zhuzhurk commented on a change in pull request #9663:
[WIP][FLINK-12433][runtime] Implement DefaultScheduler stub
URL: https://github.com/apache/flink/pull/9663#discussion_r327421441
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/executiongraph/ExecutionGraph.java
##########
@@ -1129,6 +1157,9 @@ void failGlobalIfExecutionIsStillRunning(Throwable
cause, ExecutionAttemptID fai
* @param t The exception that caused the failure.
*/
public void failGlobal(Throwable t) {
+ if (!isLegacyScheduling()) {
+ ExceptionUtils.rethrow(t);
Review comment:
> For real consistency issues caused by bugs, illegal state transitions,
etc. I think it is reasonable to terminate the JVM.
I think it's reasonable but I'm not sure whether this change can cause much
more JM failovers than it used to be. This could be an even severe problem for
session cluster which holds multiple jobs in a Dispatcher.
> One could propagate the exception to the JobMaster and in the RPC
framework check for unwanted uncaught exceptions in specially annotated RPC
methods.
If we'd like to terminate the JVM, why not do it right in the place the
fatal error happens?
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
