Docker Swarm

*****************************

SETUP OF DOCKER SWARM:

******************************

Create 3 AWS instances and name them as Manager, worker 1 and worker 2.

Go to git bash

# cd downloads/

# SSH path from AWS for manager

# sudo su -

# yum install docker

# vim /etc/hostname ==> to chnage the hostname of this machine to Manager

:wq!

# init 6 ==> restart the machine

Again open git bash

# cd downloads/

# SSH path from AWS for Worker1

# sudo su -

# yum install docker

# vim /etc/hostname ==> to chnage the hostname of this machine to Worker1

:wq!

# init 6 ==> restart the machine

Again open git bash

# cd downloads/

# SSH path from AWS for Worker2

# sudo su -

# yum install docker

# vim /etc/hostname ==> to chnage the hostname of this machine to Worker2

:wq!

# init 6 ==> restart the machine

****************************************

Initialize the Swarm Manager:

***************************************

ON MANAGER:

*************

Go to the manager machine and give below command

# systemctl start docker

# docker swarm init --advertise-addr 172.31.17.248(privateip of manager)

Curent node will be now be swarm Manager and it will generate a token id that can be executed on worker nodes so that they will join the swarm as workers.

==> Copy the token id on notepad

************************

WORKER1 SETUP

**********************

# systemctl start docker

paste the token command here from notepad // as shown below:

# docker swarm join --token SWMTKN-1-66xevqpe5r6h1gtcps1m867q8jcu93deyk87qudqdp7nyf8omk-7vna75w6jip5hmqp4h55bi3s3 172.31.17.248:2377

// the worker1 will now join the swarm as worker

It will give message as: This node joined a swarm as a worker.

MANAGER NODE:

Go to manager node and see if worker1 has joined the swarm or not

# docker node ls

2 nodes will be there .. maanager and worker1

WORKER2 SETUP

**********************

# systemctl start docker

paste the token command here from notepad // as shown below:

# docker swarm join --token SWMTKN-1-66xevqpe5r6h1gtcps1m867q8jcu93deyk87qudqdp7nyf8omk-7vna75w6jip5hmqp4h55bi3s3 172.31.17.248:2377

// the worker2 will now join the swarm as worker

It will give message as: This node joined a swarm as a worker.

MANAGER NODE:

Go to manager node and see if worker1 has joined the swarm or not

# docker node ls

3 nodes will be there .. maanager and worker1 and worker2

**********SETUP COMPLETE********************************************

SCEANRIO 1 : LOAD BALANCING or distribution of Service

************************************************

Here we will create a service nginx with 5 replicas and perform port mapping

# docker service create --name mywb -p 8989:80 --replicas 5 nginx

==>This will create 5 replicas of nginx and it is distributed on the nodes.

==> to check the services

# docker service ps mywb

List all services on SWARM

*******************************

# docker service ls

*********************************

Delete service on the SWARm

# docker service rm mywb

//removes service

# docker service ls // no service will be there

*****************************************

Scaling

****************************

Scenario2:

Using the concept of scaling we can increase or reduce the replicas

1.Create tomcat service with replica 3 with posrt mapping

# docker service create --name mytomcat -p 9090:8080 --replicas 3 tomcat

2. # docker service ps mytomcat

3. Now scale up the service to 5

# docker service scale mytomcat=5

4. check if 5 replicas of tomcat are running

# docker service ps mytomcat

// total of 5 services will be there 2 more services added.

5. Now scale down the services to 3

# docker service scale mytomcat=3

6. check if 3 replicas of tomcat are running

# docker service ps mytomcat

// total of 3 services will be there.

******************************************

SCENARIO 3: FAILOVER

**************************************

A. Create a httpd service with 6 replicas and delete one replica running on the manager

1. Create httpd service with 6 replicas

# docker service create --name mywebserver -p 9090:80 --replicas 6 httpd

2. Check if all 6 replicas are running or not.

# docker service ps mywebserver

3. On the MANAGER machine, delete a replica

# docker ps -a // replicas will be displayed

# docker rm -f containername // replica will be deleted

# docker ps -a

# docker service ps mywebserver

3. Drain worker1 from the swarm

# docker node ls

# docker node update --availability drain Worker1

4. check if all 6 replicas are running on manager and worker2

# docker service ps mywebserver

# docker service ps mywebserver | grep Running

5. Make worker1 join the swarm again

# docker node update --availability active Worker1

6. check if all 6 replicas are running on manager and worker1 and 2

# docker service ps mywebserver

7. we can also go to worker2 git bash and leave the swarm

# systemctl start docker

# docker swarm leave

8.Connect back to manager

# docker node ls

worker 2 will be in status down and Active

9. Remove the worker2 node from the swarm

# docker node rm Worker2

Worker 2 will not be in swarm now.

# docker node ls

9.Check if all the replicas are still running or not.

# docker services ps webserver

10. join back the worker2 to swarm

# docker swarm join-token worker

copy the token and paste it on worker 2 machine

docker swarm join --token SWMTKN-1-195rkd1a6332n2izpccyfn9d29q1mqs72ja0d4ueo63xyetrum-6t7d8m9tfcw9qtj86ske56p0q 172.31.44.159:2377

Worker2 will join the swarm again

on Manager machine: # docker node ls // worker 2 will be active and running

**********************************************************

ROLLING UPDATE/ ROLLING OUT:

*************************************************************

Create redis:3 with 5 replicas and update it to redis:4 and latter roll it back to redis:3

1.Create redis:3 with 5 replicas

# docker service create --name myredis --replicas 5 redis:3

2.Check if 5 replicas are running or not

Docker service ps myredis

3.perform a rolling update from redis:3 to redis:4

# docker service update --image redis:4 myredis

4.Check if 5 replicas of redis:3 are shutdown and redis:4 are up and running

# docker service ps myredis

5.Perform roll back from redis:4 to redis:3

# docker service update --rollback myredis

6.Check if 5 replicas of redis:4 are shutdown and redis:3 are up and running

# docker service ps myredis

*****************************************************

OVERLAY NETWORK

****************************************************

Create 1 custom overlay networks overlay1

Usecase 1: Create a tomcat service in swarm on overlay1 network

1.Create 1 overlay network

# docker network create --driver overlay overlay1

2. Check if the networks are created or not

# docker network ls

3. Start tomcat service with 3 replicas on overlay1 network

# docker service create --name mytomcat -p 8888:80 replicas 3 --network overlay1 tomcat

4. Check if tomcat service is running on overlay1 network

# docker service inspect mytomcat

OR

# docker service inspect mywebserver --pretty

In network you will see netwrok to be overlay1