AMQP

The AMQP connector provides Akka Stream sources and sinks to connect to AMQP 0.9.1 servers (RabbitMQ, OpenAMQ, etc.).

AMQP 1.0 is currently not supported (Qpid, ActiveMQ, Solace, etc.).

Artifacts

The Akka dependencies are available from Akka’s library repository. To access them there, you need to configure the URL for this repository.

sbt

resolvers += "Akka library repository".at("https://repo.akka.io/maven")

Maven

<project>
  ...
  <repositories>
    <repository>
      <id>akka-repository</id>
      <name>Akka library repository</name>
      <url>https://repo.akka.io/maven</url>
    </repository>
  </repositories>
</project>

Gradle

repositories {
    mavenCentral()
    maven {
        url "https://repo.akka.io/maven"
    }
}

Additionally, add the dependencies as below.

sbt

val AkkaVersion = "2.10.0"
libraryDependencies ++= Seq(
  "com.lightbend.akka" %% "akka-stream-alpakka-amqp" % "9.0.0",
  "com.typesafe.akka" %% "akka-stream" % AkkaVersion
)

Maven

<properties>
  <akka.version>2.10.0</akka.version>
  <scala.binary.version>2.13</scala.binary.version>
</properties>
<dependencies>
  <dependency>
    <groupId>com.lightbend.akka</groupId>
    <artifactId>akka-stream-alpakka-amqp_${scala.binary.version}</artifactId>
    <version>9.0.0</version>
  </dependency>
  <dependency>
    <groupId>com.typesafe.akka</groupId>
    <artifactId>akka-stream_${scala.binary.version}</artifactId>
    <version>${akka.version}</version>
  </dependency>
</dependencies>

Gradle

def versions = [
  AkkaVersion: "2.10.0",
  ScalaBinary: "2.13"
]
dependencies {
  implementation "com.lightbend.akka:akka-stream-alpakka-amqp_${versions.ScalaBinary}:9.0.0"
  implementation "com.typesafe.akka:akka-stream_${versions.ScalaBinary}:${versions.AkkaVersion}"
}

The table below shows direct dependencies of this module and the second tab shows all libraries it depends on transitively.

Direct dependencies

Organization	Artifact	Version
com.rabbitmq	amqp-client	5.21.0
com.typesafe.akka	akka-stream_2.13	2.10.0
org.scala-lang	scala-library	2.13.12

Dependency tree

com.rabbitmq    amqp-client    5.21.0    AL 2.0
    org.slf4j    slf4j-api    1.7.36
com.typesafe.akka    akka-stream_2.13    2.10.0    BUSL-1.1
    com.typesafe.akka    akka-actor_2.13    2.10.0    BUSL-1.1
        com.typesafe    config    1.4.3    Apache-2.0
        org.scala-lang    scala-library    2.13.12    Apache-2.0
    com.typesafe.akka    akka-protobuf-v3_2.13    2.10.0    BUSL-1.1
    org.reactivestreams    reactive-streams    1.0.4    MIT-0
    org.scala-lang    scala-library    2.13.12    Apache-2.0
org.scala-lang    scala-library    2.13.12    Apache-2.0

Connecting to server

All the AMQP connectors are configured using a AmqpConnectionProviderAmqpConnectionProvider and a list of DeclarationDeclaration

There are several types of AmqpConnectionProviderAmqpConnectionProvider:

• AmqpLocalConnectionProviderAmqpLocalConnectionProvider which connects to the default localhost. It creates a new connection for each stage.
• AmqpUriConnectionProviderAmqpUriConnectionProvider which connects to the given AMQP URI. It creates a new connection for each stage.
• AmqpDetailsConnectionProviderAmqpDetailsConnectionProvider which supports more fine-grained configuration. It creates a new connection for each stage.
• AmqpConnectionFactoryConnectionProviderAmqpConnectionFactoryConnectionProvider which takes a raw ConnectionFactory. It creates a new connection for each stage.
• AmqpCachedConnectionProviderAmqpCachedConnectionProvider which receive any other provider as parameter and caches the connection it provides to be used in all stages. By default it closes the connection whenever the last stage using the provider stops. Optionally, it takes automaticRelease boolean parameter so the connection is not automatically release and the user have to release it explicitly.

Sending messages

First define a queue name and the declaration of the queue that the messages will be sent to.

Scala

copysourceval queueName = "amqp-conn-it-spec-simple-queue-" + System.currentTimeMillis()
val queueDeclaration = QueueDeclaration(queueName)

Java

copysourcefinal String queueName = "amqp-conn-it-test-simple-queue-" + System.currentTimeMillis();
final QueueDeclaration queueDeclaration = QueueDeclaration.create(queueName);

Here we used QueueDeclarationQueueDeclaration configuration class to create a queue declaration.

With flow

Similarly as with Sink, the first step is to create Flow which accepts WriteMessageWriteMessages and forwards it’s content to the AMQP server. Flow emits WriteResultWriteResults informing about publication result (see below for summary of delivery guarantees for different Flow variants).

AmqpFlowAmqpFlow is a collection of factory methods that facilitates creation of flows. Here we created a simple sink, which means that we are able to pass ByteStrings to the sink instead of wrapping data into WriteMessageWriteMessages.

Last step is to materialize and run the flow we have created.

Scala

copysourceval settings = AmqpWriteSettings(connectionProvider)
  .withRoutingKey(queueName)
  .withDeclaration(queueDeclaration)
  .withBufferSize(10)
  .withConfirmationTimeout(200.millis)

val amqpFlow: Flow[WriteMessage, WriteResult, Future[Done]] =
  AmqpFlow.withConfirm(settings)

val input = Vector("one", "two", "three", "four", "five")
val result: Future[Seq[WriteResult]] =
  Source(input)
    .map(message => WriteMessage(ByteString(message)))
    .via(amqpFlow)
    .runWith(Sink.seq)

Java

copysourcefinal AmqpWriteSettings settings =
    AmqpWriteSettings.create(connectionProvider)
        .withRoutingKey(queueName)
        .withDeclaration(queueDeclaration)
        .withBufferSize(10)
        .withConfirmationTimeout(Duration.ofMillis(200));

final Flow<WriteMessage, WriteResult, CompletionStage<Done>> amqpFlow =
    AmqpFlow.createWithConfirm(settings);

final List<String> input = Arrays.asList("one", "two", "three", "four", "five");

final List<WriteResult> result =
    Source.from(input)
        .map(message -> WriteMessage.create(ByteString.fromString(message)))
        .via(amqpFlow)
        .runWith(Sink.seq(), system)
        .toCompletableFuture()
        .get();

Various variants of AMQP flow offer different delivery and ordering guarantees:

For FlowWithContextFlowWithContext counterparts of above flows see AmqpFlowWithContextAmqpFlowWithContext.

With sink

Create a sink, that accepts and forwards ByteStringByteStrings to the AMQP server.

AmqpSinkAmqpSink is a collection of factory methods that facilitates creation of sinks. Here we created a simple sink, which means that we are able to pass ByteStrings to the sink instead of wrapping data into WriteMessageWriteMessages.

Last step is to materialize and run the sink we have created.

Scala

copysourceval amqpSink: Sink[ByteString, Future[Done]] =
  AmqpSink.simple(
    AmqpWriteSettings(connectionProvider)
      .withRoutingKey(queueName)
      .withDeclaration(queueDeclaration)
  )

val input = Vector("one", "two", "three", "four", "five")
val writing: Future[Done] =
  Source(input)
    .map(s => ByteString(s))
    .runWith(amqpSink)

Java

copysourcefinal Sink<ByteString, CompletionStage<Done>> amqpSink =
    AmqpSink.createSimple(
        AmqpWriteSettings.create(connectionProvider)
            .withRoutingKey(queueName)
            .withDeclaration(queueDeclaration));

final List<String> input = Arrays.asList("one", "two", "three", "four", "five");
CompletionStage<Done> writing =
    Source.from(input).map(ByteString::fromString).runWith(amqpSink, system);

Receiving messages

Create a source using the same queue declaration as before.

The bufferSize parameter controls the maximum number of messages to prefetch from the AMQP server.

Run the source and take the same amount of messages as we previously sent to it.

Scala

copysourceval amqpSource: Source[ReadResult, NotUsed] =
  AmqpSource.atMostOnceSource(
    NamedQueueSourceSettings(connectionProvider, queueName)
      .withDeclaration(queueDeclaration)
      .withAckRequired(false),
    bufferSize = 10
  )

val result: Future[immutable.Seq[ReadResult]] =
  amqpSource
    .take(input.size)
    .runWith(Sink.seq)

Java

copysourcefinal Integer bufferSize = 10;
final Source<ReadResult, NotUsed> amqpSource =
    AmqpSource.atMostOnceSource(
        NamedQueueSourceSettings.create(connectionProvider, queueName)
            .withDeclaration(queueDeclaration)
            .withAckRequired(false),
        bufferSize);

final CompletionStage<List<ReadResult>> result =
    amqpSource.take(input.size()).runWith(Sink.seq(), system);

This is how you send and receive message from AMQP server using this connector.

Using Pub/Sub

Instead of sending messages directly to queues, it is possible to send messages to an exchange and then provide instructions to the AMQP server what to do with incoming messages. We are going to use the fanout type of the exchange, which enables message broadcasting to multiple consumers. We are going to do that by using an exchange declaration for the sink and all of the sources.

Scala

copysourceval exchangeName = "amqp-conn-it-spec-pub-sub-" + System.currentTimeMillis()
val exchangeDeclaration = ExchangeDeclaration(exchangeName, "fanout")

Java

copysourcefinal String exchangeName = "amqp-conn-it-test-pub-sub-" + System.currentTimeMillis();
final ExchangeDeclaration exchangeDeclaration =
    ExchangeDeclaration.create(exchangeName, "fanout");

The sink for the exchange is created in a very similar way.

Scala

copysourceval amqpSink = AmqpSink.simple(
  AmqpWriteSettings(connectionProvider)
    .withExchange(exchangeName)
    .withDeclaration(exchangeDeclaration)
)

Java

copysourcefinal Sink<ByteString, CompletionStage<Done>> amqpSink =
    AmqpSink.createSimple(
        AmqpWriteSettings.create(connectionProvider)
            .withExchange(exchangeName)
            .withDeclaration(exchangeDeclaration));

For the source, we are going to create multiple sources and merge them using Akka Streams operators.

Scala

copysourceval fanoutSize = 4

val mergedSources = (0 until fanoutSize).foldLeft(Source.empty[(Int, String)]) {
  case (source, fanoutBranch) =>
    source.merge(
      AmqpSource
        .atMostOnceSource(
          TemporaryQueueSourceSettings(
            connectionProvider,
            exchangeName
          ).withDeclaration(exchangeDeclaration),
          bufferSize = 1
        )
        .map(msg => (fanoutBranch, msg.bytes.utf8String))
    )
}

Java

copysourcefinal int fanoutSize = 4;
final int bufferSize = 1;

Source<Pair<Integer, String>, NotUsed> mergedSources = Source.empty();
for (int i = 0; i < fanoutSize; i++) {
  final int fanoutBranch = i;
  mergedSources =
      mergedSources.merge(
          AmqpSource.atMostOnceSource(
                  TemporaryQueueSourceSettings.create(connectionProvider, exchangeName)
                      .withDeclaration(exchangeDeclaration),
                  bufferSize)
              .map(msg -> Pair.create(fanoutBranch, msg.bytes().utf8String())));
}

We merge all sources into one and add the index of the source to all incoming messages, so we can distinguish which source the incoming message came from.

Such sink and source can be started the same way as in the previous example.

Using rabbitmq as an RPC mechanism

If you have remote workers that you want to incorporate into a stream, you can do it using rabbit RPC workflow RabbitMQ RPC

Scala

copysourceval amqpRpcFlow = AmqpRpcFlow.simple(
  AmqpWriteSettings(connectionProvider).withRoutingKey(queueName).withDeclaration(queueDeclaration)
)

val (rpcQueueF: Future[String], probe: TestSubscriber.Probe[ByteString]) = Source(input)
  .map(s => ByteString(s))
  .viaMat(amqpRpcFlow)(Keep.right)
  .toMat(TestSink())(Keep.both)
  .run()

Java

copysourcefinal Flow<ByteString, ByteString, CompletionStage<String>> ampqRpcFlow =
    AmqpRpcFlow.createSimple(
        AmqpWriteSettings.create(connectionProvider)
            .withRoutingKey(queueName)
            .withDeclaration(queueDeclaration),
        1);

Pair<CompletionStage<String>, TestSubscriber.Probe<ByteString>> result =
    Source.from(input)
        .map(ByteString::fromString)
        .viaMat(ampqRpcFlow, Keep.right())
        .toMat(TestSink.create(system), Keep.both())
        .run(system);

Acknowledging messages downstream

Committable sources return CommittableReadResultCommittableReadResult which wraps the ReadResultReadResult and exposes the methods ack and nack.

Use ack to acknowledge the message back to RabbitMQ. ack takes an optional boolean parameter multiple indicating whether you are acknowledging the individual message or all the messages up to it.

Use nack to reject a message. Apart from the multiple argument, nack takes another optional boolean parameter indicating whether the item should be requeued or not.

Scala

copysourceval amqpSource = AmqpSource.committableSource(
  NamedQueueSourceSettings(connectionProvider, queueName)
    .withDeclaration(queueDeclaration),
  bufferSize = 10
)

val result: Future[immutable.Seq[ReadResult]] = amqpSource
  .mapAsync(1)(businessLogic)
  .mapAsync(1)(cm => cm.ack().map(_ => cm.message))
  .take(input.size)
  .runWith(Sink.seq)

val nackedResults: Future[immutable.Seq[ReadResult]] = amqpSource
  .mapAsync(1)(businessLogic)
  .take(input.size)
  .mapAsync(1)(cm => cm.nack(multiple = false, requeue = true).map(_ => cm.message))
  .runWith(Sink.seq)

Java

copysourcefinal Integer bufferSize = 10;
final Source<CommittableReadResult, NotUsed> amqpSource =
    AmqpSource.committableSource(
        NamedQueueSourceSettings.create(connectionProvider, queueName)
            .withDeclaration(queueDeclaration),
        bufferSize);

final CompletionStage<List<ReadResult>> result =
    amqpSource
        .mapAsync(1, this::businessLogic)
        .mapAsync(1, cm -> cm.ack(/* multiple */ false).thenApply(unused -> cm.message()))
        .take(input.size())
        .runWith(Sink.seq(), system);

final CompletionStage<List<ReadResult>> nackedResults =
    amqpSource
        .take(input.size())
        .mapAsync(1, this::businessLogic)
        .mapAsync(
            1,
            cm ->
                cm.nack(/* multiple */ false, /* requeue */ true)
                    .thenApply(unused -> cm.message()))
        .runWith(Sink.seq(), system);