Installing Cassandra on Linux - Single Node Cluster

Install Cassandra on Linux

Prerequisites

Before installing Cassandra make sure the following prerequisites are met:

·               Root or sudo access to the install machine.

·               Oracle Java SE Runtime Environment (JRE)

              ·               Java Native Access (JNA) is required for production installations.

Check which version of Java is installed by running the following command in a terminal window:

java -version

echo $JAVA_HOME

To install Cassandra, download the binary files from the website, unpack them:

tar -xvzf apache-cassandra-1.2.16-bin.tar.gz

mv apache-cassandra-1.2.16 ~/cassandra

Configure Cassandra

Open the cassandra.yaml: file, which will be available in the

 <installation-directory>/conf directory of Cassandra.

$ vi cassandra.yaml

The above command opens the cassandra.yaml file.

Verify the following configurations.

By default, these values will be set to the specified directories.

·   data_file_directories “/var/lib/cassandra/data”

·   commitlog_directory “/var/lib/cassandra/commitlog”

·   saved_caches_directory “/var/lib/cassandra/saved_caches”

Note: To run a single-node test cluster of Cassandra, you aren’t

 going to need to change anything on the cassandra.yaml file.

Simply run:

./bin/cassandra

Next, make sure that the folders Cassandra accesses, such as the

log folder, exists and that Cassandra has the right to write on it:

mkdir /var/lib/cassandra

mkdir /var/log/cassandra

chown -R $USER:$GROUP /var/lib/cassandra

chown -R $USER:$GROUP /var/log/cassandra

Now set Cassandra’s variables by running:

export CASSANDRA_HOME=~/cassandra

export PATH=$PATH:$CASSANDRA_HOME/bin

Running Cassandra

To start Cassandra, open the terminal window, navigate to Cassandra

 home directory/home, where you unpacked Cassandra, and run

the following command to start your Cassandra server.

$ cd $CASSANDRA_HOME

$./bin/cassandra -f

Using the –f option tells Cassandra to stay in the foreground instead of

running as a background process.

If everything goes fine, you can see the Cassandra server starting.

And then run:

./bin/cassandra-cli                        

and if it says "Connected to: 'Test Cluster'", you are now running your

single-node cluster.

Starting and Stopping Cassandra

Starting and Stopping Cassandra as a service

                service cassandra start

                service cassandra stop

Cassandra operations in case of tarball installations:

                Starting Cassandra:        

                                $ cd <install_location>                  

                                $ bin/cassandra

                                - where <install_location> is where you have installed cassandra

                                - bin/cassandra will start cassandra as a background process

                                - bin/cassandra -f will start cassandra in foreground

                Stopping Cassandra:

                - As it's only a process and not a service you could do something like

                - where <pid> is what you get from ps auwx | grep cassandra

                                $ ps auwx | grep cassandra

                                $ sudo kill <pid>

Best Practices:

                - You could check if cassandra is running by checking the port, if you got a result back that
                   means it is running

                                $ lsof -i :9160

                - If you want to kill it , do kill -9 "pid"

                - Below procedure is suggested that minimizes the risk of something going wrong.

                - The other advantage of clean Cassandra restart procedure is saving some startup time.

                - A two-node cluster with nodes known as node01.hostserver.net and node02.hostserver.net :

                                 # nodetool -h node01.hostserver.net disablegossip         

                                 # nodetool -h node01.hostserver.net disablethrift

                                 # nodetool -h node01.hostserver.net drain

                disablegossip :  Because it makes node look like “dead” for other nodes.

                disablethrift   :  Turning it off makes Cassandra unable to accept user’s requests because it 

                                           disables Cassandra’s RPC server.

                drain              :   It flushes column families. In other words – converts Memtables into

                                            immutable SSTables, emptying Commit Log this way