[
https://issues.apache.org/jira/browse/FLINK-4851?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15597489#comment-15597489
]
ASF GitHub Bot commented on FLINK-4851:
---------------------------------------
Github user tillrohrmann commented on a diff in the pull request:
https://github.com/apache/flink/pull/2655#discussion_r84573260
--- Diff:
flink-runtime/src/test/java/org/apache/flink/runtime/resourcemanager/ResourceManagerJobMasterTest.java
---
@@ -145,6 +171,10 @@ public void
testRegisterJobMasterWithFailureLeaderListener() throws Exception {
Future<RegistrationResponse> declineFuture =
resourceManager.registerJobMaster(rmLeaderSessionId, jmLeaderSessionId,
jobMasterAddress, unknownJobIDToHAServices);
RegistrationResponse response = declineFuture.get(5,
TimeUnit.SECONDS);
assertTrue(response instanceof RegistrationResponse.Decline);
+
+ if (testingFatalErrorHandler.hasExceptionOccurred()) {
+ testingFatalErrorHandler.rethrowError();
+ }
--- End diff --
Yes I think this makes sense for all of the distributed components RM, JM
and TM. It would be good to instantiate default components which you can
replace with a special version of a component in case you need it for a test.
> Add FatalErrorHandler and MetricRegistry to ResourceManager
> -----------------------------------------------------------
>
> Key: FLINK-4851
> URL: https://issues.apache.org/jira/browse/FLINK-4851
> Project: Flink
> Issue Type: Sub-task
> Components: ResourceManager
> Reporter: Till Rohrmann
> Assignee: Till Rohrmann
>
> The {{ResourceManager}} currently does not contain a {{FatalErrorHandler}}.
> In order to harmonize the fatal error handling across all components, we
> should introduce a {{FatalErrorHandler}}, which handles fatal errors.
> Additionally, we should also give a {{MetricRegistry}} to the
> {{ResourceManager}} so that it can report metrics.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
