[
https://issues.apache.org/jira/browse/FLINK-15288?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16998763#comment-16998763
]
Yang Wang commented on FLINK-15288:
-----------------------------------
Hi Andrey, thanks for your comments.
The all cli options of kubernetes has been remove. For the long term, all the
cli options will be replaced with config options. So `kubernetes-session.sh`
currently will not provide any cli options. More information could be found at
[FLINK-15179|https://issues.apache.org/jira/browse/FLINK-15179].
This ticket will not introduce new config options for cutoff, we will reuse the
\{{ResourceManagerOptions#CONTAINERIZED_HEAP_CUTOFF_RATIO}}. And we will keep
the same behavior with Yarn/Mesos. I think this issue should be fixed in 1.10.
Because if we set the pod resource limit equal as jobmanager heap size, the
jobmanager pod will be killed by kubernetes of OOM when the jvm heap usage goes
up. For the new jobmanager memory configuration of FLIP-49, the change of this
ticket will not make too much differences since we keep the same behavior with
Yarn/Mesos.
> Starting jobmanager pod should respect containerized.heap-cutoff
> ----------------------------------------------------------------
>
> Key: FLINK-15288
> URL: https://issues.apache.org/jira/browse/FLINK-15288
> Project: Flink
> Issue Type: Sub-task
> Components: Deployment / Kubernetes
> Reporter: Yang Wang
> Assignee: Yang Wang
> Priority: Blocker
> Fix For: 1.10.0
>
>
> Starting jobmanager pod should respect containerized.heap-cutoff. The cutoff
> will be used to leave some memory for jvm off-heap, for example meta space,
> thread native memory and etc.
>
> {code:java}
> public static final ConfigOption<Float> CONTAINERIZED_HEAP_CUTOFF_RATIO =
> ConfigOptions
> .key("containerized.heap-cutoff-ratio")
> .defaultValue(0.25f)
> .withDeprecatedKeys("yarn.heap-cutoff-ratio")
> .withDescription("Percentage of heap space to remove from containers (YARN /
> Mesos / Kubernetes), to compensate" +
> " for other JVM memory usage.");
> {code}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
