Created
September 17, 2015 08:49
-
-
Save natewave/094344befa19b2fa0fb4 to your computer and use it in GitHub Desktop.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
def read[A, B](consumerConfig: ConsumerConfig, streamingContext: StreamingContext, maxRate: Int = DEFAULT_MAX_RATE) | |
(keyRule:GenericRecord => VA[A], valueRule: GenericRecord => VA[B]): DStream[(A, B)] = { | |
val sparkConf = streamingContext.sparkContext.getConf | |
val appName = sparkConf.get("spark.app.name") | |
val offsetsCoordinator = OffsetsCoordinator.get( | |
channel = OffsetsCoordinator.newChannel(consumerConfig.host, consumerConfig.port), | |
clientId = consumerConfig.clientId, | |
groupId = consumerConfig.groupId) | |
offsetsCoordinator.acquireAndGet { coordinator => | |
val topicPartitions: Seq[TopicAndPartition] = | |
OffsetsOperations.getTopicPartitions(consumerConfig.topic, coordinator.broker, consumerConfig.clientId) | |
val lastOffsets: Try[Map[TopicAndPartition, Long]] = | |
OffsetsOperations.fetch( | |
topicsAndPartitions = topicPartitions, | |
groupId = consumerConfig.groupId, | |
clientId = consumerConfig.clientId, | |
coordinator = coordinator) | |
lastOffsets match { | |
case Success(offsets: Map[TopicAndPartition, Long]) => { | |
// set maxRate if not already set, otherwise spark will try to load all the data into one RDD | |
if (!sparkConf.contains("spark.streaming.kafka.maxRatePerPartition")) { | |
sparkConf.set( | |
"spark.streaming.kafka.maxRatePerPartition", | |
maxRate.toString) | |
} | |
// | |
val params = consumerConfig.params | |
val messageHandler = (mmd: MessageAndMetadata[GenericRecord, GenericRecord]) => (mmd.key, mmd.message) | |
val genericKafkaStream: DStream[(GenericRecord, GenericRecord)] = KafkaUtils.createDirectStream[GenericRecord, GenericRecord, GenericRecordDecoder, GenericRecordDecoder, (GenericRecord, GenericRecord)](streamingContext, params, offsets, messageHandler) | |
val validatedStream: DStream[(VA[A], VA[B])] = genericKafkaStream.transform { rdd => | |
rdd.map(record => (keyRule(record._1), valueRule(record._2))) | |
} | |
// | |
val listener = new KafkaListener(offsetRanges => commitOffsets(offsetRanges, consumerConfig.groupId, consumerConfig.clientId, coordinator)) | |
streamingContext.sparkContext.addSparkListener(listener) | |
validatedStream.transform { rdd => | |
val offsetRanges = rdd.asInstanceOf[HasOffsetRanges].offsetRanges | |
val rddId = rdd.id | |
val successRDD = rdd.filter(kv => kv._1.isSuccess && kv._2.isSuccess).map(kv => (kv._1.get, kv._2.get)) | |
listener.registerKafkaRDD(rddId, offsetRanges) | |
successRDD | |
} | |
} | |
case Failure(reason) => throw new RuntimeException("Could not fetch offsets. Refusing to start from beginning") | |
} | |
} | |
} |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment