tillrohrmann commented on a change in pull request #9832: [FLINK-11843] Bind
lifespan of Dispatcher to leader session
URL: https://github.com/apache/flink/pull/9832#discussion_r337431357
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/dispatcher/Dispatcher.java
##########
@@ -197,19 +191,37 @@ public void onStart() throws Exception {
onFatalError(exception);
throw exception;
}
+
+ startRecoveredJobs();
Review comment:
I would like to not change this behaviour. In the long run I would rather
try to get rid of the `JobGraphListener` and the `onAddedJobGraph` and
`onRemovedJobGraph` calls. Ideally, the leader recovers what it can recover at
the beginning and then no more `JobGraphs` from the `JobGraphStore` will be
accepted.
Yes, `submitJob` works for recovered jobs because the current
`JobGraphWriter` implementations would simply overwrite the old value. Instead
of introducing a new RPC method for the submission of recovered jobs, I would
like to change this only if needed in the future. Moreover, I hope that we can
get rid of the `JobGraphListener` interface even faster. Then this would not be
needed.
----------------------------------------------------------------
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:
us...@infra.apache.org
With regards,
Apache Git Services
