XComp commented on a change in pull request #18189:
URL: https://github.com/apache/flink/pull/18189#discussion_r781224232
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/dispatcher/Dispatcher.java
##########
@@ -935,6 +942,19 @@ protected CleanupJobState
jobReachedTerminalState(ExecutionGraphInfo executionGr
archiveExecutionGraph(executionGraphInfo);
+ if (terminalJobStatus.isGloballyTerminalState()) {
+ try {
+ jobResultStore.createDirtyResult(
+
JobResult.createFrom(executionGraphInfo.getArchivedExecutionGraph()));
+ } catch (IOException e) {
+ log.error(
Review comment:
Initially, I thought that we wouldn't need more sophisticated error
handling here, because we're just falling back to the old behavior if an error
occurs while accessing the `JobResultStore`. But this reasoning is wrong. Not
marking a job as dirty in the `JobResultStore` will leave us in an inconsistent
state. I guess, we should fail fatally (similarly to when a job cannot be
recovered) in such a case to avoid rerunning the job. I'm gonna utilize the
`fatalErrorHandler`
--
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]
