Spark 1.6 Streaming consumer reading in kafka offset stuck at createDirectStream

Question

I am trying to read in the spark streaming offset into my consumer but I cannot seem to do it correctly.

Here is my code.

val dfoffset = hiveContext.sql(s"select * from $db")
dfoffset.show()
val dfoffsetArray = dfoffset.collect()
println("printing array of data")
dfoffsetArray.foreach(println)
val fromOffsets = collection.mutable.Map[TopicAndPartition, Long]()
for (i <- dfoffsetArray) {
  val topicAndPartition = (TopicAndPartition(i(1).toString, i(0).toString.toInt) -> (i(2).toString.toLong))
  fromOffsets += topicAndPartition
}

val kafkaParams = Map[String, String]("bootstrap.servers" -> serverName, "group.id" -> "test")
val topics = Array(topicName).toSet
//stuck here 
var directKafkaStream = KafkaUtils.createDirectStream[String, String, StringDecoder, StringDecoder](ssc, kafkaParams, topics)

directKafkaStream.foreachRDD(rdd1 => { ..

Here is the output from showing the dataframe

partition_number|topic_name|current_offset|
+----------------+----------+--------------+
|               0|TOPIC_NAME|          4421|

Any help is greatly appreciated.

I am using spark 1.6 , Scala 2.10.5, kafka 10

score 1 · Accepted Answer · answered Dec 15 '17 at 09:35

As the official document shown KafkaUtils.createDirectStream, you should pass the fromOffsets as the 3rd parameter of createDirectStream(and don't forget about the 4th parameter messageHandler).

The fromOffsets parameter suppose to be a collection.immutable.Map[TopicAndPartition, Long], we usually use immutable instead of mutable as possible in Scala.
You may transform the dfoffsetArray to immutable.Map[TopicAndPartition, Long] with the following:

val fromOffsets = dfoffsetArray.map( i =>
  TopicAndPartition(i(1).toString, i(0).toString.toInt) -> (i(2).toString.toLong)
).toMap

And the messageHandler is type of (MessageAndMetadata[K, V]) ⇒ R), which deals key and value of messages. You can define a simple handler as the following:

val messageHandler =
  (mmd: MessageAndMetadata[String, String]) => (mmd.key, mmd.message)

Then your createDirectStream will look like...

KafkaUtils.createDirectStream[String, String, StringDecoder, StringDecoder,
  (String, String)](ssc, kafkaParams, fromOffsets, messageHandler)

Now you are free to do some transformation to your stream. Happy streaming!

I was tutored by this article months ago. Maybe you will find it helpful.

Thanks a lot happy streaming! – javadev Dec 15 '17 at 16:59 — javadev, Dec 15 '17 at 16:59

Spark 1.6 Streaming consumer reading in kafka offset stuck at createDirectStream

1 Answers1