[Bug] [api] When forceTaskSuccess is performed on the last unsuccessful workflow instance, the workflow instance cannot be reset to the Success state.

[fanzhongyu]

Search before asking

• I had searched in the issues and found no similar issues.

What happened

通过 public void forceTaskSuccess(User loginUser, long projectCode, Integer taskInstanceId)方法对工作节点实例执行强制成功操作时，如果该节点是工作流中唯一失败/kill 的节点，工作流实例应被自动标记为
SUCCESS。但实际上工作流实例永远不会被提升为 SUCCESS。

例子：
我现在有两个工作节点实例，一个成功，一个失败

现在我将这个失败的工作节点实例强制成功 按照processService.forceWorkflowInstanceSuccessByTaskInstanceId(task);的逻辑，这里应该是会将对应的工作流实例置为成功。

但是该工作流实例状态并没有更新。

What you expected to happen

这里工作流实例状态成功更新。

How to reproduce

在forceTaskSuccess方法中，任务实例状态先被更新为 FORCED_SUCCESS 并持久化到数据库，然后才调用 processService.forceWorkflowInstanceSuccessByTaskInstanceId(task);

而在 processService.forceWorkflowInstanceSuccessByTaskInstanceId内部，会重新从数据库查询任务实例列表：
List validTaskList = taskInstanceDao.queryValidTaskListByWorkflowInstanceId(workflowInstance.getId());

此时该任务实例在数据库中的状态已经是 FORCED_SUCCESS（同一事务内自己的写操作，应该立马能读到），不再满足 isFailure() || isKill() 的过滤条件：

List failTaskList = validTaskList.stream()
.filter(instance -> instance.getState().isFailure() || instance.getState().isKill())
.map(TaskInstance::getId).collect(Collectors.toList());
if (failTaskList.size() == 1 && failTaskList.contains(task.getId())) {
// 永远无法进入 — failTaskList 中不包含已被强制成功的任务
}

由于被强制成功的任务不再是 failure/kill 状态，failTaskList 的情况为：

• 如果它是唯一的失败节点 → failTaskList 为空 → 条件不满足
• 如果还有其他失败节点 → failTaskList 中不包含当前 task → failTaskList.contains(task.getId()) 返回 false

因此，工作流实例永远无法被标记为 SUCCESS。

修复方案：
调换调用顺序：在更新任务实例状态之前先调用 forceWorkflowInstanceSuccessByTaskInstanceId()，使其内部查询时能读到原始的 failure/kill 状态：
processService.forceWorkflowInstanceSuccessByTaskInstanceId(task);
// change the state of the task instance
task.setState(TaskExecutionStatus.FORCED_SUCCESS);
task.setEndTime(new Date());
int changedNum = taskInstanceMapper.updateById(task);
if (changedNum <= 0) {
throw new ServiceException(Status.FORCE_TASK_SUCCESS_ERROR);
}

Anything else

No response

Version

dev

Are you willing to submit PR?

• Yes I am willing to submit a PR!

Code of Conduct

• I agree to follow this project's Code of Conduct

[github-actions]

Search before asking

• I had searched in the issues and found no similar issues.

What happened

When performing a forced success operation on a worker node instance through the public void forceTaskSuccess(User loginUser, long projectCode, Integer taskInstanceId) method, if the node is the only failed/kill node in the workflow, the workflow instance should be automatically marked as
SUCCESS. But in reality the workflow instance is never promoted to SUCCESS.

Example:
I now have two worker node instances, one successful and one failed

Now I force this failed worker node instance to succeed According to the logic of processService.forceWorkflowInstanceSuccessByTaskInstanceId(task);, the corresponding workflow instance should be set to success.

But the workflow instance status is not updated.

What you expected to happen

The status of this workflow instance was successfully updated.

How to reproduce

In the forceTaskSuccess method, the task instance status is first updated to FORCED_SUCCESS and persisted to the database, and then processService.forceWorkflowInstanceSuccessByTaskInstanceId(task); is called.

Within processService.forceWorkflowInstanceSuccessByTaskInstanceId, the task instance list will be re-queried from the database:
List validTaskList = taskInstanceDao.queryValidTaskListByWorkflowInstanceId(workflowInstance.getId());

At this time, the status of the task instance in the database is FORCED_SUCCESS (its own write operations in the same transaction should be read immediately), and it no longer meets the filtering conditions of isFailure() || isKill():

List failTaskList = validTaskList.stream()
.filter(instance -> instance.getState().isFailure() || instance.getState().isKill())
.map(TaskInstance::getId).collect(Collectors.toList());
if (failTaskList.size() == 1 && failTaskList.contains(task.getId())) {
// Never able to enter - failTaskList does not contain tasks that have been forced to succeed
}

Since the task that is forced to succeed is no longer in the failure/kill state, the situation of failTaskList is:

• If it is the only failed node → failTaskList is empty → condition is not met
• If there are other failed nodes → failTaskList does not contain the current task → failTaskList.contains(task.getId()) returns false

Therefore, a workflow instance can never be marked SUCCESS.

Fix:
Change the calling order: call forceWorkflowInstanceSuccessByTaskInstanceId() before updating the task instance status, so that the original failure/kill status can be read during internal query:
processService.forceWorkflowInstanceSuccessByTaskInstanceId(task);
// change the state of the task instance
task.setState(TaskExecutionStatus.FORCED_SUCCESS);
task.setEndTime(new Date());
int changedNum = taskInstanceMapper.updateById(task);
if (changedNum <= 0) {
throw new ServiceException(Status.FORCE_TASK_SUCCESS_ERROR);
}

Anything else

No response

Version

dev

Are you willing to submit PR?

• Yes I am willing to submit a PR!

Code of Conduct

• I agree to follow this project's Code of Conduct

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity for 30 days. It will be closed in next 7 days if no further activity occurs.