Ansible Playbook install for the gEAR repository (DHART) and LAMP stack

---

python3 -m venv ~/.venv/ansible
source ~/.venv/ansible/bin/activate
pip install --upgrade pip
pip install ansible

ansible --version
# ansible [core 2.16.2]
#   python version = 3.11.2
#   jinja version = 3.1.3
#   libyaml = True

The inventory file defines the nodes in which the stack is configured. Check Ansible (quick setup).

Basic setup instructions were taken from the gEAR setup page. This playbook is partly adpated from the examples and best practices for building Ansible Playbooks.

Quickstart

The stack is deployed using the following command:

ansible-playbook site.yml --tags [tag] -v 2>&1 | tee output.log

Each role can be deployed in turn with the --tags option, e.g.

• setup-install: Install required libraries and python packages
• setup-config: Configure setup
• web-install: Install Apache and PHP modules
• web-config: Configure web server
• db-install: Install MySQL
• db-config: Configure MySQL, create DB user, etc.

To perform a dry-run, use

ansible-playbook site.yml --tags [tag] --check -v 2>&1 | tee output.log

Note: This playbook installs the DHART branch, and whatever latest commits were made, for instance values in www/site_domain_prefs.json are used as is. Default DB-related variables (group_vars/all) must match the settings in gear.ini, see also create_schema.sql.

Note: When running CGI scripts through the web server, PATH does not see the virtual environment. Solutions such as How to use Python virtual environments with mod_wsgi do not seem to work for CGI scripts. So far, the solution is to specify the full path to the interpreter (#!/usr/local/envs/dhart/bin/python), so that the shell can find the local installation when it attempts to execute CGI programs. Hence default variables (group_vars/all) such as python_venv_name must match the shebang in all CGI scripts. I think this also holds for the gEAR/www/p script. For practical purposes, this has also been done for the gEAR/bin scripts (in particular to run db-config). If installing the DHART branch, then there is nothing to do.

Note: Running db-config will create the DB, download the annotations, and load them to the respective DB tables. This is currently done using scripts under gEAR/bin (another way would be to use MySQL dumps, but we would need an appropriate loader, i.e. we don't want to install the whole Ensembl data, or redefine tables). Some default variables are specified in group_vars/all, such as release number, etc., but URLs are hard coded in dump_db.yaml. The list of organisms (db/configure/vars/main.yaml) is fixed, and must match the one defined in the DB (create_schema.sql)! Currently, re-running dump_db.yaml is NOT harmless, this will (re-)load annotations each time to the DB (or results in an error for loading gene ontologies).

Install

For recent changes to the installation, see CHANGELOG. For installation notes, see INSTALL

Ansible (quick setup)

Defaults are set in config file = /etc/ansible/ansible.cfg. A minimal configuration file looks like:

[defaults]
inventory = /etc/ansible/hosts/hosts.ini, /home/eboileau/.ansible_local_inventory.ini
remote_user = eboileau

[ssh_connection]
pipelining = True

[privilege_escalation]
become = True

where /etc/ansible/hosts/hosts.ini is e.g.

gear ansible_host=gear_IP
dhart ansible_host=dhart_IP

and /home/eboileau/.ansible_local_inventory.ini is

[all:vars]
ansible_become_password=your_remote_password

To list available hosts, use ansible all --list-hosts, or list all ansible-inventory --list. For more information on how to use Ansible, consult the Ansible documentation.