Create a Simple Cassandra Cluster With 3 Nodes

Create a Simple Cassandra Cluster With 3 Nodes

Category : How-to

Get Social!

datastax-logoApache Cassandra™ is a massively scalable open source NoSQL database. Cassandra is built from day 1 to be clustered to tick the usual clustering boxes; no single point of failure, and capacity and throughput scales with cluster size. This guide will look at creating a three node basic cluster within a single data centre.

It is assumed that the following is already in place:

  • Cassandra 3.2 is installed on 3 nodes.
  • Each node has open communication between the other nodes. Take a look here if you use a firewall between nodes.
  • The IP addresses of each node are known.
  • No data is stored on the 3 Cassandra instances.
  • The default file paths are used, for example from a yum or apt-get install.

Clear existing Cassandra data

If you’ve already started your Cassandra instance you’ll need to stop it and remove the data it contains. The main reason for this is because the cluster_name needs to be the same on all nodes, and it’s best to choose one for yourself rather than use the default Test Cluster.

service cassandra stop
rm -rf /var/lib/cassandra/data/system/*

Set the Cassandra cluster configuration

Cassandra is configured using various files in the /etc/cassandra directory. The cassandra.yaml contains most of the Cassandra configuration, such as ports used, file locations and seed node IP addresses. Edit your cassandra.yaml file and fill in the details as below. The below example is for brevity and your file may contain many other settings.

/etc/cassandra/cassandra.yaml
  • cluster_name can be anything chosen by you to describe the name of the cluster. Space is allowed but make sure you wrap everything in quotes. All members of this cluster must have the same name.
  • num_tokens is the number of virtual nodes within a Cassandra instance. This is used to partition the data and spread the data throughout the cluster. A good starting point is 256.
  • seeds are the IP addresses of the clusters seed servers. Seed nodes are used as known places where cluster information (such as a list of nodes in the cluster) can be obtained. They are not a single point of failure as all active nodes have this information, they are just known locations that can be relied on to have the information when other machines can come and go. It’s recommended that there are 3 seed nodes per data centre.
  • listen_address is the IP address that Cassandra will listen on for internal (Cassandra to Cassandra) communication will occur. Cassandra will try to guess your machines IP address if you leave it blank, but it’s best to specify it yourself. This will change on each node.
  • rpc_address is the IP address that Cassandra will listen on for client based communication, such as through the CQL protocol. This will change on each node.
  • endpoint_snitch is the ‘snitch’ used by Cassandra. A snitch is what tells Cassandra which data center and rack a node belongs to within a cluster. There are various types that could be used here, however describing them is outside the scope of this guide.

Example for node 1:

cluster_name: 'JC Cluster'
num_tokens: 256
seed_provider:
    - class_name: org.apache.cassandra.locator.SimpleSeedProvider
        - seeds: 10.0.0.1, 10.0.0.2
listen_address: 10.0.0.1
rpc_address: 10.0.0.1
endpoint_snitch: GossipingPropertyFileSnitch

Example for node 2:

cluster_name: 'JC Cluster'
num_tokens: 256
seed_provider:
    - class_name: org.apache.cassandra.locator.SimpleSeedProvider
        - seeds: 10.0.0.1, 10.0.0.2
listen_address: 10.0.0.2
rpc_address: 10.0.0.2
endpoint_snitch: GossipingPropertyFileSnitch

Example for node 3:

cluster_name: 'JC Cluster'
num_tokens: 256
seed_provider:
    - class_name: org.apache.cassandra.locator.SimpleSeedProvider
        - seeds: 10.0.0.1, 10.0.0.2
listen_address: 10.0.0.3
rpc_address: 10.0.0.3
endpoint_snitch: GossipingPropertyFileSnitch

Cassandra is built to be fault tolerant and will distribute data to try to minimize the risk of a failure causing loss of data or any downtime. Cassandra therefore has the understanding of a node, a rack and a data centre. Where possible, Cassandra will ensure that the data and it’s backups are stored on a different rack and a different data centre to ensure that failure, even at a data centre level isn’t catastrophic.

Edit the cassandra-rackdc.properties file on each node and set the dc and rack attributes. For this example we’ll assume everything is in the same dcdc1 however two nodes will be on rack1 and one node will be on rack2. Names are irrelevant, just come up with a naming standard that helps you understand where the Cassandra instance actually is. Everything here is case sensitive so be sure you’re consistent.

vi /etc/cassandra/cassandra-rackdc.properties

Example for node 1:

dc=uk_dc
rack=rack1

Example for node 2:

dc=uk_dc
rack=rack1

Example for node 3:

dc=uk_dc
rack=rack2

Finally, remove the config file cassandra-topology.properties from the etc dir as that’s not used with our configuration.

rm /etc/cassandra/cassandra-topology.properties

Starting your Cassandra cluster

The final steps are to start your cluster and connect to it.

First off, start your seed instances that were specified in the cassandra.yaml config file. Once these are up and running you can start the remaining nodes.

service cassandra start

Once all of your services are started you can use the nodetool status command to check the status of your nodes. Run this from any Cassandra server. As you can see with the below output, all three servers are available in the uk_dc data centre on rack1 and rack2.

nodetool status

Datacenter: uk_dc
===============
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
--  Address      Load       Tokens       Owns    Host ID                               Rack
UN  10.0.0.3  124.64 KB  256          ?       bb57bbee-3fe4-47b1-9249-cd3f90cd9718  rack2
UN  10.0.0.2  124.7 KB   256          ?       6669bac4-52c5-49fb-a68a-da065f20ae2c  rack1
UN  10.0.0.1  106.45 KB  256          ?       ddee28dd-7909-4428-bebd-023e4e560db5  rack1

Note: Non-system keyspaces don't have the same replication settings, effective ownership information is meaningless

Connect to your Cassandra Cluster

Once all of your servers have started your cluster is ready to use! Each node will have the cqlsh utility installed that you can use to interact with your Cassandra cluster. You’ll need to use one of the IP addresses Cassandra is listening on (set with rpc_address in cassandra.yaml).

cqlsh 10.0.0.1

Connected to Uptime2 at 10.0.0.1:9042.
[cqlsh 5.0.1 | Cassandra 3.2.1 | CQL spec 3.4.0 | Native protocol v4]
Use HELP for help.
cqlsh>

 


Install DataStax Cassandra 3 on Debain/ Ubuntu

Get Social!

datastax-logoThe following process will install the DataStax distribution of Cassandra on any Debian based system, such as Debian or Ubuntu. We’ll use the official DataStax apt repositories to install Cassandra using apt-get.

Cassandra is written in Java and the latest version (at the time of writing) of Cassandra requires Java 8. You can use either OpenJDK or visit my Oracle Java post on installing the Oracle distribution of Java 8.

Once you have Java 8 installed and confirmed you can continue to install Cassandra. Use java -version on the command line to check your Java version.

java -version
java version "1.8.0_91"
Java(TM) SE Runtime Environment (build 1.8.0_91-b14)
Java HotSpot(TM) 64-Bit Server VM (build 25.91-b14, mixed mode)

Install DataStax Cassandra 3.4

Note: You can use the simple Bash script to install Java 8 and a single instance of Cassandra found here.

The first step to installing Cassandra with apt is to add the DataStax Cassandra repository and repository key to your local apt list. Run the following command to add the required repository and update your local apt cache. Change the version 3.4 on the first line to match the version you’d like to install.

echo "deb http://debian.datastax.com/datastax-ddc 3.4 main" | tee -a /etc/apt/sources.list.d/cassandra.sources.list
curl -L https://debian.datastax.com/debian/repo_key | apt-key add -
apt-get update

Once you’re updated then run the install command and begin the Cassandra install.

apt-get install datastax-ddc

Debian’s policy is for software to be available locally after the install has completed and therefore your Cassandra instance will be running as a single stand alone node.

You can run the nodetool to check the install has completed and that the Cassandra instance is available.

nodetool status

Datacenter: datacenter1
=======================
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
--  Address     Load       Tokens       Owns (effective)  Host ID                               Rack
UN  10.10.10.10 1.29 MB    256          100.0%            c25d396e-fa09-4bf6-837d-676d57cd7987  rack1

The next step is to configure your Cassandra instance by editing the yaml config file /etc/cassandra/cassandra.yaml but that’s a blog post for another day.


MySQL Database Alternative

Category : How-to

Get Social!

mysql-logoThere are many alternatives to MySQL that tick some of the boxes you may require. There are plenty of free, open source databases such as PostgreSQL, MogoDB, CouchDB and Apache Derby however many of these databases have a different feature set or are a completely different type of database to the standard MySQL relational data model. Another question is what if I already have MySQL set up in my environment and would like to change? Many people are concerned that the (fairly) recent takeover of Sun Microsystems by Oracle could spell trouble for MySQL.

mariadb-logoMariaDB was created by some of the original developers of MySQL and was created by forking the MySQL code base. It[‘s maintained by the MariaDB Foundation who ensures the free availability of the database software. There are currently 2 main versions of MariaDB available as stable distributions:

  • Version 5.x – which is a drop in replacement for any version 5.x of MSSQL and promises to support all MySQL version 5.x functionality. With this version of MariaDB you can switch your applications to point at it and they will just work. All the connectors and client binaries are compatible and for most scenarios will work right out of the box.
  • Version 10.x – this is the new branch of MariaDB which drifts away from the original MySQL specification and introduces new features. Currently, most of the basics will work, but it no longer guarantees backwards compatibility with MySQL and can no longer be used as a drop in replacement. Eventually this had to happen to enable the MariaDB developers to implement the features they needed to and is the branch of MariaDB that’s being most actively developed.

As you can see, version 5.x is a version you can use straight away if you’re a MySQL 5.x user if you so choose. Version 10.x however, will take a little more time to implement as you’ll need to change the connectors in your applications and ensure that all the features you require from the database are the same. You may also wish to use some of Version 10.x’s features that would otherwise be unavailable.

Take a look at installing MariaDB on Ubuntu or installing a quick and easy database cluster.

 


Share a Single MDS Schema Between Multiple OBIEE Installations

Category : How-to

Get Social!

oracle-biIf you’ve installed OBIEE as many times as I have you’ll be more than familiar with the RCU and the two schemas it creates, MDS and BIPLATFORM. What you may not know, howerver, is that the MDS schema can be shared with multiple installations of OBIEE. A BIPLATFORM schema is still required for each OBIEE environment.

obiee_shared_mds

The above diagram shows a single MDS schema which is shared with each OBIEE environment; Dev, Test and Prod. Each of the OBIEE environments still has it’s own BI_PLATFORM schema. When running the RCU, you will need to run it 3 times in total; once with the MDS and BIPLATFORM components selected, and twice with just the BIPLATFORM component selected.

obiee-rcu-components

 

When it comes to installing OBIEE, you simpley use the same MDS details for all servers, but change the BIPLATFORM credentials to match the environment.

Obviously this method introduces it’s own problems with several environments using the same database when it comes to patching and upgrading. I’m not saying you should do it this way, just that you can!

 


Visit our advertisers

Quick Poll

Do you use GlusterFS in your workplace?

Visit our advertisers