diff --git a/examples/src/main/scala/org/apache/spark/examples/streaming/RecoverableNetworkWordCount.scala b/examples/src/main/scala/org/apache/spark/examples/streaming/RecoverableNetworkWordCount.scala
index 6af3a0f33efc2..d90a84d794d72 100644
--- a/examples/src/main/scala/org/apache/spark/examples/streaming/RecoverableNetworkWordCount.scala
+++ b/examples/src/main/scala/org/apache/spark/examples/streaming/RecoverableNetworkWordCount.scala
@@ -69,7 +69,7 @@ import org.apache.spark.util.IntParam
 
 object RecoverableNetworkWordCount {
 
-  def createContext(ip: String, port: Int, outputPath: String) = {
+  def createContext(ip: String, port: Int, outputPath: String, checkpointDirectory: String) = {
 
     // If you do not see this printed, that means the StreamingContext has been loaded
     // from the new checkpoint
@@ -79,6 +79,7 @@ object RecoverableNetworkWordCount {
     val sparkConf = new SparkConf().setAppName("RecoverableNetworkWordCount")
     // Create the context with a 1 second batch size
     val ssc = new StreamingContext(sparkConf, Seconds(1))
+    ssc.checkpoint(checkpointDirectory)
 
     // Create a socket stream on target ip:port and count the
     // words in input stream of \n delimited text (eg. generated by 'nc')
@@ -114,7 +115,7 @@ object RecoverableNetworkWordCount {
     val Array(ip, IntParam(port), checkpointDirectory, outputPath) = args
     val ssc = StreamingContext.getOrCreate(checkpointDirectory,
       () => {
-        createContext(ip, port, outputPath)
+        createContext(ip, port, outputPath, checkpointDirectory)
       })
     ssc.start()
     ssc.awaitTermination()