aws-swarm-init

Docker image responsible for automatic initialization (setup) of docker swarm cluster. Most of the logic lives in the script entry.sh.

In order to setup a swarm cluster,

swarm init must be run on one of the managers
swarm join on remaining nodes (both managers and workers) with appropriate tokens

To perform the same steps automatically, dynamodb's locking feature would be used. Due to locks, only one manager is capable of querying/inserting data into dynamodb and a manager which is successfully queries/inserts into dynamodb would perform swarm init and other utilize the data in the dynamodb for joining the cluster.

This docker image needs following environment variables

NODE_TYPE: Should be either manager or worker
DYNAMODB_TABLE: Name of the dynamodb table to be used for locking and passing cluster information
REGION: AWS region in which dynamodb table was created (if not provided will default to region of the instance)

docker-compose:

version: "2"
services:
    init-aws:
      image: "srikalyan/aws-swarm-init:<version>"
      container_name: "aws-swarm-init"
      restart: "no"
      environment:
        NODE_TYPE: "<manager|worker>"
        DYNAMODB_TABLE: "<dynamodb_table>"
      volumes:
        - /var/run/docker.sock:/var/run/docker.sock
        -  /usr/bin/docker:/usr/bin/docker
        - /var/log:/var/log

Note 1: entry.sh is mostly taken from the docker for aws with few modifications

Note 2: This image needs a docker client but does not install one as this would create unnecessary versions which we could easily skip by mount the docker binary from host ubuntu machine.

Note 3: If you need to delete all the manager nodes then clear out your dynamodb table before destroying your manager nodes.