我的项目包括ZooKeeper,Kafka和Spark Streaming。当我尝试使用Spark Streaming将Kafka偏移量写入ZooKeeper时,问题是zkClient
无法序列化。我见过几个GitHub项目,例如:https://github.com/ippontech/spark-kafka-source
//save the offsets
kafkaStream.foreachRDD(rdd => offsetsStore.saveOffsets(topic, rdd))
def saveOffsets(topic: String, rdd: RDD[_]): Unit = {
logger.info("Saving offsets to ZooKeeper")
val stopwatch = new Stopwatch()
val offsetsRanges = rdd.asInstanceOf[HasOffsetRanges].offsetRanges
offsetsRanges.foreach(offsetRange => logger.debug(s"Using ${offsetRange}"))
val offsetsRangesStr = offsetsRanges.map(offsetRange => s"${offsetRange.partition}:${offsetRange.fromOffset}").mkString(",")
logger.debug(s"Writing offsets to ZooKeeper: ${offsetsRangesStr}")
**ZkUtils.updatePersistentPath(zkClient, zkPath, offsetsRangesStr)**
logger.info("Done updating offsets in ZooKeeper. Took " + stopwatch)
}
代码:kafkaStream.foreachRDD(rdd => offsetsStore.saveOffsets(rdd))
将在对象private val zkClient = new ZkClient(zkHosts, 30000, 30000,ZKStringSerializer)
的驱动程序offsetStore
中执行,但zkClient
无法序列化,它是如何工作的?
答案 0 :(得分:0)
您可以将zkClient
定义为@transient lazy val
,这意味着它不会在驱动程序和执行程序之间进行序列化(这是@transient
部分),而是将在每个部分重新初始化以及包含上述代码的类的每个实例(这是lazy
部分)。
您可以在此处详细了解此模式: http://fdahms.com/2015/10/14/scala-and-the-transient-lazy-val-pattern/