use HDFS in tutorial notebook so that it works with yarn-client

[ejono]

https://issues.apache.org/jira/browse/ZEPPELIN-55

Assumptions:

• HDFS is available
• HDFS is configured with a default master node so that it does not need to be specified in the HDFS path
• The HDFS directory /user/zeppelin exists, and the user running Zeppelin has access to write to this directory
• The "hdfs" tool is in the PATH

Please let me know if you have any ideas for how to address this in a better way, since I realize that these assumptions might not hold for every installation.

[Leemoonsoo]

Hi, Thanks for contribution.

As you mentioned, not every installation will have hdfs too.
And i think tutorial notebook is more like for quickly taste Zeppelin.

How about add hdfs related code as a comment?
with message like 'uncomment these codes if you want to try with hdfs'.

for example

// "hdfs dfs -rm -r /user/zeppelin/data" !               // uncomment if you want to try with hdfs
// "hdfs dfs -copyFromLocal data /user/zeppelin" !       // uncomment if you want to try with hdfs
...
// val bankText = sc.textFile(s"hdfs:///user/zeppelin/data/bank-full.csv")    // uncomment if you want to try with hdfs

What do you think?