Camunda并行多实例子流程执行报错java.util.NoSuchElementException求助

环境信息 #

• Camunda BPMN Engine版本：7.17.6-ee
• Camunda Modeler版本：4.6.0

流程结构 #

• 主流程包含并行多实例子流程，子流程内包含任务A和顺序多实例任务B（任务B处于事务边界内）
• 子流程完成后进入循环逻辑，持续处理下一批未结预订，直至无剩余待处理预订

错误详情 #

运行流程时，任务B执行阶段触发NoSuchElementException，异常栈如下：

ERROR [org.camunda.bpm.engine.context] (pool-15-thread-13) ENGINE-16004 Exception while closing command context: null: java.util.NoSuchElementException
at java.base/java.util.ArrayList$Itr.next(ArrayList.java:1000)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.bpmn.behavior.MultiInstanceActivityBehavior.getElementAtIndex(MultiInstanceActivityBehavior.java:129)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.bpmn.behavior.MultiInstanceActivityBehavior.evaluateCollectionVariable(MultiInstanceActivityBehavior.java:93)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.bpmn.behavior.MultiInstanceActivityBehavior.performInstance(MultiInstanceActivityBehavior.java:78)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.bpmn.behavior.SequentialMultiInstanceActivityBehavior.complete(SequentialMultiInstanceActivityBehavior.java:58)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.pvm.runtime.operation.PvmAtomicOperationActivityEnd.execute(PvmAtomicOperationActivityEnd.java:105)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.pvm.runtime.operation.PvmAtomicOperationActivityEnd.execute(PvmAtomicOperationActivityEnd.java:35)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.interceptor.AtomicOperationInvocation.execute(AtomicOperationInvocation.java:99)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.interceptor.CommandInvocationContext.invokeNext(CommandInvocationContext.java:131)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.interceptor.CommandInvocationContext.performNext(CommandInvocationContext.java:111)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.interceptor.CommandInvocationContext.performOperation(CommandInvocationContext.java:86)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.persistence.entity.ExecutionEntity.performOperation(ExecutionEntity.java:623)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.persistence.entity.ExecutionEntity.performOperation(ExecutionEntity.java:597)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.pvm.runtime.operation.PvmAtomicOperationActivityNotifyListenerEnd$1.callback(PvmAtomicOperationActivityNotifyListenerEnd.java:47)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.pvm.runtime.operation.PvmAtomicOperationActivityNotifyListenerEnd$1.callback(PvmAtomicOperationActivityNotifyListenerEnd.java:41)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.pvm.runtime.PvmExecutionImpl.continueExecutionIfNotCanceled(PvmExecutionImpl.java:2078)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.pvm.runtime.PvmExecutionImpl.dispatchDelayedEventsAndPerformOperation(PvmExecutionImpl.java:2027)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.pvm.runtime.operation.PvmAtomicOperationActivityNotifyListenerEnd.eventNotificationsCompleted(PvmAtomicOperationActivityNotifyListenerEnd.java:41)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.pvm.runtime.operation.PvmAtomicOperationActivityNotifyListenerEnd.eventNotificationsCompleted(PvmAtomicOperationActivityNotifyListenerEnd.java:27)
at org.camunda.bpm.camunda-engine@7.17.0//org.camunda.bpm.engine.impl.core.operation.AbstractEventAtomicOperation.execute(AbstractEventAtomicOperation.java:66)
...

异常代码定位 #

异常触发于Camunda的MultiInstanceActivityBehavior类getElementAtIndex方法，由it.next()抛出，代码如下：

protected Object getElementAtIndex(int i, Collection<?> collection) {
     Object value = null;
     int index = 0;
     Iterator<?> it = collection.iterator();
     while (index <= i) {
      value = it.next();
      index++;
     }
    return value;
 }

已尝试使用线程安全集合，问题仍未解决。

排查方向 #

1. 多实例集合的动态修改 #

即使使用线程安全集合，若顺序多实例执行期间集合元素被移除/长度缩短，会导致Camunda记录的实例总数与实际集合长度不匹配。当执行到超过新集合长度的索引时，迭代器就会抛出NoSuchElementException。

• 检查任务A、任务B的执行逻辑，以及流程变量监听器、外部服务是否在修改任务B绑定的多实例集合变量。

2. 多实例参数配置错误 #

确认任务B的多实例配置是否存在不一致：

• 检查集合变量是否正确绑定，是否误引用了其他动态变化的集合
• 若手动设置了nrOfInstances（实例总数），需确认该值与集合实际长度是否匹配。若手动指定的实例数大于集合实际长度，必然触发索引越界
• 检查实例数的计算表达式，确保表达式结果与集合实际长度同步

3. 事务边界的变量刷新问题 #

任务B处于事务边界内，事务提交延迟可能导致流程变量的集合状态未及时刷新：

• 比如任务A修改了集合，但事务未提交时任务B已读取到旧状态，后续事务提交后集合长度缩短，导致后续实例执行时索引越界
• 检查事务边界内的变量操作逻辑，确保变量修改后事务及时提交，避免状态不一致

4. 依赖版本一致性问题 #

异常栈显示使用的是camunda-engine@7.17.0，但实际环境是7.17.6-ee，可能存在依赖冲突：

• 检查项目依赖，确保所有Camunda相关组件版本统一为7.17.6-ee，排除版本不一致导致的逻辑异常

5. 循环流程的变量重置逻辑 #

每次循环处理新一批预订时，需确认任务B的多实例集合变量是否正确重置：

• 若上一次循环的集合残留，或新集合初始化不完整，会导致索引越界
• 检查循环触发时的变量初始化代码，确保每次循环的多实例集合都是全新、完整的实例

内容的提问来源于stack exchange，提问作者hakson