[
https://issues.apache.org/jira/browse/FLINK-9953?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16956023#comment-16956023
]
Till Rohrmann commented on FLINK-9953:
--------------------------------------
I would suggest to leave the client side part of the per-job mode out of the
first MVP because the {{CliFrontend}} is currently a moving target. Maybe some
of the assumptions (e.g. that the cli supports a two phase job submission
protocol for the per-job mode) might no longer hold true after the client
refactoring is done. I guess [~tison] and [~kkloudas] might tell us more about
the plans for the per-job mode.
Until this part if figured out on the client side, one could concentrate on
getting a K8s session cluster up and running. This means that one needs to
deploy the K8s session cluster and can then use the cli frontend to submit a
job: {{bin/flink run -m 111.222.333.444:1234 jobJar.jar}}. I guess that there
are enough open questions for this part alone.
For the overall approach, I agree with [~felixzheng] to do the implementation
in incremental steps. For example, in the beginning one could say we don't do
the individual pod lifecycle management and instead simply have a deployment
representing our TMs. That way we would narrow the scope of the initial
implementation considerably. Based on other needs, one could then improve the
implementation step by step. For example, at the moment I don't yet fully
understand why we need to persist RM state. I guess these kind of things should
be discussed next but they don't need to necessarily block the development of
the initial very basic K8s integration.
> Active Kubernetes integration
> -----------------------------
>
> Key: FLINK-9953
> URL: https://issues.apache.org/jira/browse/FLINK-9953
> Project: Flink
> Issue Type: New Feature
> Components: Runtime / Coordination
> Reporter: Till Rohrmann
> Assignee: Yang Wang
> Priority: Major
> Labels: pull-request-available
> Time Spent: 10m
> Remaining Estimate: 0h
>
> This is the umbrella issue tracking Flink's active Kubernetes integration.
> Active means in this context that the {{ResourceManager}} can talk to
> Kubernetes to launch new pods similar to Flink's Yarn and Mesos integration.
> Phase1 implementation will have complete functions to make flink running on
> kubernetes. Phrase2 is mainly focused on production optimization, including
> k8s native high-availability, storage, network, log collector and etc.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
