[
https://issues.apache.org/jira/browse/FLINK-8328?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16320216#comment-16320216
]
ASF GitHub Bot commented on FLINK-8328:
---------------------------------------
Github user tillrohrmann commented on a diff in the pull request:
https://github.com/apache/flink/pull/5215#discussion_r160674427
--- Diff:
flink-yarn/src/main/java/org/apache/flink/yarn/cli/FlinkYarnSessionCli.java ---
@@ -660,7 +570,25 @@ public int run(
"yarn application -kill " +
applicationId.getOpt());
yarnCluster.disconnect();
} else {
- runInteractiveCli(yarnCluster, true);
+ ScheduledThreadPoolExecutor
scheduledExecutorService = new ScheduledThreadPoolExecutor(1);
+
+ try (YarnApplicationStatusMonitor
yarnApplicationStatusMonitor = new YarnApplicationStatusMonitor(
+ yarnDescriptor.getYarnClient(),
+ yarnCluster.getApplicationId(),
+ new
ScheduledExecutorServiceAdapter(scheduledExecutorService))) {
--- End diff --
The `ScheduledExecutor` gives a better abstraction because it does not
expose service control methods like shutdown to the callee. I think the Java
abstraction is slightly broken in this regard.
> Pull Yarn ApplicationStatus polling out of YarnClusterClient
> ------------------------------------------------------------
>
> Key: FLINK-8328
> URL: https://issues.apache.org/jira/browse/FLINK-8328
> Project: Flink
> Issue Type: Sub-task
> Components: Client
> Affects Versions: 1.5.0
> Reporter: Till Rohrmann
> Assignee: Till Rohrmann
> Labels: flip-6
> Fix For: 1.5.0
>
>
> In order to make the {{FlinkYarnSessionCli}} work with Flip-6, we have to
> pull the Yarn {{ApplicationStatus}} polling out of the {{YarnClusterClient}}.
> I propose to introduce a dedicated {{YarnApplicationStatusMonitor}}. This has
> also the benefit of separating concerns better.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
