Spark Streaming源码解读之State管理之updateStateByKey和mapWithState解密

Posted 2023-02-22 snail_gesture

背景：
整个Spark Streaming是按照Batch Duractions划分Job的。但是很多时候我们需要算过去的一天甚至一周的数据，这个时候不可避免的要进行状态管理，而Spark Streaming每个Batch Duractions都会产生一个Job，Job里面都是RDD，所以此时面临的问题就是怎么对状态进行维护？这个时候就需要借助updateStateByKey和mapWithState方法完成核心的步骤。
源码分析：
1. 无论是updateStateByKey还是mapWithState方法在DStream中均没有，但是是通过隐身转换函数实现其功能。

object DStream 

  // `toPairDStreamFunctions` was in SparkContext before 1.3 and users had to
  // `import StreamingContext._` to enable it. Now we move it here to make the compiler find
  // it automatically. However, we still keep the old function in StreamingContext for backward
  // compatibility and forward to the following function directly.

  implicit def toPairDStreamFunctions[K, V](stream: DStream[(K, V)])
      (implicit kt: ClassTag[K], vt: ClassTag[V], ord: Ordering[K] = null):
    PairDStreamFunctions[K, V] = 
    new PairDStreamFunctions[K, V](stream)
  

updateStateByKey：
1. 在PairDStreamFunctions中updateStateByKey具体实现如下：
在已有的历史基础上，updateFunc对历史数据进行更新。该函数的返回值是DStream类型的。

/**
 * Return a new "state" DStream where the state for each key is updated by applying
 * the given function on the previous state of the key and the new values of each key.
 * Hash partitioning is used to generate the RDDs with Spark's default number of partitions.
 * @param updateFunc State update function. If `this` function returns None, then
 *                   corresponding state key-value pair will be eliminated.
 * @tparam S State type
 */
def updateStateByKey[S: ClassTag](
    updateFunc: (Seq[V], Option[S]) => Option[S]
  ): DStream[(K, S)] = ssc.withScope 
// defaultPartitioner
  updateStateByKey(updateFunc, defaultPartitioner())

2.  defaultPartitioner:

private[streaming] def defaultPartitioner(numPartitions: Int = self.ssc.sc.defaultParallelism) = 
  new HashPartitioner(numPartitions)

3.  partitioner就是控制RDD的每个patition

/**
 * Return a new "state" DStream where the state for each key is updated by applying
 * the given function on the previous state of the key and the new values of the key.
 * org.apache.spark.Partitioner is used to control the partitioning of each RDD.
 * @param updateFunc State update function. If `this` function returns None, then
 *                   corresponding state key-value pair will be eliminated.
 * @param partitioner Partitioner for controlling the partitioning of each RDD in the new
 *                    DStream.
 * @tparam S State type
 */
def updateStateByKey[S: ClassTag](
    updateFunc: (Seq[V], Option[S]) => Option[S],
    partitioner: Partitioner
  ): DStream[(K, S)] = ssc.withScope 
  val cleanedUpdateF = sparkContext.clean(updateFunc)
  val newUpdateFunc = (iterator: Iterator[(K, Seq[V], Option[S])]) => 
    iterator.flatMap(t => cleanedUpdateF(t._2, t._3).map(s => (t._1, s)))
  
  updateStateByKey(newUpdateFunc, partitioner, true)

4.  rememberPartitioner默认为true

/**
 * Return a new "state" DStream where the state for each key is updated by applying
 * the given function on the previous state of the key and the new values of each key.
 * org.apache.spark.Partitioner is used to control the partitioning of each RDD.
 * @param updateFunc State update function. Note, that this function may generate a different
 *                   tuple with a different key than the input key. Therefore keys may be removed
 *                   or added in this way. It is up to the developer to decide whether to
 *                   remember the partitioner despite the key being changed.
 * @param partitioner Partitioner for controlling the partitioning of each RDD in the new
 *                    DStream
 * @param rememberPartitioner Whether to remember the paritioner object in the generated RDDs.
 * @tparam S State type
 */
def updateStateByKey[S: ClassTag](
    updateFunc: (Iterator[(K, Seq[V], Option[S])]) => Iterator[(K, S)],
    partitioner: Partitioner,
    rememberPartitioner: Boolean
  ): DStream[(K, S)] = ssc.withScope 
   new StateDStream(self, ssc.sc.clean(updateFunc), partitioner, rememberPartitioner, None)

5.  在StateDStream中，StorageLevel是直接存储到磁盘，因为此时的数据非常大

class StateDStream[K: ClassTag, V: ClassTag, S: ClassTag](
    parent: DStream[(K, V)],
    updateFunc: (Iterator[(K, Seq[V], Option[S])]) => Iterator[(K, S)],
    partitioner: Partitioner,
    preservePartitioning: Boolean,
    initialRDD : Option[RDD[(K, S)]]
  ) extends DStream[(K, S)](parent.ssc) 

  super.persist(StorageLevel.MEMORY_ONLY_SER)

1. 在computeUsingPreiviousRDD源码如下：

private [this] def computeUsingPreviousRDD (
  parentRDD : RDD[(K, V)], prevStateRDD : RDD[(K, S)]) = 
  // Define the function for the mapPartition operation on cogrouped RDD;
  // first map the cogrouped tuple to tuples of required type,
  // and then apply the update function
  val updateFuncLocal = updateFunc
  val finalFunc = (iterator: Iterator[(K, (Iterable[V], Iterable[S]))]) => 
    val i = iterator.map(t => 
      val itr = t._2._2.iterator
      val headOption = if (itr.hasNext) Some(itr.next()) else None
      (t._1, t._2._1.toSeq, headOption)
    )
    updateFuncLocal(i)
  
//cogroup每次计算的时候都会遍历prevSrateRDD中的所有parititioner的信息
//
  val cogroupedRDD = parentRDD.cogroup(prevStateRDD, partitioner)
  val stateRDD = cogroupedRDD.mapPartitions(finalFunc, preservePartitioning)
  Some(stateRDD)

所以，如果数据很多的时候不建议使用updateStateByKey。
updateStateByKey函数实现如下：

mapWithState:
1. 返回MapWithStateDStream函数，维护和更新历史状态都是基于Key。使用一个function对key-value形式的数据进行状态维护。

/**
 * :: Experimental ::
 * Return a [[MapWithStateDStream]] by applying a function to every key-value element of
 * `this` stream, while maintaining some state data for each unique key. The mapping function
 * and other specification (e.g. partitioners, timeouts, initial state data, etc.) of this
 * transformation can be specified using [[StateSpec]] class. The state data is accessible in
 * as a parameter of type [[State]] in the mapping function.
 *
 * Example of using `mapWithState`:
 * 
 *    // A mapping function that maintains an integer state and return a String
//此时的state就可以看成一张表，这张表记录了状态维护中所有的历史状态。
 *    def mappingFunction(key: String, value: Option[Int], state: State[Int]): Option[String] = 
 *      // Use state.exists(), state.get(), state.update() and state.remove()
 *      // to manage state, and return the necessary string
 *    
 *
 *    val spec = StateSpec.function(mappingFunction).numPartitions(10)
 *
 *    val mapWithStateDStream = keyValueDStream.mapWithState[StateType, MappedType](spec)
 * 
 *
 * @param spec          Specification of this transformation
 * @tparam StateType    Class type of the state data
 * @tparam MappedType   Class type of the mapped data
 */
@Experimental
def mapWithState[StateType: ClassTag, MappedType: ClassTag](
    spec: StateSpec[K, V, StateType, MappedType]
  ): MapWithStateDStream[K, V, StateType, MappedType] = 
  new MapWithStateDStreamImpl[K, V, StateType, MappedType](
    self,
// StateSpecImpl类封装了StateSpec操作。
    spec.asInstanceOf[StateSpecImpl[K, V, StateType, MappedType]]
  )

2.  MapWithStateDStream源码如下:

/**
 * :: Experimental ::
 * DStream representing the stream of data generated by `mapWithState` operation on a
 * [[org.apache.spark.streaming.dstream.PairDStreamFunctions pair DStream]].
 * Additionally, it also gives access to the stream of state snapshots, that is, the state data of
 * all keys after a batch has updated them.
 *
 * @tparam KeyType Class of the key
 * @tparam ValueType Class of the value
 * @tparam StateType Class of the state data
 * @tparam MappedType Class of the mapped data
 */
@Experimental
sealed abstract class MapWithStateDStream[KeyType, ValueType, StateType, MappedType: ClassTag](
    ssc: StreamingContext) extends DStream[MappedType](ssc) 

  /** Return a pair DStream where each RDD is the snapshot of the state of all the keys. */
  def stateSnapshots(): DStream[(KeyType, StateType)]


/** Internal implementation of the [[MapWithStateDStream]] */
private[streaming] class MapWithStateDStreamImpl[
    KeyType: ClassTag, ValueType: ClassTag, StateType: ClassTag, MappedType: ClassTag](
    dataStream: DStream[(KeyType, ValueType)],
    spec: StateSpecImpl[KeyType, ValueType, StateType, MappedType])
  extends MapWithStateDStream[KeyType, ValueType, StateType, MappedType](dataStream.context) 

  private val internalStream =
    new InternalMapWithStateDStream[KeyType, ValueType, StateType, MappedType](dataStream, spec)

  override def slideDuration: Duration = internalStream.slideDuration

  override def dependencies: List[DStream[_]] = List(internalStream)
//计算的时候是通过InternalMapWithStateDStream来实现的。
  override def compute(validTime: Time): Option[RDD[MappedType]] = 
    internalStream.getOrCompute(validTime).map  _.flatMap[MappedType]  _.mappedData  
  

3.  更新历史数据。

/**
 * A DStream that allows per-key state to be maintains, and arbitrary records to be generated
 * based on updates to the state. This is the main DStream that implements the `mapWithState`
 * operation on DStreams.
 *
 * @param parent (key, value) stream that is the source
 * @param spec Specifications of the mapWithState operation
 * @tparam K   Key type
 * @tparam V   Value type
 * @tparam S   Type of the state maintained
 * @tparam E   Type of the mapped data
 */
private[streaming]
class InternalMapWithStateDStream[K: ClassTag, V: ClassTag, S: ClassTag, E: ClassTag](
    parent: DStream[(K, V)], spec: StateSpecImpl[K, V, S, E])
  extends DStream[MapWithStateRDDRecord[K, S, E]](parent.context) 
//不断的更新内存数据结构。
  persist(StorageLevel.MEMORY_ONLY)

4.  MapWithStateDStream.Compute

/** Method that generates a RDD for the given time */
  override def compute(validTime: Time): Option[RDD[MapWithStateRDDRecord[K, S, E]]] = 
    // Get the previous state or create a new empty state RDD
    val prevStateRDD = getOrCompute(validTime - slideDuration) match 
      case Some(rdd) =>
        if (rdd.partitioner != Some(partitioner)) 
          // If the RDD is not partitioned the right way, let us repartition it using the
          // partition index as the key. This is to ensure that state RDD is always partitioned
          // before creating another state RDD using it
          MapWithStateRDD.createFromRDD[K, V, S, E](
            rdd.flatMap  _.stateMap.getAll() , partitioner, validTime)
         else 
          rdd
        
      case None =>
        MapWithStateRDD.createFromPairRDD[K, V, S, E](
          spec.getInitialStateRDD().getOrElse(new EmptyRDD[(K, S)](ssc.sparkContext)),
          partitioner,
          validTime
        )
    

//基于时间窗口创建RDD
    // Compute the new state RDD with previous state RDD and partitioned data RDD
    // Even if there is no data RDD, use an empty one to create a new state RDD
    val dataRDD = parent.getOrCompute(validTime).getOrElse 
      context.sparkContext.emptyRDD[(K, V)]
    
    val partitionedDataRDD = dataRDD.partitionBy(partitioner)
    val timeoutThresholdTime = spec.getTimeoutInterval().map  interval =>
      (validTime - interval).milliseconds
    
    Some(new MapWithStateRDD(
      prevStateRDD, partitionedDataRDD, mappingFunction, validTime, timeoutThresholdTime))
  

5.  MapWithStateRDD: 是一个RDD，他本身包含了对mapWithState操作的数据，以及对数据怎么操作，MapWithStateRDDRecord代表了每个RDD的partition。

/**
 * RDD storing the keyed states of `mapWithState` operation and corresponding mapped data.
 * Each partition of this RDD has a single record of type [[MapWithStateRDDRecord]]. This contains a
 * [[StateMap]] (containing the keyed-states) and the sequence of records returned by the mapping
 * function of  `mapWithState`.
 * @param prevStateRDD The previous MapWithStateRDD on whose StateMap data `this` RDD
  *                    will be created
 * @param partitionedDataRDD The partitioned data RDD which is used update the previous StateMaps
 *                           in the `prevStateRDD` to create `this` RDD
 * @param mappingFunction  The function that will be used to update state and return new data
 * @param batchTime        The time of the batch to which this RDD belongs to. Use to update
 * @param timeoutThresholdTime The time to indicate which keys are timeout
 */
private[streaming] class MapWithStateRDD[K: ClassTag, V: ClassTag, S: ClassTag, E: ClassTag](
    private var prevStateRDD: RDD[MapWithStateRDDRecord[K, S, E]],
    private var partitionedDataRDD: RDD[(K, V)],
    mappingFunction: (Time, K, Option[V], State[S]) => Option[E],
    batchTime: Time,
    timeoutThresholdTime: Option[Long]
  ) extends RDD[MapWithStateRDDRecord[K, S, E]](
    partitionedDataRDD.sparkContext,
    List(
      new OneToOneDependency[MapWithStateRDDRecord[K, S, E]](prevStateRDD),
      new OneToOneDependency(partitionedDataRDD))
  ) 

  @volatile private var doFullScan = false

  require(prevStateRDD.partitioner.nonEmpty)
  require(partitionedDataRDD.partitioner == prevStateRDD.partitioner)

  override val partitioner = prevStateRDD.partitioner

  override def checkpoint(): Unit = 
    super.checkpoint()
    doFullScan = true
  

  override def compute(
      partition: Partition, context: TaskContext): Iterator[MapWithStateRDDRecord[K, S, E]] = 

    val stateRDDPartition = partition.asInstanceOf[MapWithStateRDDPartition]
    val prevStateRDDIterator = prevStateRDD.iterator(
      stateRDDPartition.previousSessionRDDPartition, context)
    val dataIterator = partitionedDataRDD.iterator(
      stateRDDPartition.partitionedDataRDDPartition, context)

    val prevRecord = if (prevStateRDDIterator.hasNext) Some(prevStateRDDIterator.next()) else None
    val newRecord = MapWithStateRDDRecord.updateRecordWithData(
      prevRecord,
      dataIterator,
      mappingFunction,
      batchTime,
      timeoutThresholdTime,
      removeTimedoutData = doFullScan // remove timedout data only when full scan is enabled
    )
    Iterator(newRecord)
  

6.  updateRecordWithData: RDD本身不可变的，但是可以处理变化的数据。

def updateRecordWithData[K: ClassTag, V: ClassTag, S: ClassTag, E: ClassTag](
    prevRecord: Option[MapWithStateRDDRecord[K, S, E]],
    dataIterator: Iterator[(K, V)],
    mappingFunction: (Time, K, Option[V], State[S]) => Option[E],
    batchTime: Time,
    timeoutThresholdTime: Option[Long],
    removeTimedoutData: Boolean
  ): MapWithStateRDDRecord[K, S, E] = 
    // Create a new state map by cloning the previous one (if it exists) or by creating an empty one
    val newStateMap = prevRecord.map  _.stateMap.copy() . getOrElse  new EmptyStateMap[K, S]() 

    val mappedData = new ArrayBuffer[E]
    val wrappedState = new StateImpl[S]()

    // Call the mapping function on each record in the data iterator, and accordingly
    // update the states touched, and collect the data returned by the mapping function
    dataIterator.foreach  case (key, value) =>
      wrappedState.wrap(newStateMap.get(key))
      val returned = mappingFunction(batchTime, key, Some(value), wrappedState)
      if (wrappedState.isRemoved) 
        newStateMap.remove(key)
       else if (wrappedState.isUpdated || timeoutThresholdTime.isDefined) 
//遍历当前所有batchTime的所有数据，然后使用自定义的函数对当前的batch数据进行计算，更新newStateMap数据结构。
// newStateMap是保存历史数据
        newStateMap.put(key, wrappedState.get(), batchTime.milliseconds)
      
      mappedData ++= returned
    

    // Get the timed out state records, call the mapping function on each and collect the
    // data returned
    if (removeTimedoutData && timeoutThresholdTime.isDefined) 
      newStateMap.getByTime(timeoutThresholdTime.get).foreach  case (key, state, _) =>
        wrappedState.wrapTimingOutState(state)
        val returned = mappingFunction(batchTime, key, None, wrappedState)
        mappedData ++= returned
        newStateMap.remove(key)
      
    
// MapWithStateRDDRecord所代表的partition，从RDD的角度来说，没有变。但是内部变了。只是内部数据发送变化了。
    MapWithStateRDDRecord(newStateMap, mappedData)
  

MapWithState实现如下：

总结：