The Apache Zeppelin distribution includes a scripts directory
scripts/vagrant/zeppelin-dev
This script creates a virtual machine that launches a repeatable, known set of core dependencies required for developing Zeppelin. It can also be used to run an existing Zeppelin build if you don't plan to build from source. For pyspark users, this script also includes several helpful Python Libraries
This script requires three applications, Ansible, Vagrant and Virtual Box. All of these applications are freely available as Open Source projects and extremely easy to set up on most operating systems.
If you are running Windows and don't yet have python installed, install Python 2.7.x Python Windows Installer
sudo easy_install pip thensudo pip install ansibleansible --version should now report version 1.9.2 or highervagrant up from within the /scripts/vagrant/zeppelin-dev directoryThats it!
You can now run vagrant ssh and this will place you into the guest machines terminal prompt.
If you don't wish to build Zeppelin from scratch, run the z-manager installer script while running in the guest VM:
curl -fsSL https://raw.githubusercontent.com/NFLabs/z-manager/master/zeppelin-installer.sh | bash
You can now git clone https://github.com/apache/incubator-zeppelin.git into a directory on your host machine, or directly in your virtual machine.
Cloning zeppelin into the /scripts/vagrant/zeppelin-dev directory from the host, will allow the directory to be shared between your host and the guest machine.
Cloning the project again may seem counter intuitive, since this script likley originated from the project repository. Consider copying just the vagrant/zeppelin-dev script from the zeppelin project as a stand alone directory, then once again clone the specific branch you wish to build.
Synced folders enable Vagrant to sync a folder on the host machine to the guest machine, allowing you to continue working on your project's files on your host machine, but use the resources in the guest machine to compile or run your project. (1) Synced Folder Description from Vagrant Up
By default, Vagrant will share your project directory (the directory with the Vagrantfile) to /vagrant. Which means you should be able to build within the guest machine after you
cd /vagrant/incubator-zeppelin
Running the following commands in the guest machine should display these expected versions:
node --version should report v0.12.7
mvn --version should report Apache Maven 3.3.3 and Java version: 1.7.0_85
The virtual machine consists of:
This assumes you've already cloned the project either on the host machine in the zeppelin-dev directory (to be shared with the guest machine) or cloned directly into a directory while running inside the guest machine.
cd /incubator-zeppelin
mvn clean package -Pspark-1.5 -Ppyspark -Dhadoop.version=2.2.0 -Phadoop-2.2 -DskipTests
./bin/zeppelin-daemon.sh start
On your host machine browse to http://localhost:8080/
If you turned off port forwarding in the Vagrantfile browse to http://192.168.51.52:8080
If you plan to run this virtual machine along side other Vagrant images, you may wish to bind the virtual machine to a specific IP address, and not use port fowarding from your local host.
Comment out the forward_port line, and uncomment the private_network line in Vagrantfile. The subnet that works best for your local network will vary so adjust 192.168.*.* accordingly.
#config.vm.network "forwarded_port", guest: 8080, host: 8080
config.vm.network "private_network", ip: "192.168.51.52"
vagrant halt followed by vagrant up will restart the guest machine bound to the IP address of 192.168.51.52.
This approach usually is typically required if running other virtual machines that discover each other directly by IP address, such as Spark Masters and Slaves as well as Cassandra Nodes, Elasticsearch Nodes, and other Spark data sources. You may wish to launch nodes in virtual machines with IP Addresses in a subnet that works for your local network, such as: 192.168.51.53, 192.168.51.54, 192.168.51.53, etc..
With zeppelin running, Numpy, SciPy, Pandas and Matplotlib will be available. Create a pyspark notebook, and try
%pyspark
import numpy
import scipy
import pandas
import matplotlib
print "numpy " + numpy.__version__
print "scipy " + scipy.__version__
print "pandas " + pandas.__version__
print "matplotlib " + matplotlib.__version__
To Test plotting using matplotlib into a rendered %html SVG image, try
%pyspark
import matplotlib
matplotlib.use('Agg') # turn off interactive charting so this works for server side SVG rendering
import matplotlib.pyplot as plt
import numpy as np
import StringIO
# clear out any previous plots on this notebook
plt.clf()
def show(p):
img = StringIO.StringIO()
p.savefig(img, format='svg')
img.seek(0)
print "%html <div style='width:600px'>" + img.buf + "</div>"
# Example data
people = ('Tom', 'Dick', 'Harry', 'Slim', 'Jim')
y_pos = np.arange(len(people))
performance = 3 + 10 * np.random.rand(len(people))
error = np.random.rand(len(people))
plt.barh(y_pos, performance, xerr=error, align='center', alpha=0.4)
plt.yticks(y_pos, people)
plt.xlabel('Performance')
plt.title('How fast do you want to go today?')
show(plt)