[
https://issues.apache.org/jira/browse/FLINK-23983?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17408575#comment-17408575
]
Yun Tang commented on FLINK-23983:
----------------------------------
I could reproduced it locally by running RocksDBStateBackend tests repeatly.
Since we bumped RocksDB recently, and no obvious tests changed under
flink-statebackend-rocksdb module, I also doubted that some behavior has
changed in RocksDB itself. It took me some time to figure out this might be
caused by change in \{{StateBackendTestBase}} which changed the previous test
by creating a new keyed state-backend but not disposing the older one (refer to
https://github.com/apache/flink/commit/e8daf67ce5096da791e21d0915848c78c395822d
cc @roman [~rkhachatryan] ).
> JVM crash when running RocksDBStateBackend tests
> ------------------------------------------------
>
> Key: FLINK-23983
> URL: https://issues.apache.org/jira/browse/FLINK-23983
> Project: Flink
> Issue Type: Bug
> Components: Runtime / State Backends
> Affects Versions: 1.14.0
> Reporter: Xintong Song
> Priority: Critical
> Labels: test-stability
> Fix For: 1.14.0
>
>
> https://dev.azure.com/apache-flink/apache-flink/_build/results?buildId=22855&view=logs&j=77a9d8e1-d610-59b3-fc2a-4766541e0e33&t=125e07e7-8de0-5c6c-a541-a567415af3ef&l=11131
> You would need to compare the mvn logs "Running xxx" with "Test run xxx in
> xxx" to find out the unfinished test.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
