https://community.cloudera.com/t5/Support-Questions/Spark-How-to-make-calls-to-database-using-foreachPartition/m-p/123340#M86084 <P><A rel="user" href="https://community.cloudera.com/users/13772/chmamidala.html" nodeid="13772">@Aditya Mamidala</A></P><P>Here's a working example of foreachPartition that I've used as part of a project. This is part of a Spark Streaming process, where "event" is a DStream, and each stream is written to HBase via Phoenix (JDBC). I have a structure similar to what you tried in your code, where I first use foreachRDD then foreachPartition. </P><PRE> event.map(x =&gt; x._2 ).foreachRDD { rdd =&gt; rdd.foreachPartition { rddpartition =&gt; val thinUrl = "jdbc:phoenix:phoenix.dev:2181:/hbase" val conn = DriverManager.getConnection(thinUrl) rddpartition.foreach { record =&gt; conn.createStatement().execute("UPSERT INTO myTable VALUES (" + record._1 + ")" ) } conn.commit() } } </PRE><P>The full project is located <A href="https://github.com/zaratsian/network_topology_analysis/blob/master/SparkNetworkAnalysis/src/main/scala/com/github/zaratsian/SparkStreaming/SparkNetworkAnalysis.scala">here</A>. </P> Mon, 27 Feb 2017 21:11:03 GMT dzaratsian 2017-02-27T21:11:03Z