Aitozi commented on a change in pull request #16894:
URL: https://github.com/apache/flink/pull/16894#discussion_r692873064
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/dispatcher/Dispatcher.java
##########
@@ -446,9 +446,13 @@ private CleanupJobState handleJobManagerRunnerResult(
JobManagerRunnerResult jobManagerRunnerResult, ExecutionType
executionType) {
if (jobManagerRunnerResult.isInitializationFailure()) {
if (executionType == ExecutionType.RECOVERY) {
- return jobManagerRunnerFailed(
-
jobManagerRunnerResult.getExecutionGraphInfo().getJobId(),
- jobManagerRunnerResult.getInitializationFailure());
+ if (jobManagerRunnerResult.isJobFinishing()) {
+ return CleanupJobState.GLOBAL;
Review comment:
I have implemented this by complete the jobManagerRunner resultFuture
with
```java
resultFuture.complete(
JobManagerRunnerResult.forSuccess(
new ExecutionGraphInfo(
jobMasterServiceProcessFactory.createArchivedExecutionGraph(
JobStatus.FAILED,
new JobAlreadyDoneException(getJobID())))));
```
But, I'm not sure is it right to set state to `JobStatus.FAILED`. We do not
know the status of job at the time of recovery of an already done job. I think
we can do these things like :
- Add a special `JobStatus.UNKNOWN` to handle this case
- Or when the FLIP for
[FLINK-11813](https://issues.apache.org/jira/browse/FLINK-11813) have been
introduced , we can extract the job result directly from `JobResultStore`.
Looking forward to your opinion.
--
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]
