You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "ASF GitHub Bot (JIRA)" <ji...@apache.org> on 2017/07/05 13:35:00 UTC
[jira] [Commented] (FLINK-7082) Flip-6: Generic entry point for
Flink sessions
[ https://issues.apache.org/jira/browse/FLINK-7082?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16074757#comment-16074757 ]
ASF GitHub Bot commented on FLINK-7082:
---------------------------------------
GitHub user tillrohrmann opened a pull request:
https://github.com/apache/flink/pull/4261
[FLINK-7082] Add generic entry point for session and per-job clusters
This PR is based on #4260 and #4259.
This PR introduces a generic entry point `ClusterEntrypoint` for Flink clusters. `ClusterEntrypoint` performs common operations like parsing command line options, loading the Flink configuration and setting up common services (e.g. `RPCService`).
The two sub classes `JobClusterEntrypoint` and `SessionClusterEntrypoint` are the specialization for the per-job and session mode. The former entrypoint spawns a `ResourceManager`, tries to retrieve the `JobGraph` and then launches a `JobManagerRunner` with the retrieved `JobGraph`. The latter starts a `ResourceManager` and the `Dispatcher` component in order to receive job submissions.
The PR also includes a specialization of the `SessionClusterEntrypoint` which is the `StandaloneSessionClusterEntrypoint`. This class simply spawns the `StandaloneResourceManager` for a standalone session cluster.
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/tillrohrmann/flink addClusterEntrypoint
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/flink/pull/4261.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #4261
----
commit cba1cb11a3eaaea3511120fa7864028a2c47e6a5
Author: Till Rohrmann <tr...@apache.org>
Date: 2017-07-04T21:15:52Z
[FLINK-7103] [dispatcher] Add skeletal structure of Dispatcher component
The Dispatcher is responsible for receiving job submissions, persisting the JobGraphs,
spawning JobManager to execute the jobs and recovering the jobs in case of a master
failure. This commit adds the basic skeleton including the RPC call for job submission.
Add cleanup logic for finished jobs
Pass BlobService to JobManagerRunner
commit 476239fb52ef45926b0a28b97339158b8095bc3c
Author: Till Rohrmann <tr...@apache.org>
Date: 2017-07-04T15:47:37Z
[FLINK-7082] Add generic entry point for session and per-job clusters
commit 4c45f574d66cab076b7711696c0edbb485cb3cd4
Author: Till Rohrmann <tr...@apache.org>
Date: 2017-07-05T11:38:35Z
[FLINK-7105] Make ActorSystems non daemonic
In order to not having to explicitly wait on the termination of an ActorSystem
in the main thread, we now create the ActorSystems in non-daemonic mode. That
way the process won't terminate if there is still an active actor.
----
> Flip-6: Generic entry point for Flink sessions
> ----------------------------------------------
>
> Key: FLINK-7082
> URL: https://issues.apache.org/jira/browse/FLINK-7082
> Project: Flink
> Issue Type: Bug
> Components: Cluster Management
> Reporter: Till Rohrmann
> Assignee: Till Rohrmann
> Labels: flip-6
>
> Implement a generic entry point for Flink sessions. This {{ClusterEntryPoint}} has to start a {{ResourceManager}}, the {{Dispatcher}} component and the cluster's RESTful endpoint. This class could serve as the basis for a {{Mesos-}} and {{YarnEntryPoint}} to run Flink sessions.
> Maybe we can use a common base for the session and the per-job mode. The session has to start a dispatcher component and the per-job mode retrieves the {{JobGraph}} and directly starts a {{JobManager}} with this job.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)