Sometimes Kubernetes is the answer, but other times not. redeployster recreates containers in scrappy, non-enterprise docker-compose setups. It's only purpose is to run docker-compose up -d $SERVICE and to expose that functionality through HTTP so that your build pipelines can trigger it remotely.

Attach a redeployster.token label to the services you want to make deployable:

# docker-compose.yml
version: "3"
services:
  hello:
    image: hello-world
    labels:
      - 'redeployster.token=dolphin'

Make sure that the service got launched manually once so the container(s) have the label:

docker-compose up -d

The deployable service is now configured:

• The name of the docker-compose service will be a new exposed path on redeployster: call POST /hello here to deploy the hello service.
• The token will be required by reployster to trigger the deploy: add Authorization: Bearer dolphin http header to the call.
• Adding a token also acts as an opt-in flag for a service to be deployable. Without a token, redeployster will ignore the service.

go run .

# In another shell:
curl -i -XPOST -H'Authorization: Bearer dolphin' http://localhost:4711/hello

Redeployster replies with HTTP Status 200 as soon as a deploy job starts because it then streams the deploy output. To access the exit code of a job, you can read the Exit-Code HTTP trailer.

Here is an example with curl that saves the headers in a separate file and then uses grep to exit with a non-zero code if the deployment fails:

curl -XPOST -H'Authorization: Bearer dolphin' http://localhost:4711/hello -D headers.txt
grep -q '^Exit-Code: 0\b' headers.txt

Redeployster currently doesn't have options to provide a certificate in order to listen directly on port 443.

It is meant to run on the same host as the services it needs to deploy. These Docker-managed services are most likely exposed to the HTTPS port via a reverse proxy like Nginx, Traefik, Caddy etc.

One typical scenario is to add a forwarder service under the reverse-proxy. The forwarder will then proxy requests to redeployster running directly on the host. See this example using Caddy as the forwarder, and assuming Traefik as the main reverse-proxy.

# docker-compose.yml
services:
  forwarder:
    image: caddy:2.6.4-alpine
    command: 'caddy reverse-proxy --from :3000 --to host.docker.internal:4711'
    extra_hosts:
      - "host.docker.internal:host-gateway"
    labels:
      - "traefik.enable=true"
      - "traefik.http.routers.forwarder.rule=Host(`deploy.example.com`)"
      - "traefik.http.services.forwarder.loadbalancer.server.port=3000"
      - "traefik.http.routers.forwarder.tls.certresolver=default"

Note: If you have a firewall, you might need to allow Docker's network interface to access the Redeployster's port (4711)

We could also run it within a Docker container and let it handle the other containers by mounting the Docker socket. But redeployster actually needs to call docker-compose, and for this it needs access to the docker-compose.yml file from the host.

Requirements: go

Useful commands:

• Build the binary: go build .
• Build & run from source: go run .
• Format the code: go fmt .