[
https://issues.apache.org/jira/browse/FLINK-6217?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15953865#comment-15953865
]
ASF GitHub Bot commented on FLINK-6217:
---------------------------------------
Github user haohui commented on the issue:
https://github.com/apache/flink/pull/3648
The test failure seems unrelated. Cannot reproduce locally.
> ContaineredTaskManagerParameters sets off heap memory size incorrectly
> ----------------------------------------------------------------------
>
> Key: FLINK-6217
> URL: https://issues.apache.org/jira/browse/FLINK-6217
> Project: Flink
> Issue Type: Bug
> Components: YARN
> Reporter: Haohui Mai
> Assignee: Haohui Mai
>
> Thanks [~bill.liu8904] for triaging the issue.
> When {{taskmanager.memory.off-heap}} is disabled, we observed that the total
> memory that Flink allocates exceed the total memory of the container:
> For a 8G container the JobManager starts the container with the following
> parameter:
> {noformat}
> $JAVA_HOME/bin/java -Xms6072m -Xmx6072m -XX:MaxDirectMemorySize=6072m ...
> {noformat}
> The total amount of heap memory plus the off-heap memory exceeds the total
> amount of memory of the container. As a result YARN occasionally kills the
> container.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
