Wednesday, January 22, 2014

Running Cassandra 2.x.x on Windows 7 and 8

This blog post describes how to get the Cassandra 2.x.x family running on a windows machine.  It's clear that Cassandra should not be run for production on Windows (except perhaps on Azure), but if you're a student learning to use C* it may well be you have no choice to run it on Windows 7 or 8 on your laptop.  Lets get started : 

Install JRE 7

Open a command prompt and type java -version to see if it is installed properly.  If not find a jre from oracle and install it.  Make sure it's a version 7 at the least (version 6 will not work).

You'll need to set JAVA_HOME. Find the control panel (on windows 8 search for it).  Go to "system and security" and then "system". Click on "Advanced Settings" and then the "Environment Variables" button. Click on the new button an in the Variable name box type JAVA_HOME
under the value you'll need to put in the path to the java you are using.  Mine is

c:\program files\java\jre7

but yours may be different, especially if you have a jdk.  If you are going to program java clients for C* you will need a JDK but that's a different post

 

Install Cassandra

Download Cassandra from http://cassandra.apache.org/ probably a file like
apache-cassandra-2.0.4-bin.tar.gz
You'll need to unpack this file and that will depend on which flavor of windows you have.  At this point I'll assume you have a legal copy of winzip or similar.  Unpack the downloaded file to the root of the c: or d: drive on your machine

You can now change to the Cassandra install directory in your command prompt, change to the bin directory to start Cassandra, type Cassandra to start it.  The window will print a lot of information but you are looking for a line like:

 INFO 19:00:31,031 Listening for thrift clients...

to make sure it's working.

CQLSH

So now we have C* running, we need to check we can connect to it.  Start by opening another command prompt and type cqlsh. Sadly it won't start, cqlsh now needs an installation of python, so lets get one installed. Download one from  http://www.python.org/  and go to downloads then "individual release". click on the 2.x stable release and then
scroll down to the download section.  Your looking for the Windows MSI installer.  I used:

http://www.python.org/ftp/python/2.7.6/python-2.7.6.msi

Download it and run it to install Python, you'll need a version 2 of python, NOTE THIS WELL, version 3 will not work! This installs a nice windows version of python but does not install
a path to the executable.  You'll need to set it by hand I guess.  Once again Find the control panel (on windows 8 search for it).  Go to "system and security" and then "system"
Click on "Advanced Settings" and then the "Environment Variables" button.  Under the system variables find PATH.  Highlight it and click edit.

Careful!  We don't want to wipe the current contents (if you do hit cancel) go to the end of the current path and enter

;c:\python27

Note the ; at the beginning.  Again this will depend on the current version of python you've installed and should mirror the path to your python installation. Click OK to close the
dialog boxes and open a command prompt again.

Now you should be able to change to the cassandra directory and then the bin directory and type cqlsh.  With luck you should get the
cassandra cqlsh prompt:

Connected to Test Cluster at localhost:9160.
[cqlsh 4.1.0 | Cassandra 2.0.4 | CQL spec 3.1.1 | Thrift protocol 19.39.0]
Use HELP for help.
cqlsh>

type "use system;" followed by "describe  keyspaces;", cqlsh should reply:

system  system_traces

You're now connected and ready to start work.

BTW, folks at datastax and apache Cassandra, why is this so hard ?  Would Datastax Devcenter work easier ?

Setting up a Cassandra cluster on Windows with a vagrant virtualbox

Setting up Cassandra on windows can now be a pain with all it's dependancies, but it's something I'll cover in a later post.  One simpler way is to get C* running in a virtual box and perhaps even run it as a mini cluster.  This can be helped a lot by using Vagrant, but even that isn't quite straight forward.

The following has worked for me and is based heavily on the work done by calebgroom  and his github contribution vagrant-cassandra.  I've altered it a bit for use with the latest C* which adds virtual nodes etc.  Using these instructions you should be able to provisiona 3 node C* cluster with vnodes.

1: Install oracle Vm Virtualbox from https://www.virtualbox.org/ the latest version should do.
2: Install git for windows http://msysgit.github.io/ ensure you select the option to run git from the command line.
3: Install ruby for windows http://rubyinstaller.org/ V2.x.x Select all options
4: Download devkit DevKit-mingw64-32-4.7.2-20130224-1151-sfx.exe
4.1: Extract it to a permeant location
4.2 Start "commandline" with ruby from
4.3 Change to devkit location and run
    ruby dk.rb init
    ruby dk.rb install

5 At any location run gem install librarian-chef (This may take sometime)
6 Download vagrant (http://www.vagrantup.com/ ) and install it
7: Open a command prompt and git clone https://github.com/acobley/vagrant-cassandra.git
8:  change to the directory vagrant-cassandra\vagrant and run
   librarian-chef install
9: Open vagrant/cookbooks/java/attributes and edit default.rb so that
    default['java']['jdk_version'] = '7'
   
10: If you want comment, out the “DL is deprecated, please use Fiddle” warning at C:\HashiCorp\Vagrant\embedded\lib\ruby\2.0.0\dl.rb
11: change to the vagrant-cassandra and run
    Vagrant up
This could take some time, but once it's finished you should be able to ssh to the virtual machine if you have a ssh installed.

   vagrant ssh node1

If you don't have ssh installed the the git installation comes with a ssh client so add c:\program files\git\bin to your path

 set PATH=%PATH%;c:\program files\git\bin

 Or set the path environment variable from the control panel.

 You can then ssh to the virtual host
  
  ssh vagrant@127.0.0.1 -p 2222 -i c:/users/*username*/.vagrant.d/insecure_private_key
 
Once inside the virtual machines you can test and see if  it works by getting the c* status by typing

/usr/local/cassandra/bin/nodetool -h 192.168.2.10 status





You can bring down the cluster with vagrant halt  and remove it with vagrant destroy (but then you'll need to start again!)

Vagrant can also be run on a mac.  Make sure you have vitualbox installed, clone the https://github.com/acobley/vagrant-cassandra.git and follow the instructions in the readme.