update your Hydra application (your application repository) to define a new stage corresponding to the environment you've just set up.
We usually give our stages names that help identify the servers they run on: e.g. demo, sandbox, stage, prod, etc. Copy one of the existing stage files in your project (under config/deploy) to a new stage file.

In your stage file, set the Capistrano :deploy_to directory to match the Ansible housekeeping role's project_name variable. In config/deploy.rb and/or in config/deploy/<yourenv>.rb (depending on your project structure), add a line like:

set :deploy_to, '/opt/sufia-project'

If you use the default value for project_name in the housekeeping role, this is the exact line you need.

1. In the main project repo for your rails application, create a new cap stage by copying one of the existing sage files in the config/deploy directory
2. Edit your new stage file a. Edit the line that says set :stage, and add your stage name. It's easiest to use the same name for the file and the stage. b. Edit (or add) the line that says set :deploy_to, '/opt/...' and set ... to the project_name you set in your Ansible YAML file varialbes
3. Run cap {stage-name} deploy where {stage-name} is the file and stage name you set in steps 1 & 2.
4. After you run Capistrano the first time (and only the first time), restart Tomcat using sudo systemctl restart tomcat
5. Your application should be up and running! Check it in your browser.