Muninn Coxswain

Deployment management for Muninn using Ansible.

Setup

Execution environment

Ansible commands are run within a container called an execution environment. To build the execution environment image:

ansible-builder build --tag test_ee

You may replace the image name test_ee with a more descriptive string.

Secrets

Secrets are expected to be stored in a directory called secrets/. The following are expected to be present:

• gcp service account file
• ssh keys for the user to be used by ansible
• a yaml file setting mirroring muninn's .env settings.

The names of these files are hardcoded for now, as this project only has one user for now. (me.)

I have provided a script to convert .env files to yaml files for use here.

python3 env_to_yaml.py foo.env > foo.env.yml

Usage

Tips

To restrict the set of hosts on which a playbook is run, change the value of target_hosts.

• target_hosts=muninn-h5n1-19-nov will only run against this one instance
• target-hosts=muninn-h5n1* will target all instance with names matching the pattern

The playbook uptime.yml just prints the uptime of all running instances in GCP. This is a way to test that your environment is set up correctly.

Create a new instance in GCP

Instance names start with the prefix muninn-. Provide a suffix value to name the instance.

podman run --rm -v $PWD:/muninn test_ee:latest \
ansible-playbook /muninn/playbooks/create_instance.yml \
-i /muninn/inventory.gcp.yml \
-u james_coxswain \
--private-key /muninn/secrets/james_coxswain \
-e "suffix=hello-world"

To change boot disk size provide -e "boot_disk_size_gb=32. Default is 32.

Update muninn on existing instances

This will update instances to the latest version of master and reload the nginx config. The database will not be wiped, and its schema will not be updated.

podman run --rm -v $PWD:/muninn test_ee:latest \
ansible-playbook /muninn/playbooks/update_instances.yml \
-i /muninn/inventory.gcp.yml \
-u james_coxswain \
--private-key /muninn/secrets/james_coxswain \
-e "target_hosts=all"

This playbook needs access to a yaml file that gives the muninn .env values. By default, this file is secrets/coxswain.env.yml. You can override the values in this file by supplying another file as extra_vars. Add the following to the end of the command:

-e '@/muninn/secrets/override.env.yml'

Update the H5N1 frontend on existing instances

This updates js-outbreak and outbreak-info-h5n1 to latest versions.

podman run --rm -v $PWD:/muninn test_ee:latest \
ansible-playbook /muninn/playbooks/update_frontend.yml \
-i /muninn/inventory.gcp.yml \
-u james_coxswain \
--private-key /muninn/secrets/james_coxswain \
-e "target_hosts=all"

Stop running containers and wipe database

podman run --rm -v $PWD:/muninn test_ee:latest \
ansible-playbook /muninn/playbooks/hard_shutdown.yml \
-i /muninn/inventory.gcp.yml \
-u james_coxswain \
--private-key /muninn/secrets/james_coxswain \
-e "target_hosts=all"

What doesn't this do?

Data ingestion: No part of the muninn data ingestion is handled by these playbooks. For now, ingestion is a very manual process.

Adding instances to load balancers: New instances need to be added to an instance group to associate them with a LB and make them accessible at a domain name. This is done manually through the google cloud console.