ok guys , finally figured out how to get it running. I have detailed out the
steps I did. Perhaps its clear to all you folks. To me it was not :-) Our
Hadoop development environment
- 3 node development hadoop cluster
- Current version CDH 5.3.3
- Hive 0.13.1
- Spark 1.2.0 (standalone mode)
- node1 (worker1, master)
- node2 (worker2)
- node3 (worker3)
- Cloudera Manager to manage and update(using parcels)
Steps to get spark-sql running
- On every node(node1, node2, node3 above)
- sudo cp -avi /etc/hive/conf/hive-site.xml /etc/spark/conf
- Edit and add a line
- sudo vi /opt/cloudera/parcels/CDH/lib/spark/bin/compute-classpath.sh
- # added by sanjay for running Spark using hive metadata
- CLASSPATH="$CLASSPATH:/opt/cloudera/parcels/CDH/lib/hive/lib/*"
- Run spark SQL in CLI mode
- /opt/cloudera/parcels/CDH/lib/spark/bin/spark-sql
- Run spark SQL in async mode
- /opt/cloudera/parcels/CDH/lib/spark/bin/spark-sql -e "select * from
band.beatles where upper(first_name) like '%GEORGE%' "
- Run spark SQL in "SQL File" mode
- /opt/cloudera/parcels/CDH/lib/spark/bin/spark-sql -f get_names.hql
From: Andrew Otto <[email protected]>
To: Sanjay Subramanian <[email protected]>
Cc: user <[email protected]>
Sent: Thursday, May 28, 2015 7:26 AM
Subject: Re: Pointing SparkSQL to existing Hive Metadata with data file
locations in HDFS
val sqlContext = new HiveContext(sc)val schemaRdd = sqlContext.sql("some
complex SQL")
It mostly works, but have been having issues with tables that contains a large
amount of data:
https://issues.apache.org/jira/browse/SPARK-6910
On May 27, 2015, at 20:52, Sanjay Subramanian
<[email protected]> wrote:
hey guys
On the Hive/Hadoop ecosystem we have using Cloudera distribution CDH 5.2.x ,
there are about 300+ hive tables.The data is stored an text (moving slowly to
Parquet) on HDFS.I want to use SparkSQL and point to the Hive metadata and be
able to define JOINS etc using a programming structure like this
import org.apache.spark.sql.hive.HiveContextval sqlContext = new
HiveContext(sc)val schemaRdd = sqlContext.sql("some complex SQL")
Is that the way to go ? Some guidance will be great.
thanks
sanjay