[
https://issues.apache.org/jira/browse/FLINK-9144?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16435876#comment-16435876
]
ASF GitHub Bot commented on FLINK-9144:
---------------------------------------
GitHub user NicoK opened a pull request:
https://github.com/apache/flink/pull/5842
[FLINK-9144][network] fix SpillableSubpartition causing jobs to hang when
spilling
## What is the purpose of the change
This should fix various scenarios where the backlog accounting in the
`SpillableSubpartition` was wrong during spilling buffers and where empty
buffers have been spilled (unnecessarily).
## Brief change log
- improve logging in `SpillableSubpartition` and a minor optimisation when
getting a `MemorySegment`
- make sure that backlog accounting is right in
`SpillableSubpartition#spillFinishedBufferConsumers()` in various cases:
-- empty buffers
-- buffers being spilled multiple times (currently code led to buffers
being spilled twice, once empty, and then the final contents)
- do not spill empty buffers
- always spill all buffers when finishing a stream
## Verifying this change
This change added tests and can be verified as follows:
- adapt `SpillableSubpartitionTest` to cover more spilling scenarios, also
with partial buffers and with the same pattern that the `RecordWriter` is
actually using it
- adapt the batch e2e tests to cover the scenario for this fix (it was
blocking)
## Does this pull request potentially affect one of the following parts:
- Dependencies (does it add or upgrade a dependency): **no**
- The public API, i.e., is any changed class annotated with
`@Public(Evolving)`: **no**
- The serializers: **no**
- The runtime per-record code paths (performance sensitive): **no**
(per-buffer only)
- Anything that affects deployment or recovery: JobManager (and its
components), Checkpointing, Yarn/Mesos, ZooKeeper: **no**
- The S3 file system connector: **no**
## Documentation
- Does this pull request introduce a new feature? **no**
- If yes, how is the feature documented? **JavaDocs**
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/NicoK/flink flink-9144
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/flink/pull/5842.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 #5842
----
commit 9463553d7aae31fb54099cf91234472af3eff3db
Author: Nico Kruber <nico@...>
Date: 2018-04-06T17:35:44Z
[hotfix][checkstyle] fix warnings in LocalBufferPool
commit 01c27d091715a4d2672d430d42cd79947458d999
Author: Nico Kruber <nico@...>
Date: 2018-04-06T17:36:15Z
[hotfix][network] extend logging message in SpillableSubpartition
commit 8f6bc82722076718a75dc5234545bc7fbd917dbf
Author: Nico Kruber <nico@...>
Date: 2018-04-06T17:36:33Z
[hotfix][network] minor optimisation in LocalBufferPool
commit b759456b8f16d17b7918da4df7398c28205793ab
Author: Nico Kruber <nico@...>
Date: 2018-04-06T17:34:44Z
[FLINK-9144][network] fix SpillableSubpartition causing jobs to hang when
spilling
----
> Spilling batch job hangs
> ------------------------
>
> Key: FLINK-9144
> URL: https://issues.apache.org/jira/browse/FLINK-9144
> Project: Flink
> Issue Type: Bug
> Components: Network
> Affects Versions: 1.5.0, 1.6.0
> Reporter: Nico Kruber
> Assignee: Nico Kruber
> Priority: Blocker
> Fix For: 1.5.0
>
>
> A user on the mailing list reported that his batch job stops to run with
> Flink 1.5 RC1:
> https://lists.apache.org/thread.html/43721934405019e7255fda627afb7c9c4ed0d04fb47f1c8f346d4194@%3Cdev.flink.apache.org%3E
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
