[
https://issues.apache.org/jira/browse/FLINK-1568?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14335019#comment-14335019
]
ASF GitHub Bot commented on FLINK-1568:
---------------------------------------
GitHub user tillrohrmann opened a pull request:
https://github.com/apache/flink/pull/436
[FLINK-1568] [FLINK-1604] [FLINK-1606] [FLINK-1607] Fixes failing
JobManagerFailsITCase and SimpleRecoveryITCase
This PR contains fixes for the failing ```JobManagerFailsITCase``` and
```SimpleRecoveryITCase```.
In the wake of fixing these issues, I came across a livelock in the
```PartitionRequestClientFactory``` which is fixed as well.
Moreover, the way cached partition infos are treated has been corrected. In
case of a job restart they are now discarded.
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/tillrohrmann/flink fixJobManagerFailsITCase
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/flink/pull/436.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 #436
----
commit ee70fa7f6730ee6415e5cd1dcfcc7ebcb4fb8be1
Author: Till Rohrmann <[email protected]>
Date: 2015-02-20T11:34:19Z
[FLINK-1606] [tests] Fixes JobManagerFailsITCase for Akka 2.2.1 by setting
gate-invalid-address-for = 5s and quarantine-systems-for = off
commit 9659d392391585f5ff84e67fd5a4a79f3d110fba
Author: Till Rohrmann <[email protected]>
Date: 2015-02-20T16:02:04Z
[FLINK-1607] [tests] Fixes SimpleRecoveryITCase by rebalancing the data
sources so that all mappers receive at least one element
commit 534ebd39e8b0757f5a67f98b43f8d79d3ed02799
Author: Till Rohrmann <[email protected]>
Date: 2015-02-23T12:49:05Z
[FLINK-1604] [runtime] Fixes livelock in
PartitionRequestClientFactory.createPartitionRequestClient
Replaces recursive concurrent modification resolution by while loop
Turns off stdout-logging of Akka. Sends proper exceptions in ErrorResponse.
Proper stream closing
commit 65e22a417c29efda650440c53c3fa0f3767bd9b9
Author: Ufuk Celebi <[email protected]>
Date: 2015-02-24T11:41:35Z
[FLINK-1604] [FLINK-1568] Add initial connect failure test
commit 505cbecd005d712c25a47b64034a72a539cb1ea8
Author: Till Rohrmann <[email protected]>
Date: 2015-02-24T14:58:25Z
[FLINK-1607] [runtime] Moves PartialPartitionInfo from ExecutionVertex to
Execution to automatically clear them in case of restart
----
> Add failure-case tests for data exchange
> ----------------------------------------
>
> Key: FLINK-1568
> URL: https://issues.apache.org/jira/browse/FLINK-1568
> Project: Flink
> Issue Type: Improvement
> Components: Distributed Runtime
> Affects Versions: master
> Reporter: Ufuk Celebi
>
> There are currently no tests of the distributed runtime data exchange in
> failure scenarios.
> The tests need to verify how errors are propagated to the tasks and how they
> affect the state of the exchanged intermediate result.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
