Architecture

Components

Manager

Scheduling

Resource Availability
Labels and constrains
- Engine label and constrains
- Node label and constrains
- Example
- Labels
  - env=test, storage=ssh
- Constrains
  - Nodes
Strategy and affinity
- In Swarm mode only strategy supported is spread

Service Discovery

Runs an internal DNS server, which register are services created with Docker Swarm mode.

Nodes

Runs the contaniner using docker-engine.

As shown above, with Docker Swarm mode the order of creation is following:-

Services
- tasks
  - containers

Defining an application

RSVP application

Docker Compose

version: '2'
services:
 mongodb:
   image: mongo:3.3
   expose:
     - "27017"
   environment:
    MONGODB_DATABASE: rsvpdata
   networks:
    - rsvpnet

 web:
   image: teamcloudyuga/rsvpapp
   ports:
    - "5000:5000"
   environment:
    MONGODB_HOST: mongodb
   networks:
    - rsvpnet

networks:
  rsvpnet:

Docker Application Bundle

{
  "Services": {
    "mongodb": {
      "Env": [
        "MONGODB_DATABASE=rsvpdata"
      ],
      "Image": "mongo@sha256:919d22a962b1913a2f57cbcd772689b4061cc05a7ccdf32a1900b58462c351d4",
      "Networks": [
        "rsvpnet"
      ],
      "Ports": [
        {
          "Port": 27017,
          "Protocol": "tcp"
        }
      ]
    },
    "web": {
      "Env": [
        "MONGODB_HOST=mongodb"
      ],
      "Image": "teamcloudyuga/rsvpapp@sha256:06923b50d69ee693ea4b3fc156f720d672a56508a3d66b0fe6f558715624e019",
      "Networks": [
        "rsvpnet"
      ],
      "Ports": [
        {
          "Port": 5000,
          "Protocol": "tcp"
        }
      ]
    }
  },
  "Version": "0.1"
}

High availablity of application

Replicas of tasks running
Multiple nodes

Service discovery and Load Balancing an application

iptables
- classifies, modify and take decisions based on the packet
IPVS
- load balancer at transport layer available in the Linux kernel

Autoscaling an application

No inbuild primitive yet.

Rolling upgrade and rollback of an application

$ docker service update --image teamcloudyuga/rsvpapp:1 --update-delay 10s rsvp

Internally connecting to other application

Each service get registred with the internal discovery service
Applications can contact other using service VIP

Networking option to connect applications with-in the cluster

Overlay network
Network Plugins

Accessing the application from external world

Once a service publishes a port, routing mesh maps the published port on all the nodes
Any external client can query the published port on any nodes to reach the desired service
A Load Balancer (HAProxy/Nginx) can be configured to map an endpoint to the pubished port on the nodes of the cluster.

Managing storage for application

bind mounts

$ docker service create \
--name my-service \
--mount type=bind,source=/path/on/host,destination=/path/in/container \
nginx:alpine

named volumes

$ docker service create \
--name my-service \
--replicas 3 \
--mount type=volume,source=my-volume,destination=/path/in/container,volume-label="color=red",volume-label="shape=round" \
nginx:alpine

Detailed docs are available here.

Demo

Get the

$ git clone https://github.com/cloudyuga/container-orchestration.git
$ cd docker-swarm
$ export DO_TOKEN=adfvvfvs............
$ vagrant up --provider=digital_ocean
$ vagrant ssh manager
$ git clone https://github.com/cloudyuga/rsvpapp.git
$ cd rsvpapp

On Manager

root@manager:~/rsvpapp# docker swarm init --advertise-addr=<MANAGER_IP>
Swarm initialized: current node (<MANAGER_NODE_ID>) is now a manager.

To add a worker to this swarm, run the following command:

    docker swarm join \
    --token SWMTKN-1-<TOKEN> \
    <MANAGER_IP>:2377

To add a manager to this swarm, run 'docker swarm join-token manager' and follow the instructions.

On Worker1

root@worker1:~# docker swarm join \
>     --token SWMTKN-1-<TOKEN> \
>     <MANAGER_IP>:2377
This node joined a swarm as a worker.

On Worker2

root@worker2:~# docker swarm join \
>     --token SWMTKN-1-<TOKEN> \
>     <MANAGER_IP>:2377
This node joined a swarm as a worker.