[
https://issues.apache.org/jira/browse/FLINK-3000?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15005699#comment-15005699
]
ASF GitHub Bot commented on FLINK-3000:
---------------------------------------
GitHub user sachingoel0101 opened a pull request:
https://github.com/apache/flink/pull/1354
[FLINK-3000]Adds shutdown hook to clean up lingering yarn sessions
Adds a shutdown hook to cleanup the submitted yarn application while the
user cancels it in any but a `RUNNING` state. If running, there are two cases:
Detached and Non-detached. Non detached modes will connect to the cluster and
add a shut down hook, while detached ones, well, since they're detached, once
deployed, users have to cancel them explicitly.
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/sachingoel0101/flink yarn_cli
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/flink/pull/1354.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 #1354
----
commit e1b8778491bf8fd592bddc8fe157d8aa8c150b7f
Author: Sachin Goel <[email protected]>
Date: 2015-11-15T00:41:30Z
Adds shutdown hook to clean up lingering yarn sessions
----
> Add ShutdownHook to YARN CLI to prevent lingering sessions
> ----------------------------------------------------------
>
> Key: FLINK-3000
> URL: https://issues.apache.org/jira/browse/FLINK-3000
> Project: Flink
> Issue Type: Bug
> Components: YARN Client
> Affects Versions: 0.10
> Reporter: Ufuk Celebi
> Assignee: Sachin Goel
> Fix For: 0.10.1
>
>
> Submitting a job via
> {code}
> bin/flink run -m yarn-cluster ...
> {code}
> and terminating the client can lead to lingering YARN sessions allocating
> cluster resources.
> This was reported by a user.
> {code}
> 1) One starts a flink job in the yarn mode
> 2) He sees that containers are not allocated since cluster is busy
> 3) Presses Ctrl+C
> 4) An “empty” flink session remains in the cluster although the Flink didn’t
> print that “you can track your application on the X URL”
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
