Skip to content

Latest commit

 

History

History
320 lines (216 loc) · 9.62 KB

deployment-guide.md

File metadata and controls

320 lines (216 loc) · 9.62 KB

Deployment guide for production

  1. Getting started
  2. Development and installation
  3. Deployment for production
  4. Authentication and magic tokens
  5. Websockets for interactive communication

Contents

There are easier ways to run a blog. You will find the resource requirements for this stack quite substantive. This really is if you intend to run some complex web service and need all the architecture.

NOTE: this is a more focused, step-by-step version of the generated "README". You will find a few more details there, especially about customising larger deployments, or changing your configuration after deployment. This is designed to get you up-and-running with your first production deployment in a controlled way. No guarantees, though.

Preparation

Committing to GitHub

Prepare your code and resources for your first commit. There are three files which must not be committed unless you're quite positive your data will never leak.

  • /.env
  • /cookiecutter-config-file.yml
  • /frontend/.env

These files will also need to be customised for production deployment. Make alternative arrangements for these files. Don't trust .gitignore to save you.

DigitalOcean Droplets

This guide uses DigitalOcean Droplets, so customise as required. Deploy to the smallest (currently 500MiB memory, 1 vCPU and 10GiB SSD for $4/month). You can upgrade later when you know your resource requirements.

Ensure you add your SSH encryption keys on launch so that your server can be secure from the beginning.

Deploy on whatever server image your prefer, although the default would be Ubuntu 20.04 (22.04 is the latest). End-of-life for 20.04 is April 2030, and for 22.04 is April 2032. You have time. The underlying image isn't that critical, as you'll be using the Docker images at their current versions.

Domain name and email

Get your settings and redirects at your registrar, and then set up the various DNS records at DigitalOcean, pointing at the IP address for the droplet you set up.

For reference:

Don't forget to create DNS A records for flower, traefik.

Now you should be able to login to your server and begin deployment.

Deployment

Docker

Update your server, and install all required packages:

# Install the latest updates
apt-get update
apt-get upgrade -y

Then:

# Download Docker 
curl -fsSL get.docker.com -o get-docker.sh
# Install Docker using the stable channel (instead of the default "edge") 
CHANNEL=stable sh get-docker.sh
# Remove Docker install script 
rm get-docker.sh

Clone your repository

The basic approach is to clone from GitHub then set up the appropriate .env files and any custom conf files called from docker-compose. If yours is a private repo, review the GitHub docs for how to set that up.

Remember you can create new passwords as follows:

openssl rand -hex 32
# Outputs something like: 99d3b1f01aa639e4a76f4fc281fc834747a543720ba4c8a8648ba755aef9be7f

From /srv:

git clone https://github.com/<user-name>/<project-name>.git

Then continue from the project directory /srv/<project-name>. You can always pull your latest code from that directory, with:

git pull

Docker Swarm Mode

Deploy the stack to a Docker Swarm mode cluster with a main Traefik proxy, set up using the ideas from DockerSwarm.rocks. And you can use CI (continuous integration) systems to do it automatically.

This stack expects the public Traefik network to be named traefik-public.

export USE_HOSTNAME=example.com
# Set up the server hostname 
echo $USE_HOSTNAME > /etc/hostname 
hostname -F /etc/hostname

Set up Swarm Mode:

docker swarm init

If this fails, you'll need to explicitly link the public IP for the droplet:

docker swarm init --advertise-addr 123.123.123.123

You can add additional manager and worker nodes. This is optional and you can read the DockerSwarm.rocks link for more.

Check that the nodes are connected and running:

docker node ls

Which would output something like:

ID                            HOSTNAME             STATUS    AVAILABILITY    MANAGER STATUS    ENGINE VERSION
ndcg2iavasdfrm6q2qwere2rr *   dog.example.com      Ready     Active          Leader            18.06.1-ce

Traefik Proxy with HTTPS

Follow the documentation from DockerSwarm.rocks to get automatic HTTPS certificates.

Create a network that will be shared with Traefik and the containers that should be accessible from the outside, with

docker network create --driver=overlay traefik-public

Get the Swarm node ID of this node and store it in an environment variable (use the code below exactly):

export NODE_ID=$(docker info -f '{{.Swarm.NodeID}}')

Create a tag in this node, so that Traefik is always deployed to the same node and uses the same volume:

docker node update --label-add traefik-public.traefik-public-certificates=true $NODE_ID

Create an environment variable with your email, to be used for the generation of Let's Encrypt certificates, e.g.:

export EMAIL=someone@example.com

Create an environment variable with the domain you want to use for the Traefik UI (user interface), e.g.:

export DOMAIN=traefik.example.com

You will access the Traefik dashboard at this domain, e.g. traefik.example.com.

Create an environment variable with a username (you will use it for the HTTP Basic Auth for Traefik and Consul UIs), for example:

export USERNAME=admin

Create an environment variable with the password, e.g.:

export PASSWORD=changethis

Use openssl to generate the "hashed" version of the password and store it in an environment variable:

export HASHED_PASSWORD=$(openssl passwd -apr1 $PASSWORD)

Download the file traefik.yml:

curl -L dockerswarm.rocks/traefik.yml -o traefik.yml

Deploy the stack with:

docker stack deploy -c traefik.yml traefik

It will use the environment variables you created above. Check if the stack was deployed with:

docker stack ps traefik

It will output something like:

ID             NAME                IMAGE          NODE              DESIRED STATE   CURRENT STATE          ERROR   PORTS
w5o6fmmln8ni   traefik_traefik.1   traefik:v2.2   dog.example.com   Running         Running 1 minute ago

You can check the Traefik logs with:

docker service logs traefik_traefik

Deploy to a Docker Swarm mode cluster

There are 4 (5) steps:

  1. Pull your git repo
  2. Build your app images
  3. Deploy your stack
  4. Restart your docker service

Here are the steps in detail:

  1. Pull your git repo
cd /srv/example
sudo git pull
  1. Build your app images
  • Set these environment variables, right before the next command:
    • TAG=prod
    • FRONTEND_ENV=production
  • Use the provided scripts/build.sh file with those environment variables:
TAG=prod DOMAIN=example.com STACK_NAME=example-com TRAEFIK_TAG=example.com FRONTEND_ENV=production bash -x scripts/build.sh

Persisting Docker named volumes

You can use docker-auto-labels to automatically read the placement constraint labels in your Docker stack (Docker Compose file) and assign them to a random Docker node in your Swarm mode cluster if those labels don't exist yet.

To do that, you can install docker-auto-labels:

pip install docker-auto-labels

And then run it passing your docker-stack.yml file as a parameter:

docker-auto-labels docker-stack.yml

You can run that command every time you deploy, right before deploying, as it doesn't modify anything if the required labels already exist.

  1. Deploy your stack
  • Set these environment variables:
    • DOMAIN=example.com
    • TRAEFIK_TAG=example.com
    • STACK_NAME=example-com
    • TAG=prod
  • Use the provided scripts/deploy.sh file with those environment variables:
DOMAIN=example.com TRAEFIK_TAG=example.com STACK_NAME=example-com TAG=prod bash -x scripts/deploy.sh
  1. Restart your docker service
sudo service docker restart

You may need to prune regularly while developing if you find yourself running out of space:

docker system prune

Production URLs

These are the URLs served in production (replace example.com with your own):