[
https://issues.apache.org/jira/browse/FLINK-5791?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15882342#comment-15882342
]
ASF GitHub Bot commented on FLINK-5791:
---------------------------------------
Github user shuai-xu commented on a diff in the pull request:
https://github.com/apache/flink/pull/3304#discussion_r102909687
--- Diff:
flink-yarn/src/main/java/org/apache/flink/yarn/YarnResourceManager.java ---
@@ -245,7 +248,7 @@ public void onContainersAllocated(List<Container>
containers) {
try {
/** Context information used to start a
TaskExecutor Java process */
ContainerLaunchContext
taskExecutorLaunchContext =
-
createTaskExecutorLaunchContext(container.getResource(),
container.getId().toString(), container.getNodeId().getHost());
+
createTaskExecutorLaunchContext(container.getResource(),
container.getId().toString(), container.getNodeId().getHost(),
container.getPriority());
--- End diff --
Yes, this is a problem. But if we use the resource profile retrieved from
container, it may be different from the original request, they can not be
matched. And only when the resource requested is too small, yarn will return a
container bigger than requested to avoid resource fragmentation, or else it
will return what we requested.
> Resource should be strictly matched when allocating for yarn
> ------------------------------------------------------------
>
> Key: FLINK-5791
> URL: https://issues.apache.org/jira/browse/FLINK-5791
> Project: Flink
> Issue Type: Improvement
> Components: YARN
> Reporter: shuai.xu
> Assignee: shuai.xu
> Labels: flip-6
>
> In flip6, for yarn mode, resource should be assigned as requested to avoid
> resource wasting and OOM.
> 1. YarnResourceManager will request container according to ResourceProfile
> in slot request form JM.
> 2. RM will pass the ResourceProfile to TM for initializing its slots.
> 3. RM should match the slots offered by TM with SlotRequest from JM strictly.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
