HowTo

Use Powershell

Create resource group

New-AzureRmResourceGroup -Location -Name

Deploy infrastructure

New-AzureRmResourceGroupDeployment -Name -ResourceGroupName -TemplateFile deploy_infra.json

This will setup following infrastructure

• 1 Public IP with an Azure Load Balancer
• 3 Master node VMs that can be sshed to via 2200, 2201, 2202 respectively and with :80 load balanced across 8080 on the 3 nodes
  • Docker Engine (Swarm Manager)
  • Consul server (Cluster of 3, already bootstrapped) with DNS integration to host and containers
  • Consul registrator
  • Kong Api Gateway
• 1 Worker Node VM scale set with initially 3 Workers
  • Docker Engine (Swarm Worker)
  • Consul agent
  • Consul registrator
• 1 MySql server VM
  • mysql with GMT zone, utf8 listening on 3306

Docker Swarm services

Deploy infrastructure services

publish service ports to host directly using --publish mode=host services will run on the docker host's bridge network and expose ports via port mapping to the host Example: kong running on 10.0.0.4 node (swarmMaster0), sees interfaces eth0 and localhost eth0 has 172.17.0.2 and gateway 172.17.0.1 using DNS server 168.63.129.16 (which is the only entry in its resolv.conf)

postgre (required by kong for now)

docker service create --publish mode=host,target=5432,published=5432 --env SERVICE_NAME=postgres --env POSTGRES_USER=kong --env POSTGRES_DB=kong --env POSTGRES_PASSWORD=kong --env SERVICE_5432_CHECK_TCP=true --env SERVICE_5432_CHECK_INTERVAL=15s --env SERVICE_5432_CHECK_TIMEOUT=3s --constraint "node.role == worker" postgres:9.4

redis

docker service create --mount type=volume,source=redis-data,target=/data --publish mode=host,target=6379,published=6379 --env SERVICE_NAME=redis --env REDIS_AUTH=notSecureP455w0rd --env SERVICE_6379_CHECK_TCP=true --env SERVICE_6379_CHECK_INTERVAL=15s --env SERVICE_6379_CHECK_TIMEOUT=3s --constraint "node.role == worker" ocbesbn/redis:latest

idpro

on mysql0, execute create database idpro; against mysql

docker service create --publish mode=host,target=3005,published=3005 --env MYSQL_HOST=$(curl swarmMaster0:8500/v1/catalog/service/mysql | grep -o -e "\"Address\":\"[^,]*" | grep -o "[^\"]*" | grep -v "Address"| grep -v ":") --env MYSQL_PORT=3306 --env SERVICE_NAME=idpro --env APP_HOST=52.233.155.169 --env APP_HOST_PROTOCOL=http --env APP_HOST_PORT=3005 --env GATEWAY_SCHEME=http --env GATEWAY_IP=52.233.155.169 --env GATEWAY_PORT=80 --env GATEWAY_CALLBACK=auth/callback --env CONSUL_HOST=172.17.0.1 --env SERVICE_3005_CHECK_HTTP=/register --env SERVICE_3005_CHECK_INTERVAL=15s --env SERVICE_3005_CHECK_TIMEOUT=3s ocbesbn/idpro:latest

kong

docker service create --mode global --publish mode=host,target=8080,published=8080 --publish mode=host,target=8443,published=8443 --publish mode=host,target=8001,published=8001 --env DNS_RESOLVER=172.17.0.1 --env SERVICE_NAME=kong --env KONG_DATABASE=postgres --env REDIS_AUTH=notSecureP455w0rd --env KONG_PG_HOST=postgres.service.consul --env KONG_PG_PORT=5432  --env APP_HOST=idpro.service.consul --env APP_HOST_PROTOCOL=http --env APP_HOST_PORT=3005 --env GATEWAY_SCHEME=http --env GATEWAY_IP=52.233.155.169 --env GATEWAY_PORT=80 --env GATEWAY_CALLBACK=auth/callback --env POSTGRES_USER=kong --env POSTGRES_PASSWORD=kong --env POSTGRES_DB=kong --env LOG_PORT=5000 --env LOG_HOST=172.17.0.1 --env REDIS_HOST=redis.service.consul --env REDIS_PORT=6379 --env SERVICE_8001_CHECK_HTTP=/ --env SERVICE_8001_CHECK_INTERVAL=15s --env SERVICE_8001_CHECK_TIMEOUT=3s --env SERVICE_8001_NAME=kong-api --env SERVICE_8080_NAME=kong --env SERVICE_8443_NAME=kong-https --constraint "node.role == manager" ocbesbn/api-gw:latest

api-registrator

docker service create --env SERVICE_NAME=api-registrator --env CONSUL_HOST=consul --env GATEWAY_CALLBACK=auth/callback --env KONG_HOST=kong-api --env KONG_PORT=8001 --env API_REGISTRY_PORT=3004 --publish mode=host,target=3004,published=3004 --constraint "node.role == worker" ocbesbn/api-registrator:latest

Deploy application services

Similar to infrastructure, only out of scope of this document

Configure Automated Deployments

Assuming that the service has initially been started with docker service create on target env. The depployment can update the existing service and will take into account any update parallelism and deployment settings configured on the service since creation. The deployment script will just figure out the service ID based on docker image repository name, then call docker service update --force specifying new image to load. This defaults to a rolling restart of the service and will wait for the deployment to complete before the build goes green.

1. Add project to circle CI
2. Add private key to circleci project SSH permissions
3. edit circle.yml and add a deployment (assuming it has been built to docker image already) Example

  development:
    branch: develop
    commands:
      - docker login -u $DOCKER_USER -p $DOCKER_PASS -e $DOCKER_EMAIL
      - docker tag ocbesbn/api-registrator:latest ocbesbn/api-registrator:dev
      - docker push ocbesbn/api-registrator:dev
      - curl https://raw.githubusercontent.com/gr4per/azureswarm/master/deploy_service.sh > deploy_service.sh
      - chmod +x deploy_service.sh
      - ./deploy_service.sh dm 52.233.155.169 ocbesbn/api-registrator dev