Question

在Akka集群应用程序的上下文中，我遇到了一个关于Akka期望的一个属性的问题：每个（cas）类和使用的每个消息都必须是可序列化的。我有以下上下文：我想从redis集群中使用数据，为此，我决定采用集群感知路由器池来添加节点以拥有更多的工作者。工作人员从redis读取数据并将一些元数据存储在mongodb中。在第一个版本中，我这样做了：

object MasterWorkers {

  def props
  (  awsBucket : String,
     gapMinValueMicroSec : Long,
     persistentCache: RedisCache,
     mongoURI : String,
     mongoDBName : String,
     mongoCollectioName : String
  ) : Props =
    Props(MasterWorkers(awsBucket, gapMinValueMicroSec, persistentCache, mongoURI, mongoDBName, mongoCollectioName))

  case class JobRemove(deviceId: DeviceId, from : Timestamp, to : Timestamp)
}

case class MasterWorkers
(
  awsBucket : String,
  gapMinValueMicroSec : Long,
  persistentCache: RedisCache,
  mongoURI : String,
  mongoDBName : String,
  mongoCollectioName : String
) extends Actor with ActorLogging {

  val workerRouter =
    context.actorOf(FromConfig.props(Props(classOf[Worker],awsBucket,gapMinValueMicroSec, self, persistentCache, mongoURI, mongoDBName, mongoCollectioName)),
    name = "workerRouter")

工人阶级：

object Worker {

  def props
  (
    awsBucket : String,
    gapMinValueMicroSec : Long,
    replyTo : ActorRef,
    persistentCache: RedisCache,
    mongoURI : String,
    mongoDBName : String,
    mongoCollectioName : String
  ) : Props =
    Props(Worker(awsBucket, gapMinValueMicroSec, replyTo, persistentCache, mongoURI, mongoDBName, mongoCollectioName))

  case class JobDumpFailed(deviceId : DeviceId, from: Timestamp, to: Timestamp)
  case class JobDumpSuccess(deviceId : DeviceId, from: Timestamp, to: Timestamp)

  case class JobRemoveFailed(deviceId : DeviceId, from: Timestamp, to: Timestamp)
}

case class Worker
(
  awsBucket : String,
  gapMinValueMicroSec : Long,
  replyTo : ActorRef,
  persistentCache: RedisCache,
  mongoURI : String,
  mongoDBName : String,
  mongoCollectioName : String
) extends Actor with ActorLogging {

但是当我启动两个节点时，这会引发以下异常：

[info] akka.remote.MessageSerializer$SerializationException: Failed to serialize remote message [class akka.remote.DaemonMsgCreate] using serializer [class akka.remote.serialization.DaemonMsgCreateSerializer].
[info] at akka.remote.MessageSerializer$.serialize(MessageSerializer.scala:61)
[info] at akka.remote.EndpointWriter$$anonfun$serializeMessage$1.apply(Endpoint.scala:895)
[info] at akka.remote.EndpointWriter$$anonfun$serializeMessage$1.apply(Endpoint.scala:895)
[info] at scala.util.DynamicVariable.withValue(DynamicVariable.scala:58)
[info] at akka.remote.EndpointWriter.serializeMessage(Endpoint.scala:894)
[info] at akka.remote.EndpointWriter.writeSend(Endpoint.scala:786)
[info] at akka.remote.EndpointWriter$$anonfun$4.applyOrElse(Endpoint.scala:761)
[info] at akka.actor.Actor$class.aroundReceive(Actor.scala:497)
[info] at akka.remote.EndpointActor.aroundReceive(Endpoint.scala:452)
[info] at akka.actor.ActorCell.receiveMessage(ActorCell.scala:526)
[info] at akka.actor.ActorCell.invoke(ActorCell.scala:495)
[info] at akka.dispatch.Mailbox.processMailbox(Mailbox.scala:257)
[info] at akka.dispatch.Mailbox.run(Mailbox.scala:224)
[info] at akka.dispatch.Mailbox.exec(Mailbox.scala:234)
[info] at scala.concurrent.forkjoin.ForkJoinTask.doExec(ForkJoinTask.java:260)
[info] at scala.concurrent.forkjoin.ForkJoinPool$WorkQueue.runTask(ForkJoinPool.java:1339)
[info] at scala.concurrent.forkjoin.ForkJoinPool.runWorker(ForkJoinPool.java:1979)
[info] at scala.concurrent.forkjoin.ForkJoinWorkerThread.run(ForkJoinWorkerThread.java:107)
[info] Caused by: java.io.NotSerializableException: akka.actor.ActorSystemImpl

redis缓存是一个简单的案例类，其伴随对象实现了如下界面：

object RedisCache { // some static functions }

case class RedisCache
(
  master : RedisServer,
  slaves : Seq[RedisServer]
)(implicit actorSystem : ActorSystem)
  extends PersistentCache[DeviceKey, BCPPackets] with LazyLogging {
// some code here
}

然后为了解决这个问题，我在工作中移动了redisCache，而我没有将它交给主节点：

case class Worker
(
  awsBucket : String,
  gapMinValueMicroSec : Long,
  replyTo : ActorRef,
  mongoURI : String,
  mongoDBName : String,
  mongoCollectioName : String
) extends Actor with ActorLogging {

// redis cache here now 
val redisCache = ...

但是有了这样的设计，每个routee都会创建一个新的redis缓存实例，而这不是预期的行为。我想要的是拥有我的redis缓存的一个实例，然后与我的所有路由共享它，但在集群应用程序的上下文中，似乎不可能，所以我不知道这是设计失败还是缺少一些经验与Akka。如果有人遇到类似问题，我会高兴地提出建议！

Akka Cluster感知路由器 - 将redis实例共享给所有routee

0 个答案: