Extract the time stamp from kafka messages in spar

Question

Trying to read from kafka source. I want to extract timestamp from message received to do structured spark streaming. kafka(version 0.10.0.0) spark streaming(version 2.0.1)

Answer 1

spark.read
  .format("kafka")
  .option("kafka.bootstrap.servers", "your.server.com:9092")
  .option("subscribe", "your-topic")
  .load()
  .select($"timestamp", $"value")

Field "timestamp" is what you are looking for. Type - java.sql.Timestamp. Make sure that you are connecting to 0.10 Kafka server. There is no timestamp in earlier versions. Full list of fields described here - http://spark.apache.org/docs/latest/structured-streaming-kafka-integration.html#creating-a-kafka-source-for-batch-queries

Answer 2

I'd suggest couple things:

1. Suppose you create a stream via latest Kafka Streaming Api (0.10 Kafka)

   E.g. you use dependency: "org.apache.spark" %% "spark-streaming-kafka-0-10" % 2.0.1

   Than you create a stream, according to the docs above:

    val kafkaParams = Map[String, Object](
        "bootstrap.servers" -> "broker1:9092,broker2:9092",
        "key.deserializer" -> classOf[StringDeserializer],
        "value.deserializer" -> classOf[ByteArrayDeserializer],
        "group.id" -> "spark-streaming-test",
        "auto.offset.reset" -> "latest",
        "enable.auto.commit" -> (false: java.lang.Boolean))
   
   val sparkConf = new SparkConf()
   // suppose you have 60 second window
   val ssc = new StreamingContext(sparkConf, Seconds(60))
   ssc.checkpoint("checkpoint")
   
   val stream = KafkaUtils.createDirectStream(ssc, PreferConsistent,
   Subscribe[String, Array[Byte]](topics, kafkaParams))
   

2. Your stream will be an DStream of ConsumerRecord[String,Array[Byte]] and you can get a timestamp and key-value as simple as:

   stream.map { record => (record.timestamp(), record.key(), record.value())  }
   

Hope that helps.