Rev5 updates

to align with HDP 2.2 version of the course offering

README.md

@@ -1,6 +1,6 @@
 # DevPH_Labs
 
-Rev4 /labs files for HDP 2.1 version of the
+Rev5 /labs files for HDP 2.2 version of the
 http://hortonworks.com/training/class/hadoop-2-data-analysis-pig-hive/ course.
 
-Files are a replay of the /labs folder from https://github.com/HortonworksUniversity/DevPH_Rev4.
+Status: Currently in Production

labs/Lab6.4/sessions.pig

@@ -1,5 +1,5 @@
 REGISTER datafu-1.2.0.jar;
-REGISTER /usr/lib/pig/piggybank.jar;
+REGISTER /usr/hdp/2.2.0.0-2041/pig/piggybank.jar;
 
 DEFINE UnixToISO org.apache.pig.piggybank.evaluation.datetime.convert.UnixToISO();
 

labs/Lab7.2/Solution/mostvisits.hive

@@ -1,8 +1,5 @@
-SELECT ordertotal, orders.userid, itemlist 
-FROM orders 
-JOIN max_ordertotal ON 
-max_ordertotal.userid = orders.userid 
-AND 
-max_ordertotal.maxtotal = orders.ordertotal;
-
-select order_date, sum(ordertotal) FROM orders GROUP BY order_date;
+select count(*) as visit_count, fname, lname
+from wh_visits
+     group by fname, lname                          
+     order by visit_count DESC                    
+     limit 20; 

labs/Lab7.5/create_email_table.hive

@@ -1,16 +1,16 @@
 CREATE TABLE hive_user_email
+STORED AS AVRO
+TBLPROPERTIES ('avro.schema.url'='hdfs://sandbox.hortonworks.com:8020/user/root/mbox.avsc');
 
-ROW FORMAT
 
-SERDE 'org.apache.hadoop.hive.serde2.avro.AvroSerDe'
 
-WITH SERDEPROPERTIES (
-    'avro.schema.url'='hdfs:///user/root/mbox.avsc')
-
-STORED AS
-
-INPUTFORMAT 'org.apache.hadoop.hive.ql.io.avro.AvroContainerInputFormat'
-
-OUTPUTFORMAT 'org.apache.hadoop.hive.ql.io.avro.AvroContainerOutputFormat'
-;
+#CREATE TABLE hive_user_email
+#ROW FORMAT
+#SERDE 'org.apache.hadoop.hive.serde2.avro.AvroSerDe'
+#WITH SERDEPROPERTIES (
+#    'avro.schema.url'='hdfs:///user/root/mbox.avsc')
+#STORED AS
+#INPUTFORMAT 'org.apache.hadoop.hive.ql.io.avro.AvroContainerInputFormat'
+#OUTPUTFORMAT 'org.apache.hadoop.hive.ql.io.avro.AvroContainerOutputFormat'
+#;
 

labs/Oozie/job.properties

@@ -1,10 +1,10 @@
-oozie.wf.application.path=hdfs://namenode:8020/user/root/congress
+oozie.wf.application.path=hdfs://sandbox:8020/user/root/congress
 
 #Hadoop RM
-resourceManager=resourcemanager:8050
+resourceManager=sandbox:8050
 
 #Hadoop fs.default.name
-nameNode=hdfs://namenode:8020/
+nameNode=hdfs://sandbox:8020/
 
 #Hadoop mapred.queue.name
 queueName=default