I wish running Zeppelin on windows wasn't as hard as it is. Things go haiwire if you already have Spark installed on your computer. Zeppelin's embedded Spark interpreter does not work nicely with existing Spark and you may need to perform below steps (hacks!) to make it work. I am hoping that these will be fixed in newer Zeppelin versions.
If you try to run Zeppelin after extracting the package, you might encounter "The filename, directory name, or volume label syntax is incorrect."
Google search landed me to https://issues.apache.org/jira/browse/ZEPPELIN-1584, this link was helpful but wasn't enough to get Zeppelin working.
Below is what I had to do to make it work on my Windows 10 computer.
Existing software & configurations:
- Spark version 2.1.0
- Python version 3.6.1
- Zeppelin version zeppelin-0.7.2-bin-all
- My
SPARK_HOMEis "C:\Applications\spark-2.1.1-bin-hadoop2.7"
Steps
-
Extract Zeppelin package to a folder (mine was "C:\Applications\zeppelin-0.7.2-bin-all")
-
Copy jars from existing Spark installation into Zeppelin
$ cp %SPARK_HOME%\jars\*.jar %ZEPPELIN_HOME%\interpreter\spark
$ del %ZEPPELIN_HOME%\interpreter\spark\datanucleus*.jar- Copy
pysparkfrom existing Spark installation
$ cp %SPARK_HOME%\python\lib\*.zip %ZEPPELIN_HOME%\interpreter\spark- Rename
%ZEPPELIN_HOME%\conf\zeppelin-env.cmd.templateto%ZEPPELIN_HOME%\conf\zeppelin-env.cmd - Update
zeppelin-env.cmd
set PYTHONPATH=%SPARK_HOME%\python;%SPARK_HOME%\python\lib\py4j-0.10.4-src.zip;%SPARK_HOME%\python\lib\pyspark.zip- You have to subpress existing Spark installation to make it work nicely with Zeppelin. Add below line on top of
%ZEPPELIN_HOME%\bin\zeppelin.cmdfile
set SPARK_HOME=- Start Zeppelin and validate
Spark&pyspark
$ cd %ZEPPELIN_HOME%
$ bin\zeppelin.cmd- Open http://localhost:8080 and create a new notebook and try below code
Validate
pyspark
%pyspark
a=5*4
print("value = %i" % (a))
sc.versionIt should print: value = 20
'2.1.0'
%spark
sc.versionIt should print: '2.1.0'