Transaction Integration by Example

The following explains the SpringTransactionIntegrationTest of the core codebase step by step. Below is the Spring configuration file that we use in this example (you can find it in SpringTransactionIntegrationTest-context.xml). The section shown below contains the dataSource, transactionManager, processEngine and the process engine services.

When passing the DataSource to the SpringProcessEngineConfiguration (using property “dataSource”), the Camunda engine uses a org.springframework.jdbc.datasource.TransactionAwareDataSourceProxy internally, which wraps the passed DataSource. This is done to make sure the SQL connections retrieved from the DataSource and the Spring transactions play well together. This implies that it’s no longer needed to proxy the dataSource yourself in Spring configuration, although it’s still allowed to pass a TransactionAwareDataSourceProxy into the SpringProcessEngineConfiguration. In this case no additional wrapping will occur.

Make sure when declaring a TransactionAwareDataSourceProxy in Spring configuration yourself, that you don’t use it for resources that are already aware of Spring-transactions (e.g., DataSourceTransactionManager and JPATransactionManager need the un-proxied dataSource).

<beans xmlns="http://www.springframework.org/schema/beans"
       xmlns:context="http://www.springframework.org/schema/context"
       xmlns:tx="http://www.springframework.org/schema/tx"
       xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
       xsi:schemaLocation="http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans.xsd http://www.springframework.org/schema/context http://www.springframework.org/schema/context/spring-context-2.5.xsd http://www.springframework.org/schema/tx http://www.springframework.org/schema/tx/spring-tx-3.0.xsd">

  <bean id="dataSource" class="org.springframework.jdbc.datasource.SimpleDriverDataSource">
    <property name="driverClass" value="org.h2.Driver" />
    <property name="url" value="jdbc:h2:mem:camunda;DB_CLOSE_DELAY=1000" />
    <property name="username" value="sa" />
    <property name="password" value="" />
  </bean>

  <bean id="transactionManager" class="org.springframework.jdbc.datasource.DataSourceTransactionManager">
    <property name="dataSource" ref="dataSource" />
  </bean>

  <bean id="processEngineConfiguration" class="org.camunda.bpm.engine.spring.SpringProcessEngineConfiguration">
    <property name="dataSource" ref="dataSource" />
    <property name="transactionManager" ref="transactionManager" />
    <property name="databaseSchemaUpdate" value="true" />
    <property name="jobExecutorActivate" value="false" />
  </bean>

  <bean id="processEngine" class="org.camunda.bpm.engine.spring.ProcessEngineFactoryBean">
    <property name="processEngineConfiguration" ref="processEngineConfiguration" />
  </bean>

  <bean id="repositoryService" factory-bean="processEngine" factory-method="getRepositoryService" />
  <bean id="runtimeService" factory-bean="processEngine" factory-method="getRuntimeService" />
  <bean id="taskService" factory-bean="processEngine" factory-method="getTaskService" />
  <bean id="historyService" factory-bean="processEngine" factory-method="getHistoryService" />
  <bean id="managementService" factory-bean="processEngine" factory-method="getManagementService" />

  ...
</beans>

The remainder of that Spring configuration file contains the beans and configuration that we’ll use in this particular example:

<beans>
  ...
  <tx:annotation-driven transaction-manager="transactionManager"/>

  <bean id="userBean" class="org.camunda.bpm.engine.spring.test.UserBean">
    <property name="runtimeService" ref="runtimeService" />
  </bean>

  <bean id="printer" class="org.camunda.bpm.engine.spring.test.Printer" />

</beans>

First, the application context is created with any of the Spring ways to do that. In this example you could use a classpath XML resource to configure our Spring application context:

ClassPathXmlApplicationContext applicationContext =
    new ClassPathXmlApplicationContext("mytest/SpringTransactionIntegrationTest-context.xml");

or, since it is a test:

@ContextConfiguration("classpath:mytest/SpringTransactionIntegrationTest-context.xml")

Then we can get the service beans and invoke methods on them. The ProcessEngineFactoryBean will have added an extra interceptor to the services that applies Propagation.REQUIRED transaction semantics on the engine service methods. So, for example, we can use the repositoryService to deploy a process like this:

RepositoryService repositoryService = (RepositoryService) applicationContext.getBean("repositoryService");
String deploymentId = repositoryService
  .createDeployment()
  .addClasspathResource("mytest/hello.bpmn20")
  .addClasspathResource("mytest/hello.png")
  .deploy()
  .getId();

The other way around also works. In this case, the Spring transaction will be around the userBean.hello() method and the engine service method invocation will join that same transaction.

UserBean userBean = (UserBean) applicationContext.getBean("userBean");
userBean.hello();

The UserBean looks like this. Remember from above in the Spring bean configuration we injected the repositoryService into the userBean.

public class UserBean {

  // injected by Spring
  private RuntimeService runtimeService;

  @Transactional
  public void hello() {
    // here you can do transactional stuff in your domain model
    // and it will be combined in the same transaction as
    // the startProcessInstanceByKey to the RuntimeService
    runtimeService.startProcessInstanceByKey("helloProcess");
  }

  public void setRuntimeService(RuntimeService runtimeService) {
    this.runtimeService = runtimeService;
  }
}

Using Inner Spring Transactions

When engine API calls are executed in the Spring context, Spring doesn’t make it transparent to the Process Engine when nested API calls need to be executed in a separate transaction (the Propagation.REQUIRES_NEW transaction behavior). This doesn’t play well with how the Process Engine Context is used to separate transaction data described in detail in Transactions and the Process Engine Context. Due to this, if the inner transaction fails, the changes made by the engine API calls are still committed to the Camunda database.

A solution for this is to explicitly declare to the Process Engine that a new Process Engine Context is to be created, where all the following Engine API calls will store their changes for the database. In case the inner transaction fails, the changes in this new Process Engine Context will be reverted.

The Process Engine Context must be declared whenever a Spring Propagation.REQUIRES_NEW inner transaction is defined in an already running transaction.

Example

In the following code-snippet, we can see a Spring Propagation.REQUIRED transaction, defined on the execute method, and a Spring Propagation.REQUIRES_NEW transaction, defined on the InnerProcessServiceImpl#startInnerProcess method. The InnerProcessServiceImpl#startInnerProcess method is called through the execute method, resulting in an inner transaction.We can also assume that, in the startProcessInstanceByKey engine API call, some custom code continues to execute in the inner transaction, and throws an exception.

@Transactional(value = "transactionManager", propagation = Propagation.REQUIRED)
public void execute(DelegateExecution execution) throws Exception {
  try {
    innerProcessService.startInnerProcess(execution);
  } catch (Exception ex) {
    // noop
  }
}

/* InnerProcessService implementation */

@Service
public class InnerProcessServiceImpl implements InnerProcessService {

  @Override
  @Transactional(value = "transactionManager", propagation = Propagation.REQUIRES_NEW, rollbackFor = {Throwable.class})
  public void startInnerProcess(DelegateExecution execution) {

    execution.getProcessEngineServices()
      .getRuntimeService()
      .startProcessInstanceByKey("EXAMPLE_PROCESS");

    // custom code continues
  }
}

In a case like this one, since Spring doesn’t make the Process Engine aware that the startProcessInstanceByKey engine API call is executed in a new, inner transaction, when the custom code fails, the Spring inner transaction will be rolled back, but the data from the started Process Instance data will be committed with the outer transaction.

The described problem can be solved through the static methods of the org.camunda.bpm.engine.context.ProcessEngineContext class. First, we call the ProcessEngineContext.requiresNew() method to declare that a new Context is required. Finally, we call the ProcessEngineContext.clear() method to clear the declaration for a new Process Engine Context. This is important since a new Process Engine Context will be declared for each following engine API call until the ProcessEngineContext#clear method is called.

Using a try-finally block is recommended to ensure that the ProcessEngineContext static methods are called and cleared even in the presence of exceptions. Below, you can find the solution applied to the startProcessInstanceByKey engine API call of the example above:

try {
  
  // declare new Process Engine Context
  ProcessEngineContext.requiresNew();
  
  // call engine APIs
  execution.getProcessEngineServices()
    .getRuntimeService()
    .startProcessInstanceByKey("EXAMPLE_PROCESS");

} finally {
  // clear declaration for new Process Engine Context
  ProcessEngineContext.clear();
}