added ssh tunneling instructions

Sparking_Water_EMR_instructions.md

@@ -6,8 +6,7 @@ Run thes
 
 	# Log in as Hadoop user and make ec2-user directories
 	# If you don't do this step your Spark code will immediately fail with permission issues
-	sudo su
-	su hadoop
+	sudo su hadoop
 	hadoop fs -mkdir -p /user/ec2-user
 	hadoop fs -chown ec2-user /user/ec2-user
 
@@ -29,4 +28,26 @@ Run thes
 	export MASTER="yarn-client"
 
 	# Start up pysparking
-	/home/ec2-user/sparkling-water-1.6.8/bin/pysparkling
+	/home/ec2-user/sparkling-water-1.6.8/bin/pysparkling --deploy-mode client
+
+	# Note the shell's tracking URL, which will look something like this:
+	http://ip-10-0-0-123.ec2.internal:20888/proxy/application_1477154041215_0004/
+
+	# Open up a new Terminal tab. We're doing to do port forwarding / ssh tunneling to view the Spark UI
+	ssh -i /Users/ryanzotti/Documents/private_keys/ML.pem -L 20888:[localhost]:20888 [email protected]
+
+	# Open up your web browser to the tracking URL. Replace the IP with localhost
+	http://localhost:20888/proxy/application_1477154041215_0004/
+
+
+## FAQ
+
+**Question:** You get a never-ending stream of: 
+
+    Client: Application report for application_1477154041215_0013 (state: ACCEPTED)
+
+**Answer:** You probably have too many simultaneous (potentially abandoned) spark shells running.
+
+	ps -ef | grep -i spark
+	kill -9 <spark process id>
+