How to stop an Actor reloading on exception - scala

In a scenario where an exception is thrown in an Actor receive I want to prevent this actor from being reloaded. I understood that the correct way to do this is by overriding supervisorStrategy but this does not work as shown in the example below:
class MyActor extends Actor {
println("Created new actor")
def receive = {
case msg =>
println("Received message: " + msg)
throw new Exception()
}
override val supervisorStrategy = OneForOneStrategy() {
case _: Exception => Stop
}
}
val system = ActorSystem("Test")
val actor = system.actorOf(Props(new MyActor()))
actor ! "Hello"
When I run this code "Created new actor" is output twice showing that the Actor is reloaded again after the exception.
What is the correct way to prevent the Actor being reloaded?

When an actor overrides the default supervisor strategy, that strategy applies to that actor's children. Your actor is using the default supervisor strategy, which restarts actors when they throw an exception. Define a parent for your actor and override the supervisor strategy in that parent.
class MyParent extends Actor {
override val supervisorStrategy = OneForOneStrategy() {
case _: Exception => Stop
}
val child = context.actorOf(Props[MyActor])
def receive = {
case msg =>
println(s"Parent received the following message and is sending it to the child: $msg")
child ! msg
}
}
class MyActor extends Actor {
println("Created new actor")
def receive = {
case msg =>
println(s"Received message: $msg")
throw new Exception()
}
}
val system = ActorSystem("Test")
val actor = system.actorOf(Props[MyParent])
actor ! "Hello"
In the above example, a MyActor is created as a child of MyParent. When the latter receives the "Hello" message, it sends the same message to the child. The child is stopped when it throws the exception, and "Created new actor" is therefore printed only once.

Related

Create child actors based on the request pattern

I am trying to create a web socket server using Play Framework where response from server should be synchronous or asynchronous based on request.
The request will be processed in Parent actor .Based on the action in the request, child actor will be created and message will be passed to child actor for processing and response will be sent back to the controller.
There are predefined actions and sample request for some actions are as follows,
[,,]
["1234","Boot","{"system":"ABCD"}"]
["5678","Start","{"system":"EFGH", "currenTime":"1559548762638"}"]
#Singleton
class RequestController #Inject()(cc: ControllerComponents)(implicit system: ActorSystem, mat: Materializer) extends AbstractController(cc) {
def ws = WebSocket.accept[String, String] {req =>
ActorFlow.actorRef { out =>
ParentActor.props(out)
}
}
}
object ParentActor {
def props(out: ActorRef) = Props(new ParentActor(out))
}
class ParentActor(out : ActorRef) extends Actor {
override def receive: Receive = {
case msg: String =>
//String split opeartion to find the action.
//create child actor for the action and pass the message to the child actor
val action = msg.split(",")[2]
if("Boot".equals(action)){
val bootActor: ActorRef = actorSystem.actorOf(Props[BootActor])
childActor ! msg
}else if("Start".equals(action)){
val startActor: ActorRef = actorSystem.actorOf(Props[StartActor])
startActor ! msg
}
case msg: Response => out ! msg
}
}
case class Response(name:String, msg:String)
class BootActor extends Actor{
override def receive: Receive = {
case msg : String =>
sender() ! Response("ABC",msg)
}
}
class StartActor extends Actor{
override def receive: Receive = {
case msg : String =>
sender() ! Response("Efgh",msg)
}
}
Right now i am getting the action from the request and create a child actor for the action and pass the message to the child actor for processing.
But i am not sure is there any better way or design pattern to process the request and create a child actor instead of String operation?
First of all, there appears to be a typo in your code:
if ("Boot".equals(action)) {
val bootActor: ActorRef = actorSystem.actorOf(Props[BootActor])
childActor ! msg
} else if ("Start".equals(action)) {
val startActor: ActorRef = actorSystem.actorOf(Props[StartActor])
startActor ! msg
}
The message in the first conditional clause should be sent to bootActor instead of childActor, which is undefined in your code snippet.
Another issue is that you're using actorSystem.actorOf to create the child actors. This method creates "top-level" actors, which should be kept to a minimum. Actors created with actorSystem.actorOf are under the supervision of the guardian actor. What this means in relation to your code is that when ParentActor is stopped (i.e., when a WebSocket is closed, Play stops the actor used in ActorFlow, as documented here), the multiple instances of BootActor and StartActor will not be stopped, leaving you with a bunch of idle top-level actors. The remedy is to use context.actorOf to create instances of BootActor and StartActor: doing so makes these instances children of ParentActor.
Also, you should use the == operator instead of the equals method.
Here are the aforementioned changes:
if ("Boot" == action) {
val bootActor: ActorRef = context.actorOf(Props[BootActor])
bootActor ! msg
} else if ("Start" == action) {
val startActor: ActorRef = context.actorOf(Props[StartActor])
startActor ! msg
}
The above could be slightly simplified to the following:
val childActor =
if (action == "Boot") context.actorOf(Props[BootActor])
else context.actorOf(Props[StartActor])
childActor ! msg
To further simplify your code, don't create child actors, which in this case aren't necessary. Move all the logic of interacting with the out actor into a single actor.

How to restart a killed actor?

I have a supervisor and a child actor that looks as the following:
SupervisorActor.scala
import akka.actor.{Actor, ActorLogging, Props, Terminated}
object SupervisorActor {
def props: Props = Props(new SupervisorActor)
object KillFoo
object TerminateFoo
object RestartFoo
}
final class SupervisorActor extends Actor with ActorLogging {
import com.sweetsoft.SupervisorActor._
log.info(s"Start the actor ${self}")
private val foo = context.actorOf(FooActor.props, "FOOCHILD")
context.watch(foo)
override def receive: Receive = {
case KillFoo =>
context.stop(foo)
case Terminated(actor) =>
log.info(s"Kill actor $actor")
case TerminateFoo =>
log.info("Terminate FooActor")
foo ! new IllegalArgumentException
case RestartFoo =>
}
}
FooActor.scala
import akka.actor.SupervisorStrategy.{Escalate, Restart, Resume, Stop}
import akka.actor.{Actor, ActorLogging, OneForOneStrategy, Props}
import scala.concurrent.duration._
object FooActor {
def props: Props = Props(new FooActor)
}
final class FooActor extends Actor with ActorLogging {
log.info(s"Start the actor ${ self }")
override val supervisorStrategy: OneForOneStrategy =
OneForOneStrategy(maxNrOfRetries = 1, withinTimeRange = 1.minute) {
case _: ArithmeticException => Resume
case _: NullPointerException => Restart
case _: IllegalArgumentException =>
println("STOP exception raised.")
Stop
case _: Exception => Escalate
}
override def receive: Receive = {
case _ => log.info("I got killed.")
}
}
and Main.scala
import akka.actor.ActorSystem
import com.sweetsoft.SupervisorActor.{TerminateFoo}
import scala.concurrent.Future
import scala.io.StdIn
object Main extends App {
val system = ActorSystem("monarch")
implicit val dispatcher = system.dispatcher
try {
// Create top level supervisor
val supervisor = system.actorOf(SupervisorActor.props, "mupervisor")
// Exit the system after ENTER is pressed
Future {
Thread.sleep(2000)
supervisor ! TerminateFoo
}
StdIn.readLine()
} finally {
system.terminate()
}
}
After the FooActor got killed, I would like to restart it manually again, like I did:
private val foo = context.actorOf(FooActor.props, "FOOCHILD")
How to do it?
I am thinking about to create a function, that is going to create the FooActor and after it got killed, just call the function to start a new FooActor.
There are a few problems with the code. supervisorStrategy should be in the SupervisorActor as it is responsible for supervision, not the child actor itself.
foo ! new IllegalArgumentException won't cause the child actor to terminate as actors can accept any object as a message and Exception derived objects are not treated specially. It'll just print "I got killed." but otherwise ignore the message and keep running. In particular, it won't invoke supervisorStrategy
handler.
You can either:
send the pre-defined system PoisonPill message to the child actor
to gracefully stop it (i.e. after processing all pending messages)
send the pre-defined system Kill message to the child actor to stop
it immediately after processing the current message and ignore the
other queued messages if any.
forward TerminateFoo message to it
and throw an exception in its handler. In this case the child
object's destiny is decided by supervisorStrategy handler for
IllegalArgumentException exception type (i.e. Stop in your case).
.
override def receive: Receive = {
case TerminateFoo =>
log.info("Stopping FooActor by throwing an unhandled exception for supervisorStrategy to process it")
throw new IllegalArgumentException
case m => log.info(s"I received a message $m")
}
See https://doc.akka.io/docs/akka/current/actors.html#stopping-actors for details

Is it possible for the supervisor to pass the exception to the calling actor if the child actor fails even after retrying N times?

I have two actors Computer and Printer. Computer is the parent of Printer and has a one for one strategy defined for Printer.
I have listed the code below.
class Computer extends Actor with ActorLogging{
import Computer._
import Printer._
implicit val timeout: Timeout = 2 seconds
val printer: ActorRef = context.actorOf(Props[Printer], "printer-actor")
override def receive: Receive = {
case Print(text) => {
val printJob: Future[Any] = printer ? PrintJob(Random.nextInt, text)
printJob.mapTo[Page].map {
case Page(text) => {
log.info(s"Received page containing text ${text}")
context.system.shutdown()
}
}.onFailure {
case t: Throwable => sender ! akka.actor.Status.Failure(t)
}
}
}
override val supervisorStrategy =
OneForOneStrategy(maxNrOfRetries = 3, withinTimeRange = 1 minute) {
case e : Exception => {
log.info(s"caught exception of type ${e.getClass}")
SupervisorStrategy.Restart
}
}
}
class Printer extends Actor with ActorLogging{
import Printer._
override def receive: Receive = {
case PrintJob(id, text) => {
log.info(s"Received ${PrintJob(id, text)}")
if (Random.nextBoolean) sender ! Page(text)
else throw new NoPaperException(id)
}
}
override def preRestart(cause: Throwable, message: Option[Any]) = {
log.info(s"Restarting actor ${self} because of ${cause}. Queueing message ${message}")
postStop()
message.map(self forward _)
}
}
The Printer throws an exception based on the random generator. The code works fine, the supervisor restarts the and retries the child actor on failure just as instructed.
However the ask pattern val printJob: Future[Any] = printer ? PrintJob(Random.nextInt, text) fails with a AkkaTimeoutException in case all attempts to get the Printer actor work fails.
Is there a way to pass back the exact exception which caused the actor to fail ? In this case NoPapperException.
Cheers,
Utsav
to pass the exception back to the sender you need to sender ! Status.Failure(e) - where e is the exception
You can either do that directly from the actor, or if you want to do that from the supervisor you need to have a subclass of exception that would hold the sender ref with it so that the supervisor would be able to send the exception back

Akka - test supervision strategy

I have the following scenario: the parent supervisor actor creates a child for each message using a factory (function) passed in the constructor.
class supervisorActor(childActorMaker: ActorRefFactory => ActorRef)
extends Actor with ActorLogging{
def receive: Receive = {
case "testThis" =>
val childActor = childActorMaker(context)
childActor!"messageForChild"
}
override val supervisorStrategy =
OneForOneStrategy() {
case _ => Stop
}
}
class childActor extends Actor {
def receive:Receive = {
case _ => /** whatever **/
}
}
In the test I override the child Receive to force an exception. My expectation is the child actor to be stopped every time because of the Supervision Strategy i set.
"When the child Actor throws an exception the Supervisor Actor " should " " +
" stop it" in {
val childActorRef = TestActorRef(new childActor() {
override def receive = {
case msg: String => throw new IllegalArgumentException("kaboom")
}
})
watch(childActorRef)
val maker = (_: ActorRefFactory) => childActorRef
val supervisorActorRef = system.actorOf(Props(new supervisorActor(maker)))
supervisorActorRef!"testThis"
expectTerminated(childActorRef, 1.second)
}
I would expect the child actor to be stoped because of the supervisorStrategy Stop.
Instead I get this error:
java.lang.AssertionError: assertion failed: timeout (1 second) during expectMsg: Terminated
Any idea of why is this happening? Thank you
It seems that the childActorRef is not created with supervisorActorRef's context (I mean ActorContext of supervisorActor you created in the test code).
So childActor(childActorRef) is not a child of supervisorActor(supervisorActorRef) in your test code. That's why supervisor strategy of supervisorActor doesn't serve the purpose.

Akka supervisor actor do not handle exception when child actor throws an exception within onFailure of a future

I'm facing a problem with an Akka supervisor actor. When the child actor throws an exception within onFailure method of a future result, the supervisor does not handle the error (I want to restart the child in the case of a ConnectException).
I'm using Akka 2.3.7.
This is the supervisor actor:
class MobileUsersActor extends Actor with ActorLogging {
import Model.Implicits._
import Model.MobileNotifications
override val supervisorStrategy =
OneForOneStrategy(maxNrOfRetries = 3, withinTimeRange = 1 minute) {
case _: java.net.ConnectException => {
Logger.error("API connection error. Check your proxy configuration.")
Restart
}
}
def receive = {
case Start => findMobileUsers
}
private def findMobileUsers = {
val notis = MobileNotificationsRepository().find()
notis.map(invokePushSender)
}
private def invokePushSender(notis: List[MobileNotifications]) = {
notis.foreach { n =>
val pushSender = context.actorOf(PushSenderActor.props)
pushSender ! Send(n)
}
}
}
And this is the child actor:
class PushSenderActor extends Actor with ActorLogging {
def receive = {
case Send(noti) => {
val response = sendPushNotification(noti) onFailure {
case e: ConnectException => throw e
}
}
}
private def sendPushNotification(noti: MobileNotifications): Future[WSResponse] = {
val message = "Push notification message example"
Logger.info(s"Push Notification >> $message to users " + noti.users)
PushClient.sendNotification(message, noti.users)
}
}
I tried to notify sender with an akka.actor.Status.Failure(e) as is suggested here, but did not work, the exception keep unhandled by the supervisor.
As a workaround, I found this way to get it work:
class PushSenderActor extends Actor with ActorLogging {
def receive = {
case Send(noti) => {
val response = sendPushNotification(noti) onFailure {
case e: ConnectException => self ! APIConnectionError
}
}
case APIConnectionError => throw new ConnectException
}
private def sendPushNotification(noti: MobileNotifications): Future[WSResponse] = {
val message = "Push notification message example"
Logger.info(s"Push Notification >> $message to users " + noti.users)
PushClient.sendNotification(message, noti.users)
}
}
Is this an Akka bug or am I doing something wrong?
Thanks!
I think that the problem is that the exception thrown inside the Future doesn't belong to the same thread (potentially) as the one the Actor is running (someone more experienced can elaborate on this). So, the problem is that the exception thrown inside the Future body is "swallowed" and not propagated to the Actor. Since this is the case, the Actor doesn't fail and so there's no need to apply the supervision strategy. So, the first solution that comes to my mind is to wrap the exception inside the Future in some message, send it to yourself, and then throw it from inside the Actor context itself. This time, the Exception will be caught and the supervision strategy will be applied. Note, however, that unless you send the Send(noti) message again, you will not see the Exception happening since the Actor was restarted. All in all, the code would be like this:
class PushSenderActor extends Actor with ActorLogging {
case class SmthFailed(e: Exception)
def receive = {
case Send(noti) => {
val response = sendPushNotification(noti) onFailure {
case e: ConnectException => self ! SmthFailed(e) // send the exception to yourself
}
}
case SmthFailed(e) =>
throw e // this one will be caught by the supervisor
}
private def sendPushNotification(noti: MobileNotifications): Future[WSResponse] = {
val message = "Push notification message example"
Logger.info(s"Push Notification >> $message to users " + noti.users)
PushClient.sendNotification(message, noti.users)
}
}
Hope it helped.