Running the Cassandra cluster example

Presentation

The accompanying presentation from OpenSlava 2014 can be found from http://www.slideshare.net/IlkkaAnttonen/open-slava-2014ilkkaanttonen.

Checking out the project and contents

You can check out the project from GitHub with

git clone https://github.com/SirIle/openslava-docker.git <dirname>

This README document assumes that dirname directory is the root.

Under it are four relevant directories for setting up and running the container demo:

• dockerfiles (contains the scripts and container descriptors)
• maestro (MaestroNG orchestration files)
• vagrant (contains the Vagrant configuration files)
• weave (contains the shell scripts for running containers and weave networking)

Prerequisites

The environment has been tested in Windows 7 and Mac OS X. It should also run in Linux.

Following software has to be installed on the host (tested version numbers in parenthesis):

• VirtualBox (4.3.20)
• Vagrant (1.7.1)

On OS X and Linux you also need

• Ansible (1.8.2)

The Vagrant scripts also assume that you have a directory called ~/Projects created. It can be empty. On Windows, it should be c:\users<username>\Projects and in other operating systems under your home directory.

Running in one node with MaestroNG orchestration

Setting up the environment

Start and enter node1

cd vagrant/ubuntu_docker
vagrant up
vagrant ssh

You can also connect to the running node1 with other terminals. It has the host only network IP of 10.10.10.30. You need to use the Vagrant ssh key for the connection. Instructions for doing this for PuTTY can be found from https://blog.accenture.com/ilkka_anttonen/2014/04/15/connecting-to-vagrant-vm-on-windows-using-putty/.

This launches the node1 virtual machine and provisions it using Ansible. You can use tmux for multiple sessions.

Building the images

Build the container images

cd ~/dockerfiles
./buildContainers.sh

Running the containers

Use MaestroNG to start the containers

cd ~/maestro
maestro -f cassandra.yaml start

Running in two nodes with Weave and registry

Setting up the environment

Start node1

cd vagrant/ubuntu_docker
vagrant up

Start node2

vagrant up node2

Start docker registry

cd ../ubuntu_docker_registry
vagrant up

Building and pushing the images

On node1 build the images and push core and cassandra to the local registry

cd dockerimages
./buildContainers.sh
./pushContainers.sh

Running containers in node1

On node1 use the shell script to start containers

cd ~/weave
./startNode1.sh

Running containers in node2

On node2 use the shell script to start containers. This will pull them from the local registry first.

cd ~/weave
./startNode2.sh

Connecting the nodes

On either of the nodes, connect to the core instance and join the Consul master nodes. For example on node2

ssh root@localhost -p 1000 'consul join -wan 10.0.1.1'

Restart Cassandra instance on that node so that it reconnects to the other datacenter

ssh root@localhost -p 1021 'supervisorctl restart cassandra'

After a while the nodes have joined to the same ring. This can be checked with

ssh root@localhost -p 1021 'nodetool status'

If you want to run cqlsh on one of the Cassandra containers, it can be done with

ssh root@localhost -p 1021 -t cqlsh

Cleanup

To stop the running containers and clear out the instances use

~/weave/stopNode.sh

Checking that everything is working

Different services can be reached with a browser from the host. Services can be found from:

Service	URL
Logstash on node1	http://10.10.10.30
Logstash on node2	http://10.10.10.40
ConsulUI	http://10.10.10.30:8500/ui/dist
Opscenter	http://10.10.10.30:8888

Inside the Docker node the different containers can be accessed using ssh (username/password is root/root):

Container	Ssh port	Weave IP
Core	1000	10.0.1.1
Cassandra1	1021	10.0.1.2
Cassandra2	1022	10.0.1.3
Cassandra3	1023	10.0.1.4
Opscenter	1121	10.0.1.10
Node	1001	10.0.1.20
Core (node2)	1000	10.0.2.1
Cassadra4 (node2)	1021	10.0.2.2
Cassandra5 (node2)	1022	10.0.2.3

The command to connect to for example Cassandra1 on node1 is:

ssh root@localhost -p 1021

Creating a keyspace and few test rows

Ssh into one of the Cassandra containers and start up cqlsh

ssh root@localhost -p 1021 -t cqlsh

Enter the following (you can exit cqlsh with either exit or ctrl+d)

CREATE KEYSPACE demo WITH REPLICATION = { 'class' : 'SimpleStrategy', 'replication_factor' : 2 };

CREATE TABLE demo.users (
	userid text,
	firstname text,
	lastname text,
	PRIMARY KEY (userid)
);

INSERT INTO demo.users (userid, firstname, lastname) VALUES ('ile','Ilkka','Anttonen');
INSERT INTO demo.users (userid, firstname, lastname) VALUES ('toinen','Some','User');

SELECT * FROM demo.users where userid = 'ile';

Status of the Cassandra ring can be displayed with nodetool status.

Using icedcoffeescript to query the cluster

Connect to the node container

ssh root@localhost -p 1001

Install the node-cassandra-cql module with npm

npm install node-cassandra-cql

Create the following program, use your favourite editor and name the file cql.iced

cql = require 'node-cassandra-cql'
client = new cql.Client({hosts:['cassandra.service.consul'],keyspace: 'demo'})
client.execute 'SELECT * FROM demo.users where userid = ?', ['ile'], (err, result) ->
  console.log 'Result: ' + result.rows[0].firstname
  process.exit 0

Now this program can be run

iced cql.iced

You should get the result if you have previously created the test data as described in previous step.

Cleanup scripts

To stop the containers and clean out unused images and remove instances use

/vagrant/scripts/cleanDocker.sh