Docker Stack

This is used for creating a muiltcontainer architecture using docker compose and deploy it in the swarm cluster.

Docker compose + swarm = docker stack.



Commands

1.To see list of stack

# docker stack ls

2.To create a stck

# docker stack deploy -c stack_filename/docker_composefilename stackname

3.To see list of nodes where the stack service is running

# docker stack ps stackname

4.To see list of services in docker stack

# docker stack service stack_name

5.To delete a stck

# docker stack rm stack_name


Example:

Create a directory==>goinside the directory

Create a yml file ==> vim stack1.yml

---
version: 3

services:
 mydb:
  image: mysql:5
  environment: 
   MYSQL_ROOT_PASSWORD: edureka

 myworsdpress:
  image: wordpress
  ports:
   -5050:80
  deploy:
   replicas: 3
...

To deploy the stack service

# docker stack deploy -c stack1.yml wordpress

Here -c : allows you to create docker compose file with any name

3.To see where the stack replicas are running

# docker stack ps wordpress

3 replicas of wordpress and 1 mysql replica will be created

4.To remove the stack

# docker stack rm wordpress

Example2:

Starting a jenkins linked with 2 tomcat servers, creating a stack file and adding details like no: of replicas and which service to run on which node

# vim stack2.yml
---
version: 3

services:
 jenkins:
  image: jenkins
  ports:
   -5050:8080
  deploy:
   replicas: 2
   placement:
    constraints:
     -node.hostname == Manager
 
 qaserver:
  image: tomcat
  ports:
   -6060:8080
  deploy:
   replicas: 2
   placement:
    constraints:
     -node.hostname == Worker1
 
 prodserver:
  image: tomcat
  ports:
   -7070:8080
  deploy:
   replicas: 3
   placement:
    constraints:
     -node.hostname == Worker2
 
…

:wq!

# docker stack deploy -c stack2.yml stack-ci-cd

# docker stack ps stackci-cd

So all the replicas will be created will be running on thesre respective manager, worker 1 and worker 2 node as mentioned in stack file

Here under constraints we have give 1 node, but if you want we can give more than 1 node also, so if one node is not available then the service will start running on another node.
If in conatraint you have given only 1 node, then if the node is out of resources or is down, then the services will wait until that node comes alive again.

Example2: Advance:
Create a docker stack file to setup the selenium testing environment and also put an upper limit on the h/w allocation

---
version: '3'
services:
 hub:
  image: selenium/hub
  ports:
   -4444:4444
  deploy:
   replicas: 1
   resources:
    limits:
     cpus: "0.1"
     memory: "200M"
 chrome:
  image: selenium/node-chrome-debug
  ports: 
   -5901:5900
  deploy:
   replicas: 2
   resources:
    limits:
     cpus: "0.01"
     memory: "100M"   

firefox:
 image: selenium/node-firefox-debug
 ports: 
  -5901:5900
 deploy:
   replicas: 1

… 

2.To deploy the services from above stack file:

# docker stack deploy -c stack3.yml selenium

3.View the services

# docker stack ps selenium