Set up the Mes Aides stack.
The ansible scripts in this repository have been tested only Debian 11 x86_64 server. However, older or newer versions of Debian may be compatible.
You will need at most the following ressources:
- an SSH connection as a priviledged user to the remote server
- Ansible >2.14.5 with Python >3.9 installed on your local machine. See the documentation for your operating system.
Then, duplicate the file vps.yaml
in the inventories
folder and modifiy it to match your needs.
The options are as follow :
virtualmachines:
hosts:
vps:
ansible_host: vps-45bb7a36.vps.ovh.net # The server adress
ansible_port: 22 # The ssh port used to connect to the server
ansible_ssh_user: debian # The user name of the priviledged account on the server
ansible_ssh_private_key_file: ~/.ssh/id_rsa # The path to the SSH key used to connect to the server
ansible_host_ip: 51.38.232.135 # The ip of the server, used to generate DNS records
fullname: solstice.aides-jeunes.leonides.org #
dns_root: leonides.org # The root of domain name use by your server
email: [email protected] # The email used to register Certbot
github_users: # The github users that will be able to connect to the server
- github_username_example
monitor: # If set, a monitoring service will be deployed on specified port
port: 8887
applications: # List all applications that will be deployed
- name: aides_jeunes
repository: https://github.com/betagouv/aides-jeunes.git
branch: main
default_site: true
https: true
domain: aides-jeunes.leonides.org
node_server_port: 8001
node_instance_number: 4
openfisca_server_port: 2001
openfisca_worker_number: 4
You can set some basic security settings on your server by running ansible-playbook -i ./inventories/vps.yaml initialize.yaml
.
This will disable SSH connection to the server using password. This step will not run if the server user file ~/.ssh/authorized_keys
is either empty or missing. The following properties will be modified in /etc/ssh/sshd_config
:
- set
PasswordAuthentication no
- set
ChallengeResponseAuthentication no
It is advised to run this command on a newly installed server, while keeping an active ssh connection in parallel and only if you understand the implication of those parameters. Your hosting service should provide you with an emergency access if you get locked out of the server.
In order for the server to be successfully deployed, some sub domain name must be specified in the DNS record. In order to know which values are required, simply run ansible-playbook -i ./inventories/vps.yaml dns-record.yaml
. You will get an output such as this one:
monitor.solstice.aides-jeunes 3600 IN A 5.135.137.147
solstice.aides-jeunes 3600 IN A 5.135.137.147
www.solstice.aides-jeunes 3600 IN A 5.135.137.147
openfisca.solstice.aides-jeunes 3600 IN A 5.135.137.147
aides-jeunes 3600 IN A 5.135.137.147
www.aides-jeunes 3600 IN A 5.135.137.147
openfisca.aides-jeunes 3600 IN A 5.135.137.147
Then you will have to add all those entries to the associated domain name DNS record.
In order to enable continuous deployment of this ops repository on your server you need to run the following command once:
ansible-playbook -i ./inventories/vps.yaml synchronize.yaml
A copy of this repository will be created in the folder /opt/mes-aides
of the server. This repository will be automatically updated and new modifications applied every time an ssh connection is made with the private key associated with the update_key
defined in the inventory.
Note: Private and public keys should be generated manually:
- Private should be keys added to Github secrets variable (used here for instance)
- Public keys added to
ops.update_key
variable in the inventory
Warning: Launching the synchronize.yaml
playbook alone will remove continuous deploiement of server stack. You will need to run the bootstrap.yaml
playbook again to re-enable it.
Run the command ansible-playbook -i ./inventories/vps.yaml bootstrap.yaml
in order to bootstrap the server basic configuration.
Once done, every applications should be up and running on the server.
Note that you only need to run this command once, but you can re-run it if you modify either Nginx, Python, Mongo configuration or if the bootstrap process failed at some point. All unaltered steps that ran successfully will be automatically skipped by Ansible.
In order to setup continuous deployment, you will need to:
- Run manually the
synchronize.yaml
playbook - Run manually the
bootstrap.yaml
playbook - Connect to the server using one of the private keys associated to your Github account
- switch user to
main
- run
cd ~/
and cd the application folder you want to deploy - get the private key (see
ansible_ssh_private_key_file
in inventory) - set it up in your Github repository as a secret (see here)
It is possible to dump mongodb collections from a server and restore them on another.
In order do dump data, you will need to had specific configuration lines in your inventory application to specify which mongodb collection to target and with which query :
mongodb_collections_migration:
simulations: '{"createdAt": {"$gte": { "$date": "2023-01-01T00:00:00.000Z" }}}'
followups: '{"createdAt": {"$gte": { "$date": "2023-01-01T00:00:00.000Z" }}}'
Then run the following command to download the selected collections locally in a ./.tmp
folder :
ansible-playbook -i ./inventories/localhost.yaml --tags="dump" mongodb-migration.yaml
To restore that data on another server you will need to add the following lines to the inventory applications :
mongodb_collections_migration:
simulations: ""
followups: ""
Then run the command :
ansible-playbook -i ./inventories/vps.yaml --tags="restore" mongodb-migration.yaml
In order to run ansible on a local image you will need to have both Vagrant and Docker installed on your machine. You will also need to have a valid public/secret key pair in your local ssh folder (~/.ssh/
) called id_rsa.pub
and id_rsa
.
Navigate to the local
folder and run the command :
vagrant up --provider=virtualbox
to create a VirtualBox VMvagrant up --provider=docker
to create a docker container (recommended if running on an arm64 processor)
Once the image is successfully created, you should be able to run any of the above commands.