[
https://issues.apache.org/jira/browse/FLINK-7805?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16380212#comment-16380212
]
ASF GitHub Bot commented on FLINK-7805:
---------------------------------------
GitHub user GJL opened a pull request:
https://github.com/apache/flink/pull/5597
[FLINK-7805][flip6] Add HA capabilities to YarnResourceManager
## What is the purpose of the change
*Recover previously running containers after a restart of the
ApplicationMaster.
This is a port of a feature that was already implemented prior to FLIP-6.*
cc: @tillrohrmann
## Brief change log
- *Extract `RegisterApplicationMasterResponseReflector` class into
separate file.*
- *Use `RegisterApplicationMasterResponseReflector` from within
`YarnResourceManager`*
## Verifying this change
This change added tests and can be verified as follows:
- *Added unit tests for `RegisterApplicationMasterResponseReflector`*
- *Manually deployed a cluster on YARN with HA enabled. Submitted a job,
and killed the master several times. Verified that the right log messages were
generated ("Recovered X containers from previous attempts"), and that the
taskmanager resource ids remained the same.*
## Does this pull request potentially affect one of the following parts:
- Dependencies (does it add or upgrade a dependency): (yes / **no**)
- The public API, i.e., is any changed class annotated with
`@Public(Evolving)`: (yes / **no**)
- The serializers: (yes / **no** / don't know)
- The runtime per-record code paths (performance sensitive): (yes /
**no** / don't know)
- Anything that affects deployment or recovery: JobManager (and its
components), Checkpointing, Yarn/Mesos, ZooKeeper: (**yes** / no / don't know)
- The S3 file system connector: (yes / **no** / don't know)
## Documentation
- Does this pull request introduce a new feature? (yes / **no**)
- If yes, how is the feature documented? (**not applicable** / docs /
JavaDocs / not documented)
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/GJL/flink FLINK-7805-2
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/flink/pull/5597.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 #5597
----
commit 012f656705466cea4ba2e037f7e52f46c0a1bf9f
Author: gyao <gary@...>
Date: 2018-02-27T15:58:53Z
[FLINK-8787][flip6] Do not copy flinkConfiguration in
AbstractYarnClusterDescriptor
commit bf8211112c16a16bd29aba754c1c34f4de960af4
Author: gyao <gary@...>
Date: 2018-02-28T12:04:19Z
[hotfix] Add missing space to log message in ZooKeeperLeaderElectionService
commit a435c6503dcd8e65e34b807d5dc1c7045911e788
Author: gyao <gary@...>
Date: 2018-02-28T12:06:00Z
[hotfix][Javadoc] Fix typo in YARN Utils: teh -> the
commit 3e4c28c38cc04aa29bdb8928fa26daf3c4ab1e69
Author: gyao <gary@...>
Date: 2018-02-28T12:07:04Z
[hotfix][Javadoc] Fix typo in YarnTestBase: teh -> the
commit 6b1efdb290c605d93f17fc0aadfb07205eaf60fd
Author: gyao <gary@...>
Date: 2018-02-28T12:08:25Z
[hotfix][tests] Fix wrong assertEquals in YARNSessionCapacitySchedulerITCase
Test swapped actual and expected arguments.
Remove catching Throwable in test; instead propagate all exceptions.
commit 615c19acfc87bad021b3dce02b6dc5fac2aac784
Author: gyao <gary@...>
Date: 2018-02-28T12:20:23Z
[FLINK-7805][flip6] Recover YARN containers after AM restart.
Recover previously running containers after a restart of the
ApplicationMaster.
This is a port of a feature that was already implemented prior to FLIP-6.
Extract RegisterApplicationMasterResponseReflector class into separate file.
----
> Add HA capabilities to YarnResourceManager
> ------------------------------------------
>
> Key: FLINK-7805
> URL: https://issues.apache.org/jira/browse/FLINK-7805
> Project: Flink
> Issue Type: Improvement
> Components: Distributed Coordination, YARN
> Affects Versions: 1.4.0
> Reporter: Till Rohrmann
> Assignee: Gary Yao
> Priority: Major
> Labels: flip-6
>
> The new {{YarnResourceManager}} implementation does not retrieve allocated
> containers from previous attempts in HA mode like the old
> {{YarnFlinkResourceManager}} did. We should add this functionality in order
> to properly support long running Yarn applications [1].
> [1]
> https://de.hortonworks.com/blog/apache-hadoop-yarn-hdp-2-2-fault-tolerance-features-long-running-services/
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
