[
https://issues.apache.org/jira/browse/FLINK-7668?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16182428#comment-16182428
]
ASF GitHub Bot commented on FLINK-7668:
---------------------------------------
Github user tillrohrmann commented on a diff in the pull request:
https://github.com/apache/flink/pull/4728#discussion_r141319310
--- Diff:
flink-runtime/src/main/java/org/apache/flink/runtime/webmonitor/RestfulGateway.java
---
@@ -38,5 +44,36 @@
* @param timeout for this operation
* @return Future REST endpoint address
*/
- CompletableFuture<String> requestRestAddress(Time timeout);
+ CompletableFuture<String> requestRestAddress(@RpcTimeout Time timeout);
+
+ /**
+ * Requests the AccessExecutionGraph for the given jobId. If there is
no such graph, then
+ * the future is completed with a {@link JobNotFoundException}.
+ *
+ * @param jobId identifying the job whose AccessExecutionGraph is
requested
+ * @param timeout for the asynchronous operation
+ * @return Future containing the AccessExecutionGraph for the given
jobId, otherwise {@link JobNotFoundException}
+ */
+ CompletableFuture<AccessExecutionGraph> requestJob(JobID jobId,
@RpcTimeout Time timeout);
--- End diff --
It makes sense to accumulate (all) methods required for the REST handlers
in a common interface because it will be implemented by the `Dispatcher` as
well as the `JobMaster`. Right now, the `RestfulInterface` is used two
purposes: 1. retrieving the REST address and 2. all REST handler related
requests. Once all information which is required by the REST handlers is
serializable, we no longer have to redirect. Then we can drop the
`requestRestAddress` method. Then I would also rename this interface into
something like `RestInformationGateway`. Until then, I would like to keep it
how it is.
> Add AccessExecutionGraph refresh interval to ExecutionGraphHolder
> -----------------------------------------------------------------
>
> Key: FLINK-7668
> URL: https://issues.apache.org/jira/browse/FLINK-7668
> Project: Flink
> Issue Type: Sub-task
> Components: REST
> Affects Versions: 1.4.0
> Reporter: Till Rohrmann
> Assignee: Till Rohrmann
> Labels: flip-6
>
> Once we support offline {{AccessExecutionGraph}} implementation (see
> FLINK-7667) we should add a refresh interval to the {{ExecutionGraphHolder}}
> after which the {{AccessExecutionGraph}} is retrieved again from the
> {{JobMaster}}.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
