Skip to content

Latest commit

 

History

History
1494 lines (1150 loc) · 57 KB

README.md

File metadata and controls

1494 lines (1150 loc) · 57 KB

Zeebe Test Container

Java CI Compatible with: Camunda Platform 8

Easily test your application against a containerized, configurable Zeebe instance.

Please refer to testcontainers.org for general documentation on how to use containers for your tests, as well as general prerequisites.

When to use this project?

In most cases, you probably do not need to use this project, and instead want to use zeebe-process-test.

zeebe-process-test is built to allow you to write unit and integration tests of your BPMN process definitions, DMN models, and how they interact together with your workers. It provides a stripped down Zeebe engine, one which fully supports the same BPMN and DMN features as Zeebe, but without the administrative capabilities and other advanced features (e.g. multiple partitions, backups, rebalancing, monitoring, etc.)

If all you want to do is test the correctness of your BPMN process definitions or DMN models, and how they interact with your workers, then stick with zeebe-process-test. However, if you want to do any of the following, the zeebe-test-container is the right tool:

  • write acceptance/integration tests for Zeebe itself
  • write acceptance/integration tests for a Zeebe gateway interceptor
  • write acceptance/integration tests for a Zeebe exporter
  • write automated acceptance tests that use advanced features (e.g. parallelism across multiple partitions, automated rebalancing of leadership, chaos/resilience tests, etc.)

This tool is not made for load testing, as you will most likely not achieve the same level of performance as you would in a production deployment.

Supported Zeebe versions

NOTE: version 1.0 is incompatible with Zeebe versions pre 0.23.x

Version 1.x and 2.x is compatible with the following Zeebe versions:

  • 0.23.x
  • 0.24.x
  • 0.25.x
  • 0.26.x

Version 3.x is compatible with the following Zeebe versions:

  • 1.x
  • 8.x

Installation

Add the project to your dependencies:

<dependency>
  <groupId>io.zeebe</groupId>
  <artifactId>zeebe-test-container</artifactId>
  <version>3.5.0</version>
</dependency>
testImplementation 'io.zeebe:zeebe-test-container:3.5.0'

Requirements

Zeebe Test Container is built for Java 8+, and will not work on lower Java versions.

You will need Java 21+ for development purposes, however, as many of our tests rely on shared Zeebe libraries which are built for Java 21+.

Additionally, you will need to comply with all the Testcontainers requirements, as defined here.

Compatibility guarantees

As there is currently only a single maintainer, only the latest major version will be maintained and supported.

zeebe-test-container uses API guardian to declare the stability and guarantees of its API.

Every class/interface/etc. is annotated with an @API annotation describing its status. Additionally, at times, inner members of a class/interface/etc. may be annotated with an overriding @API annotation. In that case, it would take precedence over its parent annotation. For example, if you have the following:

import org.apiguardian.api.API;
import org.apiguardian.api.API.Status;

@API(status = Status.STABLE)
public class MyClass {

  @API(status = Status.EXPERIMENTAL)
  public void myExperimentalMethod() {

  }

  public void myStableMethod() {

  }
}

Then we can assume the contract for MyClass is stable and will not be changed until the next major version, except for its method myExperimentalMethod, which is an experimental addition which may change at any time, at least until it is marked as stable or dropped.

NOTE: for contributors, please remember to annotate new additions, and to maintain the compatibility guarantees of pre-annotated entities.

Quickstart

Using with junit4

If you're using junit4, you can add the container as a rule: it will be started and closed around each test execution. You can read more about Testcontainers and junit4 here.

package com.acme.zeebe;

import io.camunda.zeebe.client.ZeebeClient;
import io.camunda.zeebe.client.api.response.DeploymentEvent;
import io.camunda.zeebe.client.api.response.ProcessInstanceResult;
import io.zeebe.containers.ZeebeContainer;
import io.camunda.zeebe.model.bpmn.Bpmn;
import io.camunda.zeebe.model.bpmn.BpmnModelInstance;
import org.assertj.core.api.Assertions;
import org.junit.Rule;
import org.junit.Test;

public class MyFeatureTest {

  @Rule
  public ZeebeContainer zeebeContainer = new ZeebeContainer();

  @Test
  public void shouldConnectToZeebe() {
    // given
    final ZeebeClient client =
      ZeebeClient.newClientBuilder()
        .gatewayAddress(zeebeContainer.getExternalGatewayAddress())
        .usePlaintext()
        .build();
    final BpmnModelInstance process =
      Bpmn.createExecutableProcess("process").startEvent().endEvent().done();

    // when
    // do something (e.g. deploy a process)
    final DeploymentEvent deploymentEvent =
      client.newDeployCommand().addProcessModel(process, "process.bpmn").send().join();

    // then
    // verify (e.g. we can create an instance of the deployed process)
    final ProcessInstanceResult processInstanceResult =
      client
        .newCreateInstanceCommand()
        .bpmnProcessId("process")
        .latestVersion()
        .withResult()
        .send()
        .join();
    Assertions.assertThat(processInstanceResult.getProcessDefinitionKey())
      .isEqualTo(deploymentEvent.getProcesses().get(0).getProcessDefinitionKey());
  }
}

Using with junit5

If you're using junit5, you can use the Testcontainers extension. It will manage the container lifecycle for you. You can read more about the extension here.

package com.acme.zeebe;

import io.camunda.zeebe.client.ZeebeClient;
import io.camunda.zeebe.client.api.response.DeploymentEvent;
import io.camunda.zeebe.client.api.response.ProcessInstanceResult;
import io.zeebe.containers.ZeebeContainer;
import io.camunda.zeebe.model.bpmn.Bpmn;
import io.camunda.zeebe.model.bpmn.BpmnModelInstance;
import org.assertj.core.api.Assertions;
import org.junit.jupiter.api.Test;
import org.testcontainers.junit.jupiter.Container;
import org.testcontainers.junit.jupiter.Testcontainers;

@Testcontainers
public class MyFeatureTest {

  @Container
  private final ZeebeContainer zeebeContainer = new ZeebeContainer();

  @Test
  void shouldConnectToZeebe() {
    // given
    final ZeebeClient client =
      ZeebeClient.newClientBuilder()
        .gatewayAddress(zeebeContainer.getExternalGatewayAddress())
        .usePlaintext()
        .build();
    final BpmnModelInstance process =
      Bpmn.createExecutableProcess("process").startEvent().endEvent().done();

    // when
    // do something (e.g. deploy a process)
    final DeploymentEvent deploymentEvent =
      client.newDeployCommand().addProcessModel(process, "process.bpmn").send().join();

    // then
    // verify (e.g. we can create an instance of the deployed process)
    final ProcessInstanceResult processInstanceResult =
      client
        .newCreateInstanceCommand()
        .bpmnProcessId("process")
        .latestVersion()
        .withResult()
        .send()
        .join();
    Assertions.assertThat(processInstanceResult.getProcessDefinitionKey())
      .isEqualTo(deploymentEvent.getProcesses().get(0).getProcessDefinitionKey());
  }
}

Usage

zeebe-test-container provides three different containers:

  • ZeebeContainer: a Zeebe broker with an embedded gateway
  • ZeebeBrokerContainer: a Zeebe broker without an embedded gateway
  • ZeebeGatewayContainer: a standalone Zeebe gateway
  • ZeebeCluster: an experimental cluster builder API which lets you manage a Zeebe cluster made of various containers

If you're unsure which one you should use, then you probably want to use ZeebeContainer, as it is the quickest way to test your application against Zeebe.

Broker with embedded gateway

ZeebeContainer will start a new Zeebe broker with embedded gateway. For most tests, this is what you will want to use. It provides all the functionality of a Zeebe single node deployment, which for testing purposes should be enough.

The container is considered started if and only if:

  1. The monitoring, command, cluster, and gateway ports are open and accepting connections (read more about the ports here) .
  2. The broker ready check returns a 204 (see more about this check here) .
  3. The gateway topology is considered complete.

A topology is considered complete if there is a leader for all partitions.

Once started, the container is ready to accept commands, and a client can connect to it by setting its gatewayAddress to ZeebeContainer#getExternalGatewayAddress().

Standalone broker without gateway

ZeebeBrokerContainer will start a new Zeebe broker with no embedded gateway. As it contains no gateway, the use case for this container is to test Zeebe in clustered mode. As such, it will typically be combined with a ZeebeGatewayContainer or a ZeebeContainer.

The container is considered started if and only if:

  1. The monitoring, command, and cluster ports are open and accepting connections (read more about the ports here) .
  2. The broker ready check returns a 204 (see more about this check here) .

Once started, the container is ready to accept commands via the command port; you should therefore link a gateway to it if you wish to use it.

Standalone gateway

ZeebeGatewayContainer will start a new Zeebe standalone gateway. As it is only a gateway, it should be linked to at least one broker - a ZeebeContainer or ZeebeBrokerContainer. By default, it will not expose the monitoring port, as monitoring is not enabled in the gateway by default. If you enable monitoring, remember to expose the port as well via GenericContainer#addExposedPort(int).

The container is considered started if and only if:

  1. The cluster and gateway ports are open and accepting connections (read more about the ports here) .
  2. The gateway topology is considered complete.

A topology is considered complete if there is a leader for all partitions.

Once started, the container is ready to accept commands, and a client can connect to it by setting its gatewayAddress to ZeebeContainer#getExternalGatewayAddress().

Configuring your container

Configuring your Zeebe container of choice is done exactly as you normally would - via environment variables or via configuration file. You can find out more about it on the Zeebe documentation website .

Zeebe 0.23.x and upwards use Spring Boot for configuration - refer to their documentation on how environment variables are mapped to configuration settings. You can read more about this here

Testcontainers provide mechanisms through which environment variables can be injected , or configuration files mounted. Refer to their documentation for more.

Examples

A series of examples are included as part of the tests, see test/java/io/zeebe/containers/examples.

Note that these are written for junit5.

Below you will find examples which are not part of the project itself, mostly due to requiring external resources or systems.

Testing a custom exporter

You can use zeebe-test-container to test your custom exporter. There are two ways to go about this: with a pre-built Docker image containing your exporter, or by loading a pre-packaged JAR containing your exporter.

Loading exporter JAR

The idea here is to load your custom exporter, pre-packaged into a self-contained JAR, either as a file or a classpath resource.

Here are some projects which do exactly this:

You can do this in two ways:

  1. Ensure the artifact is packaged separately and made available via the file system
  2. Hook into your build pipeline to always have the latest JAR available to your QA tests

The second option is more work, but it provides the better experience as running mvn verify for your QA module will always run your tests with the latest code changes seamlessly.

Build and test the exporter JAR

You can do this by creating a second module or project specifically to run integration tests for your exporter.

Note

The following assumes you're using Maven, but it should be possible to adapt it to Gradle.

First, create the new project, say, exporter-qa. It will need to list your exporter module, say exporter, as a dependency. This will ensure that the exporter module is built before exporter-qa.

For example:

<dependencies>
  <!-- ensure the exporter project is built before the QA project -->
  <dependency>
    <groupId>com.acme</groupId>
    <artifactId>exporter</artifactId>
    <scope>provided</scope>
    <optional>true</optional>
  </dependency>
</dependencies>

Next, you we need to copy the packaged JAR of the exporter module into the exporter-qa module:

<build>
  <plugins>
    <!-- copy over the exporter JAR as a resource -->
    <plugin>
      <groupId>org.apache.maven.plugins</groupId>
      <artifactId>maven-dependency-plugin</artifactId>
      <version>${plugin.version.dependency}</version>
      <configuration>
        <!-- you may require this if you aren't actually referencing the exporter itself -->
        <ignoredUnusedDeclaredDependencies>
          <dep>com.acme:exporter</dep>
        </ignoredUnusedDeclaredDependencies>
      </configuration>
      <dependencies>
        <!-- Fake dependency on the exporter to ensure it's packaged beforehand -->
        <dependency>
          <groupId>com.acme</groupId>
          <artifactId>exporter</artifactId>
          <version>${project.version}</version>
        </dependency>
      </dependencies>
      <executions>
        <execution>
          <id>copy</id>
          <goals>
            <goal>copy</goal>
          </goals>
          <phase>package</phase>
          <configuration>
            <artifactItems>
              <artifactItem>
                <groupId>com.acme</groupId>
                <artifactId>exporter</artifactId>
                <version>${project.version}</version>
                <type>jar</type>
                <outputDirectory>${project.basedir}/src/main/resources</outputDirectory>
                <destFileName>exporter.jar</destFileName>
              </artifactItem>
            </artifactItems>
            <overWriteReleases>false</overWriteReleases>
            <overWriteSnapshots>true</overWriteSnapshots>
          </configuration>
        </execution>
      </executions>
    </plugin>
  </plugins>
</build>

Note that since we're copying the exporter in the package phase, you will need to run your QA tests in the integration-test phase. To simplify this, we recommend you use the Maven failsafe plugin.

Once done, you have everything you need to write an integration test in the exporter-qa module. For example:

package com.acme.exporter.qa;

import io.zeebe.containers.ZeebeContainer;
import org.junit.jupiter.api.Test;
import org.testcontainers.junit.jupiter.Container;
import org.testcontainers.junit.jupiter.Testcontainers;
import org.testcontainers.utility.MountableFile;

@Testcontainers
public final class MyExporterIT {
  @Container
  private final ZeebeContainer zeebe = new ZeebeContainer()
    .withCopyFileToContainer(MountableFile.forClasspathResource("exporter.jar", 0775), "/usr/local/zeebe/exporters/exporter.jar")
    .withEnv("ZEEBE_BROKER_EXPORTERS_ACME_CLASSNAME", "com.acme.exporter.Exporter")
    .withEnv("ZEEBE_BROKER_EXPORTERS_ACME_JARPATH", "/usr/local/zeebe/exporters/exporter.jar")
    .withEnv("ZEEBE_BROKER_EXPORTERS_ACME_ARGS_0_PROP", "foo");

  @Test
  void shouldExportData() {
    // write your test here
  }
}

Loading Docker image

The idea here is to prepare a Docker image based on Zeebe's official image which already contains your packaged exporter JAR, so no file mounting or packaging is necessary in advance. Here's an example of such an image built using Jib: Hazelcast Exporter. However, you do need to re-package your Docker image after every code change to your exporter before running your QA test.

Here are some projects which do exactly this:

Let's say your build pipeline has already built the image as ghcr.io/acme/exporter, and the exporter JAR is located at /usr/local/zeebe/exporters/exporter.jar in that image. Then you could write the following test:

package com.acme.exporter.qa;

import io.zeebe.containers.ZeebeContainer;
import org.junit.jupiter.api.Test;
import org.testcontainers.junit.jupiter.Container;
import org.testcontainers.junit.jupiter.Testcontainers;
import org.testcontainers.utility.DockerImageName;

@Testcontainers
public final class MyExporterIT {
  private static final DockerImageName TEST_IMAGE = DockerImageName.parse("ghcr.io/acme/exporter");

  @Container
  private final ZeebeContainer zeebe = new ZeebeContainer(TEST_IMAGE)
    .withEnv("ZEEBE_BROKER_EXPORTERS_ACME_CLASSNAME", "com.acme.exporter.Exporter")
    .withEnv("ZEEBE_BROKER_EXPORTERS_ACME_JARPATH", "/usr/local/zeebe/exporters/exporter.jar")
    .withEnv("ZEEBE_BROKER_EXPORTERS_ACME_ARGS_0_PROP", "foo");

  @Test
  void shouldExportData() {
    // write your test here
  }
}

Continuous Integration

If you wish to use this with your continous integration pipeline (e.g. Jenkins, CircleCI), the Testcontainers has a section explaining how to use it, how volumes can be shared, etc.

Experimental features

There are currently several experimental features across the project. As described in the compatibility guarantees, we only guarantee backwards compatibility for stable APIs (marked by the annotation @API(status = Status.STABLE)). Typically, you shouldn't be using anything else. However, there are some features which already provide value, but for which the correct API is unclear; these are marked with @API(status = Status.EXPERIMENTAL). These may be changed, or dropped depending on their usefulness.

NOTE: you should never use anything marked as @API(status = Status.INTERNAL). These are there purely for internal purposes, and cannot be relied on at all.

Cluster

NOTE: the cluster API is currently an experimental API. You're encouraged to use it and give feedback, as this is how we can validate it. Keep in mind however that it is subject to change in the future.

A typical production Zeebe deployment will be a cluster of nodes, some brokers, and possibly some standalone gateways. It can be useful to test against such deployments for acceptance or E2E tests.

While it's not too hard to manually link several containers, it can become tedious and error prone if you want to test many different configurations. The cluster API provides you with an easy way to programmatically build Zeebe deployments while minimizing the surface of configuration errors.

NOTE: if you have a static deployment that you don't need to change programmatically per test, then you might want to consider setting up a static Zeebe cluster in your CI pipeline (either via Helm or docker-compose), or even using Testcontainer's docker-compose feature.

Usage

The main entry point is the ZeebeClusterBuilder, which can be instantiated via ZeebeCluster#builder(). The builder can be used to configure the topology of your cluster. You can configure the following:

  • the number of brokers in the cluster (by default 1)
  • whether brokers should be using the embedded gateway (by default true)
  • the number of standalone gateways in the cluster (by default 0)
  • the number of partitions in the cluster (by default 1)
  • the replication factor of each partition (by default 1)
  • the network the containers should use (by default Network#SHARED)

Container instances (e.g. ZeebeBrokerContainer or ZeebeContainer) are only instantiated and configured once you call #build(). At this point, your cluster is configured in a valid way, but isn't started yet - that is, no real Docker containers have been created/started.

Once your cluster is built, you can access its brokers and gateways via ZeebeCluster#getBrokers and ZeebeCluster#getGateways. This allows you to further configure each container as you wish before actually starting the cluster.

Examples

Here is a short example on how to set up a cluster for testing with junit5.

package com.acme.zeebe;

import io.camunda.zeebe.client.ZeebeClient;
import io.camunda.zeebe.client.api.response.BrokerInfo;
import io.camunda.zeebe.client.api.response.Topology;
import io.zeebe.containers.cluster.ZeebeCluster;
import java.util.List;
import java.util.concurrent.TimeUnit;
import org.assertj.core.api.Assertions;
import org.junit.jupiter.api.AfterEach;
import org.junit.jupiter.api.Test;
import org.junit.jupiter.api.Timeout;
import org.testcontainers.junit.jupiter.Container;
import org.testcontainers.junit.jupiter.Testcontainers;

/**
 * Showcases how you can create a test with a cluster of two brokers and one standalone gateway.
 * Configuration is kept to minimum, as the goal here is only to showcase how to connect the
 * different nodes together.
 */
@Testcontainers
class ZeebeClusterWithGatewayExampleTest {

  @Container
  private final ZeebeCluster cluster =
    ZeebeCluster.builder()
      .withEmbeddedGateway(false)
      .withGatewaysCount(1)
      .withBrokersCount(2)
      .withPartitionsCount(2)
      .withReplicationFactor(1)
      .build();

  @Test
  @Timeout(value = 15, unit = TimeUnit.MINUTES)
  void shouldStartCluster() {
    // given
    final Topology topology;

    // when
    try (final ZeebeClient client = cluster.newClientBuilder().build()) {
      topology = client.newTopologyRequest().send().join(5, TimeUnit.SECONDS);
    }

    // then
    final List<BrokerInfo> brokers = topology.getBrokers();
    Assertions.assertThat(topology.getClusterSize()).isEqualTo(3);
    Assertions.assertThat(brokers)
      .hasSize(2)
      .extracting(BrokerInfo::getAddress)
      .containsExactlyInAnyOrder(
        cluster.getBrokers().get(0).getInternalCommandAddress(),
        cluster.getBrokers().get(1).getInternalCommandAddress());
  }
}

You can find more examples by looking at the test/java/io/zeebe/containers/examples/cluster package.

Cluster startup time

There are some caveat as well. For example, if you want to create a large cluster with many brokers and need to increase the startup time:

package com.acme.zeebe;

import io.zeebe.containers.cluster.ZeebeCluster;
import java.time.Duration;
import java.util.concurrent.TimeUnit;
import org.junit.jupiter.api.AfterEach;
import org.junit.jupiter.api.Test;
import org.junit.jupiter.api.Timeout;
import org.testcontainers.junit.jupiter.Container;
import org.testcontainers.junit.jupiter.Testcontainers;

@Testcontainers
class ZeebeHugeClusterTest {

  @Container
  private final ZeebeCluster cluster =
    ZeebeCluster.builder()
      .withEmbeddedGateway(false)
      .withGatewaysCount(3)
      .withBrokersCount(6)
      .withPartitionsCount(6)
      .withReplicationFactor(3)
      // configure each container to have a high start up time as they get started in parallel
      .withNodeConfig(node -> node.self().withStartupTimeout(Duration.ofMinutes(5)))
      .build();

  @Test
  @Timeout(value = 30, unit = TimeUnit.MINUTES)
  void shouldStartCluster() {
    // test things
  }
}

Debugging

There might be cases where you want to debug a container you just started in one of your tests. You can use the RemoteDebugger utility for this. By default, it will start your container and attach a debugging agent to it on port 5005. The container startup is then suspended until a debugger attaches to it.

NOTE: since the startup is suspended until a debugger connects to it, it's possible for a the startup strategy to time out if no debugger connects to it.

You can use it with any container as:

package com.acme.zeebe;

import io.zeebe.containers.ZeebeContainer;
import io.zeebe.containers.util.RemoteDebugger;
import org.junit.jupiter.api.Test;
import org.testcontainers.junit.jupiter.Container;
import org.testcontainers.junit.jupiter.Testcontainers;

@Testcontainers
public class MyFeatureTest {

  @Container
  private final ZeebeContainer zeebeContainer = RemoteDebugger.configure(new ZeebeContainer());

  @Test
  void shouldTestProperty() {
    // test...
  }
}

Note that RemoteDebugger#configure(GenericContainer<?>) returns the same container, so you can use the return value to chain more configuration around your container.

package com.acme.zeebe;

import io.zeebe.containers.ZeebeContainer;
import io.zeebe.containers.util.RemoteDebugger;
import org.junit.jupiter.api.Test;
import org.testcontainers.junit.jupiter.Container;
import org.testcontainers.junit.jupiter.Testcontainers;

@Testcontainers
public class MyFeatureTest {

  @Container
  private final ZeebeContainer zeebeContainer = RemoteDebugger.configure(new ZeebeContainer())
    .withEnv("ZEEBE_BROKER_NETWORK_HOST", "0.0.0.0");

  @Test
  void shouldTestProperty() {
    // test...
  }
}

You can also configure the port of the debug server, or even configure the startup to not wait for a debugger to connect by using RemoteDebugger#configure(GenericContainer<?>, int, boolean). See the Javadoc for more.

Volumes and data

Zeebe brokers store all their data under a configurable data directory (default to /usr/local/zeebe/data). By default, when you start a container, this data is ephemeral and will be deleted when the container is removed.

If you want to keep the data around, there's a few ways you can do so. One option is to use a folder on the host machine, and mount it as the data directory. Here's an example:

package com.acme.zeebe;

import static org.assertj.core.api.Assertions.assertThat;

import io.zeebe.containers.ZeebeBrokerContainer;
import io.zeebe.containers.ZeebeHostData;
import java.util.concurrent.TimeUnit;
import org.junit.jupiter.api.Test;
import org.junit.jupiter.api.Timeout;

final class HostDataExampleTest {

  @Test
  @Timeout(value = 5, unit = TimeUnit.MINUTES)
  void shouldSaveDataOnTheHost(@TempDir final Path tempDir) {
    // given
    final ZeebeHostData data = new ZeebeHostData(tempDir.toString());
    try (final ZeebeBrokerContainer container = new ZeebeBrokerContainer().withZeebeData(data)) {
      // when
      container.start();
    }

    // then
    assertThat(tempDir).isNotEmptyDirectory();
  }
}

This will have Zeebe write its data directly to a directory on your host machine. Note that the permissions of the written files will be assigned to the user the container runs in. This means if the broker process in the container runs as root, then the files written will belong to root, and your user may not be able to run it. To circumvent this, you will need to run your container as your user as well. You can do so by modifying the create command passing the correct user to the host config, e.g. container.withCreateContainerCmdModifier(cmd -> cmd.withUser("1000:0")).

NOTE: this may or may not work on Windows. I have no access to a Windows machine, so I can't really say how permissions should look like for a Windows machine. It's recommended instead to use volumes, and extract the data from it if you need.

On the other hand, you can also use a Docker volume for this. The volume is reusable across multiple runs, and can also be mounted on different containers. Here's an example:

package com.acme.zeebe;

import static org.assertj.core.api.Assertions.assertThatCode;

import io.zeebe.containers.ZeebeBrokerContainer;
import io.zeebe.containers.ZeebeVolume;
import java.util.concurrent.TimeUnit;
import org.junit.jupiter.api.Test;
import org.junit.jupiter.api.Timeout;

final class VolumeExampleTest {

  @Test
  @Timeout(value = 5, unit = TimeUnit.MINUTES)
  void shouldUseVolume() {
    // given
    final ZeebeVolume volume = ZeebeVolume.newVolume();
    try (final ZeebeContainer container = new ZeebeContainer().withZeebeData(volume)) {
      // when
      container.start();
      container.stop();

      // then
      assertThatCode(() -> container.start()).doesNotThrowExceptions();
    }
  }
}

You can see a more complete example here.

Extracting data

At times, you may want to extract data from a running container, or from a volume. This can happen if you want to run a broker locally based on test data for debugging. It can also be used to generate Zeebe data which can be reused in further test as a starting point to avoid always regenerating that data.

There are two main interfaces for this. If you want to extract the data from a running container, you can directly use ContainerArchive. This represents a reference to a zipped, TAR file on a given container, which can be extracted to a local path.

Here's an example which will extract Zeebe's default data directory to /tmp/zeebe. This will result in a copy of the Zeebe data directory at /tmp/zeebe/usr/local/zeebe/data.

package com.acme.zeebe;

import static org.assertj.core.api.Assertions.assertThat;

import io.zeebe.containers.ZeebeBrokerContainer;
import io.zeebe.containers.archive.ContainerArchive;
import java.io.IOException;
import java.nio.file.Path;
import org.junit.jupiter.api.Test;
import org.testcontainers.junit.jupiter.Container;
import org.testcontainers.junit.jupiter.Testcontainers;

@Testcontainers
final class ExtractDataLiveExampleTest {

  @Container
  private final ZeebeBrokerContainer container = new ZeebeBrokerContainer();

  @Test
  void shouldExtractData() {
    // given
    final Path destination = Paths.get("/tmp/zeebe");
    final ContainerArchive archive = ContainerArchive.builder().withContainer(container).build();

    // when
    archive.extract(destination);

    // then
    assertThat(destination).isNotEmptyDirectory();
    assertThat(destination.resolve("usr/local/zeebe/data")).isNotEmptyDirectory();
  }
}

NOTE: if all you wanted was to download the data, you could simply use ContainerArchive#transferTo(Path). This will download the zipped archive to the given path as is.

You can find more examples for this feature under examples/archive.

Time traveling

Since version 1.3, Zeebe provides a coarse time traveling API via the actor clock actuator endpoint. While this actuator is enabled by default, the internal clock is itself immutable. The only thing you can do out of the box is get the current time as seen by Zeebe's actor framework.

In order to test time based events - such as triggering a timer catch event from one of your deployed processes - you can start the node with a mutable clock. To do so, you need to set the configuration flag zeebe.clock.controlled = true. This can be done with Testcontainers by passing the following environment variable:

package com.acme.zeebe;

import io.zeebe.containers.ZeebeNode;
import org.junit.jupiter.api.BeforeEach;
import org.junit.jupiter.api.Test;

final class SetupMutableClockExample {

  private ZeebeNode<?> node; // make sure this value is instantiated

  @BeforeEach
  void beforeEach() {
    node.withEnv("ZEEBE_CLOCK_CONTROLLED", "true");
  }

  // any subsequent tests can now mutate the clock
  @Test
  void shouldMutateTheClock() {
  }
}

NOTE: if you forget to mutate the clock, the actuator will return a 403 with an explicit error message about the clock's immutability.

Once the clock is mutable, you can then use the provided high level API, ZeebeClock.

Here's a basic example which will simply advance the broker's time:

package io.zeebe.containers.clock;

import static org.assertj.core.api.Assertions.assertThat;

import io.zeebe.containers.ZeebeBrokerContainer;
import java.time.Duration;
import java.time.Instant;
import org.junit.jupiter.api.AfterEach;
import org.junit.jupiter.api.Test;
import org.junit.jupiter.api.parallel.Execution;
import org.junit.jupiter.api.parallel.ExecutionMode;
import org.testcontainers.junit.jupiter.Container;
import org.testcontainers.junit.jupiter.Testcontainers;

@Execution(ExecutionMode.SAME_THREAD)
@Testcontainers
final class ZeebeClockTest {

  @Container
  private static final ZeebeBrokerContainer BROKER =
    new ZeebeBrokerContainer().withEnv("ZEEBE_CLOCK_CONTROLLED", "true");

  private final ZeebeClock clock = ZeebeClock.newDefaultClock(BROKER);

  @AfterEach
  void afterEach() {
    ZeebeClock.newDefaultClock(BROKER).resetTime();
  }

  @Test
  void shouldAddTime() {
    // given
    final Instant pinnedTime = clock.pinTime(Instant.now());
    final Duration offset = Duration.ofDays(1);

    // when
    final Instant modifiedTime = clock.addTime(offset);

    // then
    final Instant expectedTime = pinnedTime.plus(offset);
    assertThat(modifiedTime).isEqualTo(pinnedTime.plus(offset)).isEqualTo(expectedTime);
  }
}

You can find more examples for this feature under examples/clock.

Tips and limitations

Reusing the same container

If you wish to reuse the same container across multiple tests where you modify the time, make sure to include a trigger after each test (e.g. @AfterEach in Junit 5) to reset the clock (i.e. ZeebeClock#resetTime()). It's also recommended you ensure you run all your tests against that one node sequentially to avoid flaky tests.

Embedded gateway issues

Some requests will not play nice with time travel when you modify a broker's clock, if the request is going through that same broker's embedded gateway. For example, if you create a process instance and await its result (i.e. #withResult()), then advance the clock, the request will time out since the clock is shared between the broker and the embedded gateway.

Debug exporter

The module comes with a built-in, injectable debug exporter and receiver pair. This allows you to stream records out of a broker container into your local tests. You can use this to view the log, assert certain events are happening, etc. If you're an exporter author, you can use this to verify that your exporter is exporting the expected events as well.

To use it, you must first start an instance of a DebugReceiver, which is a lightweight server binding to any InetSocketAddress. The safest way to construct one is to use the default constructor, DebugReceiver(Consumer<Record<?>>), which will pick a random, available, local port to bind to. One receiver per cluster should be used, as receivers can also acknowledge records.

Exported records will be streamed to the given consumer. This is done in a separate thread, thus you must ensure the consumer is thread-safe. Records are streamed in the order they come in, but will be sequenced relatively to their partition. This means, all records on partition P will be ordered by increasing position, but ordering between partitions is undefined (though it's roughly time based, i.e. as the records are exported by the brokers).

Here is the smallest example usage:

package com.acme;

import static org.assertj.core.api.Assertions.assertThat;

import io.camunda.zeebe.protocol.record.Record;
import io.zeebe.containers.ZeebeContainer;
import io.zeebe.containers.exporter.DebugReceiver;
import java.util.List;
import java.util.concurrent.CopyOnWriteArrayList;
import org.junit.jupiter.api.Test;

final class WithDebugExporterTest {

  @Test
  void shouldReadExportedRecords() {
    final List<Record<?>> records = new CopyOnWriteArrayList<>();
    try (final DebugReceiver receiver = new DebugReceiver(records::add).start()) {
      final ZeebeContainer container =
        new ZeebeContainer().withDebugExporter(receiver.serverAddress().getPort());

      // ... send some commands that will publish some records.
      // records are then streamed to the consumer, which in our case appends them to the records
      // list.
      assertThat(records).hasSize(1);
    }
  }
}

You can view a longer example here.

Acknowledging records

The receiver can acknowledge records, such that Zeebe will mark them as having been successfully exported and will not re-export them (and may delete them).

By default, records are not acknowledged, and are never deleted. To acknowledge records, you can use the DebugReceiver#acknowledge(int, long) method, passing the partition ID and the highest acknowledged position for that partition.

If you wish records to be auto-acknowledged, you can simply set Long.MAX_VALUE as the position for each known partition before any records are exported.

There is one limitation, which is that the acknowledged position is the one returned by the server. This means, if you receive record 1, and then acknowledge that, only when receiving record 2 will the acknowledged position take effect.

Zeebe Process Test Compatibility

It's possible to use a container or cluster as the backing engine when working with zeebe-process-test. This will let you reuse all the assertions you are used to, while running integration tests using one or more actual Zeebe instances.

Usage

In order to use it, you will need to add the following dependency to your project:

Maven:

<dependency>
  <groupId>io.zeebe</groupId>
  <artifactId>zeebe-test-container-engine</artifactId>
  <version>3.5.0</version>
</dependency>

Gradle:

testImplementation 'io.zeebe:zeebe-test-container-engine:3.5.0'

The usage differs a little from normal zeebe-process-test usage. Whereas there you would use the @ZeebeProcessTest annotation as your Junit 5 extension, here we stick with the standard Testcontainers annotations, i.e. the combination of @Testcontainers and @Container. Doing so means you keep using familiar tools to manage the lifecycle of your containers, and it lets us be more flexible when it comes to customizing said containers.

To illustrate, here is a minimal example which deploys a process definition, creates an instance, and waits for its completion:

package com.acme;

import io.camunda.zeebe.client.ZeebeClient;
import io.camunda.zeebe.client.api.response.ProcessInstanceEvent;
import io.camunda.zeebe.model.bpmn.Bpmn;
import io.camunda.zeebe.model.bpmn.BpmnModelInstance;
import io.camunda.zeebe.process.test.assertions.BpmnAssert;
import io.zeebe.containers.ZeebeContainer;
import io.zeebe.containers.engine.ContainerEngine;
import java.time.Duration;
import java.util.concurrent.TimeoutException;
import org.junit.jupiter.api.Test;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.testcontainers.containers.output.Slf4jLogConsumer;
import org.testcontainers.junit.jupiter.Container;
import org.testcontainers.junit.jupiter.Testcontainers;

@Testcontainers
final class ExampleTest {

  @Container
  private final ContainerEngine engine = ContainerEngine.createDefault();

  @Test
  void shouldCompleteProcessInstance() {
    // given
    final BpmnModelInstance processModel =
      Bpmn.createExecutableProcess("process").startEvent().endEvent().done();
    final ProcessInstanceEvent processInstance;

    // when
    try (final ZeebeClient client = engine.createClient()) {
      client.newDeployResourceCommand().addProcessModel(processModel, "process.bpmn").send().join();
      processInstance =
        client.newCreateInstanceCommand().bpmnProcessId("process").latestVersion().send().join();
    }

    // then
    engine.waitForIdleState(Duration.ofSeconds(5));
    BpmnAssert.assertThat(processInstance).isStarted().isCompleted();
  }
}

As you can see, you can keep writing your tests the exact same way as you would with zeebe-process-test, the only difference is in the creation of the ZeebeTestEngine.

You can look at ContainerEngine for an explanation of how it differs from the normal ZeebeTestEngine, and for hints on how to build customized engines. There are also some examples which hopefully illustrate how to use this with all the customization options.

Reference

The integration is done by leveraging a few other experimental features. When you create a container engine, the underlying containers are automatically configured to have a controllable clock and to export their records using the debug exporter.

This means when configuring the containers manually, do not disable or override these two features.

Caveats

Idle/busy state

The bundled, stripped down engine available with zeebe-process-test allows for much tighter integration with the assertions, such that it can accurately report if it is idle or busy. When running with a real Zeebe instance, this is hidden from us.

Instead, we define idle to mean that no records are exported by the debug exporter for a certain period of time (the ContainerEngineBuilder#withIdlePeriod configuration). The idle state is defined over all partitions, not only a single partition.

Similarly, we define a busy state as at least one record was exported during the timeout period.

Awaitility issues

If you wish to wrap your BpmnAssert calls with Awaitility for resilience, note that BpmnAssert sets the record stream source as a thread local. Since Awaitility, by default, polls on different threads, you may run into issues where no record stream source is found, or the wrong record stream source is used.

To properly use it, make sure to either configure Awaility#pollInSameThread, or in your callback, overwrite the thread local stream source with BpmnAssert#initRecordStreamSource. This latter is not recommended if you run your tests in parallel, however, as you may affect other tests.

Tips

Tailing your container's logs during development

As containers are somewhat opaque by nature (though Testcontainers already does a great job of making this more seamless), it's very useful to add a LogConsumer to a container. This will consume your container logs and pipe them out to your consumer. If you're using SLF4J, you can use the Slf4jLogConsumer and give it a logger, which makes it much easier to debug if anything goes wrong. Here's an example of how this would look like:

package com.acme.zeebe;

import io.zeebe.containers.ZeebeContainer;
import org.junit.jupiter.api.Test;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.testcontainers.containers.output.Slf4jLogConsumer;
import org.testcontainers.junit.jupiter.Container;
import org.testcontainers.junit.jupiter.Testcontainers;

@Testcontainers
public class MyFeatureTest {

  private static final Logger LOGGER =
    LoggerFactory.getLogger("com.acme.zeebe.MyFeatureTest.zeebeContainer");

  @Container
  private final ZeebeContainer zeebeContainer =
    new ZeebeContainer().withLogConsumer(new Slf4jLogConsumer(LOGGER));

  @Test
  void shouldTestProperty() {
    // test...
  }
}

Configuring GenericContainer specific properties with a Zeebe*Node interface

There are three container types in this module: ZeebeContainer, ZeebeBrokerContainer, ZeebeGatewayContainer. ZeebeContainer is a special case which can be both a gateway and a broker. To support a more generic handling of broker/gateway concept, there are two types introduced which are ZeebeBrokerNode and ZeebeGatewayNode. These types, however, are not directly extending GenericContainer.

At times, when you have a ZeebeBrokerNode or ZeebeGatewayNode, you may want to additionally configure it in ways that are only available to GenericContainer instances. You can use Container#self() to get the node's GenericContainer representation.

For example:

package com.acme.zeebe;

import io.zeebe.containers.cluster.ZeebeCluster;
import java.util.concurrent.TimeUnit;
import org.junit.jupiter.api.AfterEach;
import org.junit.jupiter.api.Test;
import org.junit.jupiter.api.Timeout;

class ZeebeHugeClusterTest {

  private final ZeebeCluster cluster =
    ZeebeCluster.builder()
      .withBrokersCount(1)
      .build();

  @AfterEach
  void tearDown() {
    cluster.stop();
  }

  @Test
  @Timeout(value = 5, unit = TimeUnit.MINUTES)
  void shouldStartCluster() {
    // given
    final ZeebeBrokerNode<?> broker = cluster.getBrokers().get(0);
    broker.self().withStartupTimeout(Duration.ofMinutes(2));
    cluster.start();

    // test more things
  }
}

Advanced Docker usage via the DockerClient

Remember that you have access to the raw docker client via GenericContainer#getDockerClient(). This lets you do all kinds of things at runtime, such as fiddling with volumes, networks, etc., which is a great way to test failure injection.

Gracefully restarting a container

NOTE: this is an advanced feature which is experimental and may still cause issues, so use at your own risk.

Testcontainers itself does not provide a convenience method to stop or pause a container. Using GenericContainer#stop() will actually kill and remove the container. To do these things, you can use the DockerClient directly, and send standard docker commands (e.g. stop, pause, etc.).

Alternatively, we provide a convenience method as ZeebeNode#shutdownGracefully() (e.g. ZeebeContainer#shutdownGracefully(), ZeebeBrokerContainer#shutdownGracefully()). This will only stop the container with a grace period, after which it will kill the container if not stopped. However, the container is not removed, and can be restarted later.

Keep in mind that to restart it you need to use a DockerClient#startContainerCmd(String), as just calling GenericContainer#start() will not start your container again.

Limiting container resources

When starting many containers, you can use GenericContainer#withCreateContainerCmdModifier() on creation to limit the resources available to them. This can be useful when testing locally on a development machine and having to start multiple containers.

Contributing

Contributions are more than welcome! Please make sure to read and adhere to the Code of Conduct. Additionally, in order to have your contributions accepted, you will need to sign the Contributor License Agreement.

Build from source

Prerequisites

In order to build from source, you will need to install maven 3.6+. You can find more about it on the maven homepage.

To build the project, you will need a JDK 21 installed locally. Note however that the core and engine modules are targeting Java 8 for compatibility purposes, and we must ensure that we maintain compatibility. To do this, the CI pipeline will run the tests using Java 8.

Finally, you will need to install Docker on your local machine.

Modules

The library is split into three modules:

  • core: the core library. It's artifact ID is zeebe-test-container for backwards compatibility. This is what users will include in their project.
  • engine: the implementation of ZeebeTestEngine, the compatibility layer between this library and Zeebe Process Test.
  • exporter: the debug exporter module. It will be packaged as a fat JAR and included as a resource in the core module. It has to be a separate module as it targets Java 21, same as the zeebe-exporter-api module it implements.
  • exporter-test: a module to test the integration between DebugReceiver and DebugExporter, without having to run everything through an actual broker.

Building

With all requirements ready, you can now simply clone the repository , and from its root, run the following command:

mvn clean install

This will build the project and run all tests locally.

Should you wish to only build without running the tests, you can run:

mvn clean package

Testing

Testing is done via GitHub actions, using two workflows:

  • lint.yml: includes jobs which ensure code quality (e.g. static analysis, linting, formatting, etc.)
  • test.yml: testing jobs for each module. For the core module, there are two testing jobs, one using TestContainers Cloud, and one using the local Docker daemon. This is due to a limitation of TestContainers Cloud, which does not work with host binds. Any tests which need to run on the local job should be annotated with @DisabledIfTestcontainersCloud.

One important thing to note is that we package and copy the debug exporter into the core module during the build process. This means that any tests which relies on the debug exporter being accessible has to be an integration test run by failsafe (i.e. test files ending with IT).

Code style

The project uses Spotless to apply consistent formatting and licensing to all project files. By default, the build only performs the required checks. If you wish to auto format/license your code, run:

mvn spotless:apply

Backwards compatibility

Zeebe Test Container uses a Semantic Versioning scheme for its versions, and revapi to enforce backwards compatibility according to its specification.

Additionally, we also use apiguardian to specify backwards compatibility guarantees on a more granular level. As such, only APIs marked as STABLE are considered when enforcing backwards compatibility.

If you wish to incubate a new feature, or if you're unsure about a new API type/method, please use the EXPERIMENTAL status for it. This will give us flexibility to test out new features and change them easily if we realize they need to be adapted.

Note that this only applies to the core/ module. All other modules are currently considered implementation details, and are not released anyway.

Report issues or contact developers

Work on Zeebe Test Container is done entirely through the Github repository. If you want to report a bug or request a new feature feel free to open a new issue on GitHub issues.

Create a Pull Request

To work on an issue, follow the following steps:

  1. Ensure that an issue exists for the task you want to work on. If one does not, create one.

  2. Checkout the master branch and pull the latest changes.

    git checkout develop
    git pull
    
  3. Create a new branch with the naming scheme issueId-description.

    git checkout -b 123-my-new-feature
    
  4. Follow the Google Java Format and Zeebe Code Style while coding.

  5. Implement the required changes on your branch, and make sure to build and test your changes locally before opening a pull requests for review.

  6. If you want to make use of the CI facilities before your feature is ready for review, feel free to open a draft PR.

  7. If you think you finished the issue please prepare the branch for reviewing. In general the commits should be squashed into meaningful commits with a helpful message. This means cleanup/fix etc commits should be squashed into the related commit.

  8. Finally, be sure to check on the CI results and fix any reported errors.

Commit Message Guidelines

Commit messages use Conventional Commits format, with a slight twist. See the Zeebe commit guidelines for more .

Contributor License Agreement

You will be asked to sign our Contributor License Agreement when you open a Pull Request. We are not asking you to assign copyright to us, but to give us the right to distribute your code without restriction. We ask this of all contributors in order to assure our users of the origin and continuing existence of the code. You only need to sign the CLA once.

Note that this is a general requirement of any Camunda Community Hub project.