reele commented on code in PR #11677:
URL: https://github.com/apache/dolphinscheduler/pull/11677#discussion_r965682277
##########
dolphinscheduler-dao/src/main/java/org/apache/dolphinscheduler/dao/mapper/TaskInstanceMapper.java:
##########
@@ -81,7 +81,7 @@ IPage<TaskInstance>
queryTaskInstanceListPaging(IPage<TaskInstance> page,
* @param endTime endTime
* @return task instance
*/
- TaskInstance queryLastTaskInstance(@Param("taskCode") long taskCode,
@Param("startTime") Date startTime, @Param("endTime") Date endTime);
+ TaskInstance queryLastTaskInstance(@Param("taskCode") long taskCode,
@Param("processInstanceId") long processInstanceId);
Review Comment:
because taskInstance don't keep the `scheduleTime`, but processInstance do,
if only use startTime, and processInstance start with a scheduleTime, the
dependent node will use `scheduleTime` to calculate a `bizdate` to find the
task, it may can't detect the task in original `bizdate`.
the actual situation I have encountered is:
etl task `B` processes a zipper table, it must run continuously by date,
it's previews dependent task `D` depends on `A[T]` and `B[T-1](self
yesterday)`, in some reason, `A` blocked until the `next day`, so `B.startTime`
is `[T+1]`, and next day `D` will use `bizdate[T]` calculated by
`scheduleTime[T+1]` to finding `B[T]`, there is no `B` started on time `[T]`,
so the dependent task `D[T+1]` will never stop, every time this happens, I need
to manually stop the processInstance, and set the state of `D` to `Forced
Success`, and recover the process.
In fact taskInstance can only run attached to a processInstance, so I think
it is right to find taskInstance through processInstance, and in 2.0.5, it did
in the same way.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
