micronaut-camunda-platform-7

This open source project allows you to easily integrate Camunda Platform 7 into Micronaut projects.

The Micronaut Framework is known for its efficient use of resources. With this integration you embed the BPMN 2.0 compliant and developer friendly Camunda process engine with minimal memory footprint into your application.

The integration is preconfigured with sensible defaults, so that you can get started with minimal configuration: simply add a dependency in your Micronaut project!

If you are interested in using Camunda's cloud native solution Zeebe on a Micronaut application instead, have a look at our open source project micronaut-zeebe-client.

We're not aware of all installations of our Open Source project. However, we love

• listening to your feedback,
• discussing possible use cases with you,
• aligning the roadmap to your needs!

📨 Please activate telemetry and contact us!

Do you want to try it out? Please jump to the Getting Started section.

Do you want to contribute to our open source project? Please read the Contribution Guidelines and contact us.

If you also want to run your External Task Client on a Micronaut application, have a look at the open source project micronaut-camunda-external-client.

Micronaut Framework + Camunda = :heart:

Table of Contents

• ✨ Features
• 🚀 Getting Started
  • Create Project with Micronaut Launch
  • Deploying Models
  • Camunda Integration
  • Configuration
  • Examples
  • Supported JDKs
• 🏆 Advanced Topics
  • Dependency Management
  • Camunda REST API and Webapps
  • Camunda Enterprise Edition (EE)
  • Process Engine Plugins
  • Custom Process Engine Configuration
  • Custom Job Executor Configuration
  • Transaction Management
  • Performance
  • Architectural Design
  • Keycloak
  • Eventing Bridge
  • Process Tests
  • Docker
  • Updating Camunda
  • Pitfalls
• 📚 Releases
• 📆 Publications
• 📨 Contact

✨Features

• Camunda can be integrated as an embedded process engine into a Micronaut project by simply adding a dependency in build.gradle (Gradle) or pom.xml (Maven).
• Using H2 as an in-memory database is as simple as adding a dependency. Other data sources can be configured via properties.
• BPMN process models, DMN decision tables, and Camunda Forms are automatically deployed for all configured locations.
• The Camunda process engine with its job executor is started automatically - but the job executor is disabled for tests by default.
• The process engine and related services, e.g. RuntimeService, RepositoryService, ..., are provided as lazy initialized beans and can be injected.
• Micronaut beans are resolved from the application context if they are referenced by expressions or Java class names within the process models.
• The process engine integrates with Micronaut transaction manager. Optionally, micronaut-data-jdbc or micronaut-data-jpa are supported.
• Eventing Bridge that maps Camunda Events to Micronaut ApplicationEvents.
• The process engine can be configured with generic properties.
• The Camunda REST API and the Webapps are supported (currently only for Jetty).
• The Camunda Enterprise Edition (EE) is supported.
• Process Engine Plugins are automatically activated on start.
• The job executor uses the Micronaut IO Executor's thread pools.
• The process engine configuration and the job executor configuration can be customized programmatically.
• A Camunda admin user is created if configured by properties and not present yet (including admin group and authorizations).
• Camunda's telemetry feature is automatically deactivated during test execution.

🚀Getting Started

This section describes what needs to be done to use micronaut-camunda-bpm-feature in a Micronaut project.

Create Project with Micronaut Launch

Create a new Micronaut project using Micronaut Launch and check that the "camunda-platform7" feature is selected.

This will take care of the following:

• If you don't explicitly select any database then an in-memory H2 will be included by default.
• The configuration file application.yml
  • enables the Webapps and the REST-API
  • is configured to create an admin user with credentials admin/admin with which you can login to http://localhost:8080/camunda
  • enables telemetry
• Jetty will be pre-configured (instead of Netty) to support the Webapps and REST-API by default

All you need to do is save a process model in the resources, see the following section.

Deploying Models

BPMN process models (*.bpmn), DMN decision tables (*.dmn), and Camunda Forms (*.form) should be created with the Camunda Modeler and saved in the resources.

By default, only the root of the resources will be scanned, but with the property camunda.locations you can configure the locations.

When starting the application you'll see the log output: Deploying model: classpath:xxxxxxx.bpmn

If you deploy Camunda Forms then you can reference these from your user tasks by either

• defining the form as type "Camunda Form" and setting the "Form Ref", e.g. ExampleForm (Preferred solution supported by Camunda Modeller 4.11 and newer).
• defining the form as type "Embedded or External Task Forms" and setting the "Form Key", e.g. camunda-forms:deployment:example.form (Supported by Camunda Modeller 4.10 and newer).

Camunda Integration

Process Engine and Services

Inject the process engine or any of the Camunda services using constructor injection:

import jakarta.inject.Singleton;
import org.camunda.bpm.engine.ProcessEngine;
import org.camunda.bpm.engine.RuntimeService;

@Singleton
public class MyComponent {

    private final ProcessEngine processEngine;
    private final RuntimeService runtimeService;

    public MyComponent(ProcessEngine processEngine, RuntimeService runtimeService) {
        this.processEngine = processEngine;
        this.runtimeService = runtimeService;
    }

    // ...
}

Alternatively to constructor injection, you can also use field injection, Java bean property injection, or method parameter injection.

You can then for example use the runtimeService to start new processes instances or correlate existing process instances.

Java Delegates

To invoke a Java delegate create a bean and reference it in your process model using an expression, e.g. ${loggerDelegate}:

import jakarta.inject.Singleton;
import org.camunda.bpm.engine.delegate.DelegateExecution;
import org.camunda.bpm.engine.delegate.JavaDelegate;

@Singleton
public class LoggerDelegate implements JavaDelegate {

    private static final Logger log = LoggerFactory.getLogger(LoggerDelegate.class);

    @Override
    public void execute(DelegateExecution delegateExecution) {
        log.info("Hello World: {}", delegateExecution);
    }
}

Internally, the bean will be resolved using io.micronaut.inject.qualifiers.Qualifiers.byName(...). Therefore, you can use the annotation jakarta.inject.Named to define an explicit bean name and use that name in your expression.

Configuration

Data Source

By default, an in-memory H2 data source is preconfigured. Remember to add the runtime dependency com.h2database:h2 mentioned in Dependency Management.

However, you can configure any other database supported by Camunda, e.g. in application.yml:

datasources:
  default:
    url: jdbc:postgresql://localhost:5432/postgres
    username: postgres
    password: secret
    driver-class-name: org.postgresql.Driver

after adding the appropriate driver as a dependency:

runtimeOnly("org.postgresql:postgresql:42.3.1")

Connection Pool with HikariCP

This integration uses HikariCP as a database connection pool to optimize performance. By default, the following configuration is applied:

• datasources.default.minimum-idle: 10
• datasources.default.maximum-pool-size: 50

Properties

You may use the following properties (typically in application.yml) to configure the Camunda integration.

Generic Properties

The process engine can be configured using generic properties listed in Camunda's Documentation: Configuration Properties.

The properties can be set in kebab case (lowercase and hyphen separated) or camel case (indicating the separation of words with a single capitalized letter as written in Camunda's documentation). Kebab case is preferred when setting properties.

Some of the most relevant properties are:

• database-schema-update (databaseSchemaUpdate)
• history
• initialize-telemetry (initializeTelemetry)
• telemetry-reporter-activate (telemetryReporterActivate)

Example:

camunda:
  generic-properties:
    properties:
      history: audit

Telemetry

Please consider activating Camunda's telemetry feature so that the Micronaut Camunda Integration appears in Camunda's statistics.

When starting on a fresh database use:

camunda:
  generic-properties:
    properties:
      initialize-telemetry: true

If you missed the activation on an already active database or want to check/change the setting, go to the admin application and navigate to System -> Telemetry Feature. Locally, jump directly to the Telemetry Feature.

Examples

Here are some example applications:

• Onboarding Process with service tasks, user tasks, and message correlation.
• Simple application with Java/Maven
• Simple application with Kotlin/Gradle
• Internal example application used during development

Supported JDKs

We officially support the following JDKs:

• JDK 8 (LTS) up to Camunda 7.19, i.e. with release v2.15.0
• JDK 11 (LTS)
• JDK 17 (LTS)

🏆Advanced Topics

Dependency Management

The Camunda integration works with both Gradle and Maven, but we recommend using Gradle because it has better Micronaut Support.

If you create a new project then simply use the feature features=camunda-platform7 as described in Getting Started section. However, you can also manage the dependencies yourself as described here:

Note: The module micronaut-camunda-bpm-feature includes the dependency org.camunda.bpm:camunda-engine which will be resolved transitively.

Camunda REST API and Webapps

Currently, the Camunda REST API and Webapps (Cockpit, Task list, and Admin) are only supported on the server runtime Jetty.

If you create your project with Micronaut Launch the jetty feature will be preselected for you.

However, if you have an existing project, you have to set the micronaut runtime of your project to jetty, e.g.

Configuration of REST API and Webapps

By default, REST API and the Webapps are not enabled. You have to configure them e.g. in the application.yaml as follows:

camunda:
  webapps:
    enabled: true
  rest:
    enabled: true

Further Information:

• The Webapps are by default available at /camunda. By default, / will redirect you there.
• The REST API is by default available at /engine-rest, e.g. to get the engine name use GET /engine-rest/engine.
• See Configuration Properties on how to enable basic authentication for REST, create a default user, or disable the redirect.
• Enabling the REST API or the Webapps impacts the startup time. Depending on your hardware it increases by around 500-1000 milliseconds.

Advanced Webapps Configuration

The security of the Webapps can be configured with the following properties:

Camunda Enterprise Edition (EE)

Add Maven Coordinates

To use the Camunda Enterprise Edition you have to add the Camunda Enterprise repository:

Replace CE with EE Dependencies

Then remove the CE dependencies and replace them with the EE ones. Here are some example snippets on how to do that. Keep in mind using the correct version of the libraries.

Configure your EE license

Camunda will use the license configured in the Cockpit.

If no license is registered, then the following locations will be checked during startup to register the license:

1. The URL referenced by the property camunda.license-file
2. The file camunda-license.txt in the resource's root if the property camunda.license-file has an empty value
3. The path .camunda/license.txt in the user's home directory

If you want to update your license key, use the Camunda Cockpit.

Process Engine Plugins

Every bean that implements the interface org.camunda.bpm.engine.impl.cfg.ProcessEnginePlugin is automatically added to the process engine's configuration on start.

You can either

• implement a bean factory with @io.micronaut.context.annotation.Factory and add one or more methods returning ProcessEnginePlugin instances and annotate each with a bean scope annotation
• annotate your class with @jakarta.inject.Singleton and implement the ProcessEnginePlugin interface

Example with the LDAP plugin:

implementation("org.camunda.bpm.identity:camunda-identity-ldap:7.21.0")

import io.micronaut.context.annotation.Factory;
import org.camunda.bpm.engine.impl.cfg.ProcessEnginePlugin;
import org.camunda.bpm.identity.impl.ldap.plugin.LdapIdentityProviderPlugin;
import jakarta.inject.Singleton;

@Factory
public class PluginConfiguration {

    @Singleton
    public ProcessEnginePlugin ldap() {
        // Using a public online LDAP:
        // https://www.forumsys.com/tutorials/integration-how-to/ldap/online-ldap-test-server/
        LdapIdentityProviderPlugin ldap = new LdapIdentityProviderPlugin();
        ldap.setServerUrl("ldap://ldap.forumsys.com:389");
        ldap.setManagerDn("cn=read-only-admin,dc=example,dc=com");
        ldap.setManagerPassword("password");
        ldap.setBaseDn("dc=example,dc=com");
        return ldap;
    }
}

You can now log in with "einstein" / "password". Note: the automatic creation of the admin user with the property camunda.admin-user conflicts with a read-only LDAP and must not be set!

Custom Process Engine Configuration

With the following bean it's possible to customize the process engine configuration:

import info.novatec.micronaut.camunda.bpm.feature.MnProcessEngineConfiguration;
import info.novatec.micronaut.camunda.bpm.feature.ProcessEngineConfigurationCustomizer;
import io.micronaut.context.annotation.Replaces;
import jakarta.inject.Singleton;

@Singleton
@Replaces(ProcessEngineConfigurationCustomizer.class)
public class MyProcessEngineConfigurationCustomizer implements ProcessEngineConfigurationCustomizer {
    @Override
    public void customize(MnProcessEngineConfiguration processEngineConfiguration) {
        processEngineConfiguration.setProcessEngineName("CustomizedEngine");
    }
}

Custom Job Executor Configuration

With the following bean it's possible to customize the job executor:

import info.novatec.micronaut.camunda.bpm.feature.JobExecutorCustomizer;
import info.novatec.micronaut.camunda.bpm.feature.MnJobExecutor;
import io.micronaut.context.annotation.Replaces;
import jakarta.inject.Singleton;

@Singleton
@Replaces(JobExecutorCustomizer.class)
public class MyJobExecutorCustomizer implements JobExecutorCustomizer {
    @Override
    public void customize(MnJobExecutor jobExecutor) {
        jobExecutor.setWaitTimeInMillis(300);
    }
}

Transaction management

By default the process engine integrates with Micronaut's transaction manager and uses a Hikari connection pool:

• When interacting with the process engine, e.g. starting or continuing a process, the existing transaction will be propagated.
• JavaDelegates and Listeners will have the surrounding Camunda transaction propagated to them allowing the atomic persistence of data.

Optionally, micronaut-data-jdbc or micronaut-data-jpa are supported.

Using micronaut-data-jdbc

To enable embedded transactions management support with micronaut-data-jdbc please add the following dependencies to your project:

and then configure the JDBC properties as described micronaut-sql documentation.

Using micronaut-data-jpa

To enable embedded transactions management support with micronaut-data-jpa please add the following dependencies to your project:

and then configure JPA as described in micronaut-sql documentation.

Performance

The Time to First Response (TTFR) is mainly influenced by the (slow) bootstrapping of the process engine - the bottleneck is the sequential parsing of over 50 MyBatis mappings. But the REST api and WebApps also take time.

This Micronaut Camunda Integration includes some optimizations that come into play especially in multi-core environments.

Some hints:

• Newer JDKs (JDK 17) are about 20% faster than older ones (JDK 8)
• More CPU cores are better to take advantage of parallelization during startup (so don't limit Docker to one or two CPUs).
• The selected vendor for the JDK has a small influence (Java SE Development Kit from Oracle is one of the faster ones)

Most discussions regarding startup time discuss the relevance during deployment or scaling up applications.

However, there is more: for a developer startup times are also relevant when test suites are executed - either locally or in a CI environment. If the application context is created often (e.g. @MockBean dirties the context in Spring Boot...) then integration tests run quite long.

If unit and integration tests are more of a conceptional difference (and not so much regarding performance) then the developer has more freedom of choice to decide if a unit or integration test is more appropriate for his scenario.

The documentation of Micronaut Test actually says: "One of the design goals of Micronaut was to eliminate the artificial separation imposed by traditional frameworks between function and unit tests due to slow startup times and memory consumption."

Architectural Design

Separating Process Engine from Webapps/REST

If you want to activate the Webapps and/or REST it might be an option to do this in a separate application. Both are connected via a common database.

Possible aspects:

• Your main process engine can run on Netty while the Webapps and/or REST run on Jetty (with a disabled Job Executor).
• You can scale these applications independently.

Embedding External Workers

If you're intending to use the External Task Pattern it might be an option embedding them (at first) in your main application - and having them to communicate via REST on localhost.

Possible aspects:

• Having a lot of separate applications for each external worker from the beginning increases the complexity. You can put all external workers in one application - or even in the main application which provides the process engine.
• If you decide later to extract single modules to a separate microservice then this is straight forward

Here is an example application: https://github.com/tobiasschaefer/micronaut-embedded-worker

Keycloak

You can enable the Keycloak integration to

• log into the Webapps (Tasklist, Cockpit, and Admin)
• get responses from the REST API with basic-auth enabled

1. Start Keycloak, e.g. docker run -p 8080:8080 -e KEYCLOAK_USER=admin -e KEYCLOAK_PASSWORD=admin -e DB_VENDOR="h2" quay.io/keycloak/keycloak:12.0.4
2. Configure Keycloak and add a test user.
3. Add the dependency implementation("org.camunda.bpm.extension:camunda-bpm-identity-keycloak:2.2.3") to your Micronaut project
4. Add the plugin:

   @Singleton
   @ConfigurationProperties("plugin.identity.keycloak")
   public class KeyCloakPlugin extends KeycloakIdentityProviderPlugin {
   }

5. Configure the application.yml

   plugin.identity.keycloak:
   keycloakIssuerUrl: http://localhost:8080/auth/realms/master
   keycloakAdminUrl: http://localhost:8080/auth/admin/realms/master
   clientId: camunda-identity-service
   clientSecret: 42aa42bb-1234-4242-a24a-42a2b420cde1 # you get this from keycloak
   useEmailAsCamundaUserId: true
   administratorGroupName: camunda-admin

6. Start the application and log in with your created test user. Keep in mind that your user needs an e-mail address.

Eventing Bridge

The Eventing Bridge maps Camunda Events to Micronaut ApplicationEvents. It's possible to configure three different event streams:

• Task: All events depending on UserTasks (UserTasks are Created, Assigned, Completed)
• Execution: All execution events (Activities are Started, Ended and Transitions are being taken)
• History: All history events

Configuration of Eventing Bridge

camunda:
  eventing:
    task: true
    execution: true
    history: true

Event Listener Implementation

To consume Micronaut ApplicationEvents you can implement the interface ApplicationEventListener or use the @EventListener annotation.

Process Tests

Process tests can easily be implemented with JUnit 5 by adding the camunda-bpm-assert library as a dependency:

and then implement the test using the usual @MicronautTest annotation:

import io.micronaut.test.extensions.junit5.annotation.MicronautTest;
import org.camunda.bpm.engine.ProcessEngine;
import org.camunda.bpm.engine.RuntimeService;
import org.camunda.bpm.engine.runtime.ProcessInstance;
import org.junit.jupiter.api.BeforeEach;
import org.junit.jupiter.api.Test;

import jakarta.inject.Inject;

import static org.camunda.bpm.engine.test.assertions.bpmn.BpmnAwareTests.*;

@MicronautTest
class HelloWorldProcessTest {

    @Inject
    ProcessEngine processEngine;

    @Inject
    RuntimeService runtimeService;

    @BeforeEach
    void setUp() {
        init(processEngine);
    }

    @Test
    void happyPath() {
        ProcessInstance processInstance = runtimeService.startProcessInstanceByKey("HelloWorld");
        assertThat(processInstance).isStarted();

        assertThat(processInstance).isWaitingAt("TimerEvent_Wait");
        execute(job());

        assertThat(processInstance).isEnded();
    }
}

Note: the integration automatically disables the job executor and the process engine's telemetry feature during test execution. This is deduced from the "test" profile.

See also a test in our example application: HelloWorldProcessTest

Docker

When using Gradle we recommend the Micronaut Application Plugin's dockerBuild task to create a layered Docker image.

Build the Docker image:

./gradlew dockerBuild

Run the Docker image:

docker run -p 8080:8080 <IMAGE>

Updating Camunda

Generally, follow Camunda's instructions regarding Update to the next Minor Version.

If you want to automate the database schema migration you can use Liquibase or Flyway together with the migration sql scripts provided by Camunda.

The following examples are based on Liquibase.

When starting on an empty database, e.g. when using H2 for tests:

<changeSet author="Tobias" id="1a" >
  <comment>Create common baseline Camunda 7.14 for H2 based on https://artifacts.camunda.com/artifactory/public/org/camunda/bpm/distro/camunda-sql-scripts/7.14.0/camunda-sql-scripts-7.14.0.zip in directory create</comment>
  <sqlFile path="camunda/h2_engine_7.14.0.sql" relativeToChangelogFile="true" dbms="h2" />
  <sqlFile path="camunda/h2_identity_7.14.0.sql" relativeToChangelogFile="true" dbms="h2" />
</changeSet>

If you already have a persistent database with the database schema of 7.14 which is not yet managed by Liquibase, e.g. PostgreSQL:

<changeSet author="Tobias" id="1b" >
  <comment>Create common baseline Camunda 7.14 for PostgreSQL (even if schema already exists) based on https://artifacts.camunda.com/artifactory/public/org/camunda/bpm/distro/camunda-sql-scripts/7.14.0/camunda-sql-scripts-7.14.0.zip in directory create</comment>
  <preConditions onFail="MARK_RAN">
    <not>
      <tableExists tableName="ACT_RU_JOB" />
    </not>
  </preConditions>
  <sqlFile path="camunda/postgres_engine_7.14.0.sql" relativeToChangelogFile="true" dbms="postgresql" />
  <sqlFile path="camunda/postgres_identity_7.14.0.sql" relativeToChangelogFile="true" dbms="postgresql" />
</changeSet>

When updating to a new Camunda version first apply all patch updates (if available) and then update to the next minor version:

<changeSet author="Tobias" id="2" >
  <comment>Update to Camunda 7.15 based on https://artifacts.camunda.com/artifactory/public/org/camunda/bpm/distro/camunda-sql-scripts/7.15.0/camunda-sql-scripts-7.15.0.zip in directory upgrade</comment>
  <!-- no patch files available for 7.14.x ... -->
  <sqlFile path="camunda/h2_engine_7.14_to_7.15.sql" relativeToChangelogFile="true" dbms="h2" />
  <sqlFile path="camunda/postgres_engine_7.14_to_7.15.sql" relativeToChangelogFile="true" dbms="postgresql" />
</changeSet>

Pitfalls

No version information in Fat/Uber/Shadow JAR

If you create a Fat/Uber/Shadow JAR and run that you will see a warning:

WARN i.n.m.c.b.f.MnProcessEngineConfiguration - The Camunda version cannot be determined. If you created a Fat/Uber/Shadow JAR then please consider using the Micronaut Application Plugin's 'dockerBuild' task to create a Docker image.

This is because the repackaging of the jars implicitly removes the META-INF information.

Missing version information leads to

• Detailed telemetry cannot be sent to Camunda because the version is mandatory
• EE license cannot be configured

Instead, of creating a Fat/Uber/Shadow JAR, please see instructions on creating a Docker image and use the resulting image to run a Docker container.

Executing Blocking Operations on Netty's I/O Thread Pool

When using the default server implementation Netty, blocking operations must be performed on I/O instead of Netty threads to avoid possible deadlocks. Therefore, as soon as Camunda "borrows a client thread" you have to make sure that the event loop is not blocked. A frequently occurring example is the implementation of a REST endpoint which interacts with the process engine. By default, Micronaut would use a Netty thread for this blocking operation. To prevent the use of a Netty thread it is recommended to use the annotation @ExecuteOn(TaskExecutors.IO). This will make sure that an I/O thread is used.

@Post("/hello-world-process")
@ExecuteOn(TaskExecutors.IO)
public String startHelloWorldProcess() {
    return runtimeService.startProcessInstanceByKey("HelloWorld").getId();
}

Camunda Platform Assertions - Multiple process tests

If you create multiple process tests, you need to add the following initialisation code in each test:

@Inject
ProcessEngine processEngine;

@BeforeEach
void setUp() {
    init(processEngine);
}

This makes the assertions aware of your process engine. Otherwise, it tries to reuse the engine of the test that got executed first and that may already be shut down, see Camunda Platform 7's Assert User Guide.

Here is a complete example: HelloWorldProcessTest.

📚Releases

The list of releases contains a detailed changelog.

We use Semantic Versioning.

The following compatibility matrix shows the officially supported Micronaut and Camunda versions for each release. Other combinations might also work but have not been tested.

Download of Releases:

• GitHub Artifacts
• Maven Central Artifacts

📆Publications

• 2021-11: Automating Processes with Camunda and Micronaut)
  Recording of the Novatec Summit by Tobias Schäfer (45 Minutes)
• 2021-10: Camunda Question Corner
  Recording of the Community Contribution Special with Niall Deehan and Tobias Schäfer (10 minutes)
• 2021-07: Automate any Process on Micronaut
  Blogpost by Tobias Schäfer
• 2021-02: Automating Processes with Microservices on Micronaut and Camunda
  Webinar by Tobias Schäfer, Bernd Rücker, and Sergio del Amo
• 2020-04: Micronaut meets Camunda BPM
  Blogpost by Tobias Schäfer

📨Contact

If you have any questions or ideas feel free to create an issue or contact us via GitHub Discussions.

We love listening to your feedback, and of course also discussing the project roadmap and possible use cases with you!

This open source project is being developed by envite consulting GmbH and Novatec Consulting GmbH with the support of the open source community.