RestartSource.onFailuresWithBackoff

Wrap the given Source with a Source that will restart it when it fails using an exponential backoff. Notice that this Source will not restart on completion of the wrapped flow.

Error handling

Signature

RestartSource.onFailuresWithBackoff

Description

Wraps the given Source with a Source that will restart it when it fails using exponential backoff. The backoff resets back to minBackoff if there hasn’t been a restart within maxRestartsWithin (which defaults to minBackoff).

This Source will not emit a failure as long as maxRestarts is not reached. The failure of the wrapped Source is handled by restarting it. However, the wrapped Source can be cancelled by cancelling this Source. When that happens, the wrapped Source, if currently running will, be cancelled and not restarted. This can be triggered by the downstream cancelling, or externally by introducing a KillSwitch right after this Source in the graph.

This uses the same exponential backoff algorithm as BackoffOpts.

See also:

Examples

This shows that a Source is not restarted if it completes, only if it fails. Tick is only printed three times as the take(3) means the inner source completes successfully after emitting the first 3 elements.

Scala
Java
sourceval finiteSource = Source.tick(1.second, 1.second, "tick").take(3)
val forever = RestartSource.onFailuresWithBackoff(RestartSettings(1.second, 10.seconds, 0.1))(() => finiteSource)
forever.runWith(Sink.foreach(println))
// prints
// tick
// tick
// tick
sourceSource<String, Cancellable> finiteSource =
    Source.tick(Duration.ofSeconds(1), Duration.ofSeconds(1), "tick").take(3);
Source<String, NotUsed> forever =
    RestartSource.onFailuresWithBackoff(
        RestartSettings.create(Duration.ofSeconds(1), Duration.ofSeconds(10), 0.1),
        () -> finiteSource);
forever.runWith(Sink.foreach(System.out::println), system);
// prints
// tick
// tick
// tick

If the inner source instead fails, it will be restarted with an increasing backoff. The source emits 1, 2, 3, and then throws an exception. The first time the exception is thrown the source is restarted after 1s, then 2s etc, until the maxBackoff of 10s.

Scala
Java
source// could throw if for example it used a database connection to get rows
val flakySource: Source[() => Int, NotUsed] =
  Source(List(() => 1, () => 2, () => 3, () => throw CantConnectToDatabase("darn")))
val forever =
  RestartSource.onFailuresWithBackoff(
    RestartSettings(minBackoff = 1.second, maxBackoff = 10.seconds, randomFactor = 0.1))(() => flakySource)
forever.runWith(Sink.foreach(nr => system.log.info("{}", nr())))
// logs
// [INFO] [12/10/2019 13:51:58.300] [default-pekko.test.stream-dispatcher-7] [pekko.actor.ActorSystemImpl(default)] 1
// [INFO] [12/10/2019 13:51:58.301] [default-pekko.test.stream-dispatcher-7] [pekko.actor.ActorSystemImpl(default)] 2
// [INFO] [12/10/2019 13:51:58.302] [default-pekko.test.stream-dispatcher-7] [pekko.actor.ActorSystemImpl(default)] 3
// [WARN] [12/10/2019 13:51:58.310] [default-pekko.test.stream-dispatcher-7] [RestartWithBackoffSource(pekko://default)] Restarting graph due to failure. stack_trace:  (docs.stream.operators.source.Restart$CantConnectToDatabase: darn)
// --> 1 second gap
// [INFO] [12/10/2019 13:51:59.379] [default-pekko.test.stream-dispatcher-8] [pekko.actor.ActorSystemImpl(default)] 1
// [INFO] [12/10/2019 13:51:59.382] [default-pekko.test.stream-dispatcher-8] [pekko.actor.ActorSystemImpl(default)] 2
// [INFO] [12/10/2019 13:51:59.383] [default-pekko.test.stream-dispatcher-8] [pekko.actor.ActorSystemImpl(default)] 3
// [WARN] [12/10/2019 13:51:59.386] [default-pekko.test.stream-dispatcher-8] [RestartWithBackoffSource(pekko://default)] Restarting graph due to failure. stack_trace:  (docs.stream.operators.source.Restart$CantConnectToDatabase: darn)
// --> 2 second gap
// [INFO] [12/10/2019 13:52:01.594] [default-pekko.test.stream-dispatcher-8] [pekko.actor.ActorSystemImpl(default)] 1
// [INFO] [12/10/2019 13:52:01.595] [default-pekko.test.stream-dispatcher-8] [pekko.actor.ActorSystemImpl(default)] 2
// [INFO] [12/10/2019 13:52:01.595] [default-pekko.test.stream-dispatcher-8] [pekko.actor.ActorSystemImpl(default)] 3
// [WARN] [12/10/2019 13:52:01.596] [default-pekko.test.stream-dispatcher-8] [RestartWithBackoffSource(pekko://default)] Restarting graph due to failure. stack_trace:  (docs.stream.operators.source.Restart$CantConnectToDatabase: darn)
source// could throw if for example it used a database connection to get rows
Source<Creator<Integer>, NotUsed> flakySource =
    Source.from(
        Arrays.<Creator<Integer>>asList(
            () -> 1,
            () -> 2,
            () -> 3,
            () -> {
              throw new RuntimeException("darn");
            }));
Source<Creator<Integer>, NotUsed> forever =
    RestartSource.onFailuresWithBackoff(
        RestartSettings.create(Duration.ofSeconds(1), Duration.ofSeconds(10), 0.1),
        () -> flakySource);
forever.runWith(
    Sink.foreach((Creator<Integer> nr) -> system.log().info("{}", nr.create())), system);
// logs
// [INFO] [12/10/2019 13:51:58.300] [default-pekko.test.stream-dispatcher-7]
// [pekko.actor.ActorSystemImpl(default)] 1
// [INFO] [12/10/2019 13:51:58.301] [default-pekko.test.stream-dispatcher-7]
// [pekko.actor.ActorSystemImpl(default)] 2
// [INFO] [12/10/2019 13:51:58.302] [default-pekko.test.stream-dispatcher-7]
// [pekko.actor.ActorSystemImpl(default)] 3
// [WARN] [12/10/2019 13:51:58.310] [default-pekko.test.stream-dispatcher-7]
// [RestartWithBackoffSource(pekko://default)] Restarting graph due to failure. stack_trace:
// (RuntimeException: darn)
// --> 1 second gap
// [INFO] [12/10/2019 13:51:59.379] [default-pekko.test.stream-dispatcher-8]
// [pekko.actor.ActorSystemImpl(default)] 1
// [INFO] [12/10/2019 13:51:59.382] [default-pekko.test.stream-dispatcher-8]
// [pekko.actor.ActorSystemImpl(default)] 2
// [INFO] [12/10/2019 13:51:59.383] [default-pekko.test.stream-dispatcher-8]
// [pekko.actor.ActorSystemImpl(default)] 3
// [WARN] [12/10/2019 13:51:59.386] [default-pekko.test.stream-dispatcher-8]
// [RestartWithBackoffSource(pekko://default)] Restarting graph due to failure. stack_trace:
// (RuntimeException: darn)
// --> 2 second gap
// [INFO] [12/10/2019 13:52:01.594] [default-pekko.test.stream-dispatcher-8]
// [pekko.actor.ActorSystemImpl(default)] 1
// [INFO] [12/10/2019 13:52:01.595] [default-pekko.test.stream-dispatcher-8]
// [pekko.actor.ActorSystemImpl(default)] 2
// [INFO] [12/10/2019 13:52:01.595] [default-pekko.test.stream-dispatcher-8]
// [pekko.actor.ActorSystemImpl(default)] 3
// [WARN] [12/10/2019 13:52:01.596] [default-pekko.test.stream-dispatcher-8]
// [RestartWithBackoffSource(pekko://default)] Restarting graph due to failure. stack_trace:
// (RuntimeException: darn)

Finally, to be able to stop the restarting, a kill switch can be used. The kill switch is inserted right after the restart source. The inner source is the same as above so emits 3 elements and then fails. A killswitch is used to be able to stop the source being restarted:

Scala
Java
sourceval flakySource: Source[() => Int, NotUsed] =
  Source(List(() => 1, () => 2, () => 3, () => throw CantConnectToDatabase("darn")))
val stopRestarting: UniqueKillSwitch =
  RestartSource
    .onFailuresWithBackoff(RestartSettings(1.second, 10.seconds, 0.1))(() => flakySource)
    .viaMat(KillSwitches.single)(Keep.right)
    .toMat(Sink.foreach(nr => println(s"Nr ${nr()}")))(Keep.left)
    .run()
// ... from some where else
// stop the source from restarting
stopRestarting.shutdown()
sourceSource<Creator<Integer>, NotUsed> flakySource =
    Source.from(
        Arrays.<Creator<Integer>>asList(
            () -> 1,
            () -> 2,
            () -> 3,
            () -> {
              throw new RuntimeException("darn");
            }));
UniqueKillSwitch stopRestarting =
    RestartSource.onFailuresWithBackoff(
            RestartSettings.create(Duration.ofSeconds(1), Duration.ofSeconds(10), 0.1),
            () -> flakySource)
        .viaMat(KillSwitches.single(), Keep.right())
        .toMat(Sink.foreach(nr -> System.out.println("nr " + nr.create())), Keep.left())
        .run(system);
// ... from some where else
// stop the source from restarting
stopRestarting.shutdown();

Reactive Streams semantics

emits when the wrapped source emits

backpressures during backoff and when downstream backpressures

completes when the wrapped source completes or maxRestarts are reached within the given time limit

cancels when downstream cancels

1.1.3