You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by pnowojski <gi...@git.apache.org> on 2018/05/03 13:58:23 UTC
[GitHub] flink pull request #5952: [FLINK-9287][kafka] Properly clean up resources in...
GitHub user pnowojski opened a pull request:
https://github.com/apache/flink/pull/5952
[FLINK-9287][kafka] Properly clean up resources in non EXACTLY_ONCE FlinkKafkaProducer011
Previously FlinkKafkaProducer was not being closed for AT_LEAST_ONCE and NONE Semantics when closing FlinkKafkaProducer011. This was leading to resources leaking (for example increasing number of active threads).
## Verifying this change
This bug fix might be hard to test automatically. This PR adds a new test proposal in separate commit, however it might be flaky if there are other tests being executed in parallel.
## 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/pnowojski/flink f9287
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/flink/pull/5952.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 #5952
----
commit c9139a90dd6a2afb25a4eb3102e1abedf90d8f5f
Author: Piotr Nowojski <pi...@...>
Date: 2018-05-03T13:50:53Z
[FLINK-9287][kafka] Properly clean up resources in non EXACTLY_ONCE FlinkKafkaProducer011
Previously FlinkKafkaProducer was not being closed for AT_LEAST_ONCE and NONE Semantics
when closing FlinkKafkaProducer011. This was leading to resources leaking (for example
increasing number of active threads)
commit 5a1d0962f3fb92a87ee809b5a09106f5c4d05caf
Author: Piotr Nowojski <pi...@...>
Date: 2018-05-03T13:53:40Z
[FLINK-9287][kafka] Ensure threads count do not grow in FlinkKafkaProducer011
----
---
[GitHub] flink pull request #5952: [FLINK-9287][kafka] Properly clean up resources in...
Posted by asfgit <gi...@git.apache.org>.
Github user asfgit closed the pull request at:
https://github.com/apache/flink/pull/5952
---
[GitHub] flink issue #5952: [FLINK-9287][kafka] Properly clean up resources in non EX...
Posted by tzulitai <gi...@git.apache.org>.
Github user tzulitai commented on the issue:
https://github.com/apache/flink/pull/5952
As discussed offline, both approaches I mentioned wouldn't seem to work.
I'll merge as is, let's keep an eye on it to see if the test is flaky.
---
[GitHub] flink issue #5952: [FLINK-9287][kafka] Properly clean up resources in non EX...
Posted by tzulitai <gi...@git.apache.org>.
Github user tzulitai commented on the issue:
https://github.com/apache/flink/pull/5952
Would filtering the thread count by thread name be helpful here?
Another possible approach to test this:
Maybe we can simply verify that the current transaction producer is closed after the cleanup?
This would require making the `getCurrentTransaction()` method visible for tests, though. We'll also need to have a `isClosed()` method on the `FlinkKafkaProducer`.
---