nahtonaj opened a new pull request, #58217:
URL: https://github.com/apache/spark/pull/58217
<!--
Thanks for sending a pull request! Here are some tips for you:
1. If this is your first time, please read our contributor guidelines:
https://spark.apache.org/contributing.html
2. Ensure you have added or run the appropriate tests for your PR:
https://spark.apache.org/developer-tools.html
3. If the PR is unfinished, add '[WIP]' in your PR title, e.g.,
'[WIP][SPARK-XXXX] Your PR title ...'.
4. Be sure to keep the PR description updated to reflect all changes.
5. Please write your PR title to summarize what this PR proposes.
6. If possible, provide a concise example to reproduce the issue for a
faster review.
7. If you want to add a new configuration, please read the guideline first
for naming configurations in
'common/utils/src/main/scala/org/apache/spark/internal/config/ConfigEntry.scala'.
8. If you want to add or modify an error type or message, please read the
guideline first in
'common/utils/src/main/resources/error/README.md'.
-->
### What changes were proposed in this pull request?
<!--
Please clarify what changes you are proposing. The purpose of this section
is to outline the changes and how this PR fixes the issue.
If possible, please consider writing useful notes for better and faster
reviews in your PR. See the examples below.
1. If you refactor some codes with changing classes, showing the class
hierarchy will help reviewers.
2. If you fix some SQL features, you can provide some references of other
DBMSes.
3. If there is design documentation, please add the link.
4. If there is a discussion in the mailing list, please add the link.
-->
This pull request handles state server shutdown while the listener is still
waiting for the Python worker to connect.
The patch makes the following changes:
1. Reorders `TransformWithStateInPySparkPythonRunner.stop()` so that it
interrupts the state server listener thread before closing the server socket
channel. This matches the existing executor-side cleanup ordering.
2. Handles expected interruption and channel-closure exceptions thrown by
`ServerSocketChannel.accept()` before the Python worker connects.
3. Preserves the listener thread's interrupt status for interrupt-driven
shutdown and transitions the stateful processor handle to `CLOSED`.
4. Treats `ClosedChannelException`, including its
`AsynchronousCloseException` subclass, as an expected shutdown path without
incorrectly setting the interrupt status.
5. Adds regression coverage for the Java NIO exception hierarchy and
interrupt semantics.
6. Adds real loopback `ServerSocketChannel` tests for both interruption
before entering `accept()` and interruption while the listener is blocked
inside `accept()`. The tests verify bounded listener termination, no escaped
exception, channel closure, the `CLOSED` handle state, and no protocol output.
No public APIs or configuration options are added or changed.
### Why are the changes needed?
<!--
Please clarify why the changes are needed. For instance,
1. If you propose a new API, clarify the use case for a new API.
2. If you fix a bug, you can clarify why it is a bug.
-->
When a TransformWithState in PySpark runner is stopped before the Python
worker connects to its JVM state server, the listener can still be blocked in
`ServerSocketChannel.accept()`.
Previously, the driver-side cleanup closed the server socket channel before
interrupting the listener thread. Closing the channel wakes the listener with
an expected `AsynchronousCloseException`. Because the pre-connect path did not
handle this exception, it escaped
`TransformWithStateInPySparkStateServer.run()` and was wrapped as:
```text
TransformWithStateInPySpark state server daemon thread exited unexpectedly
(crashed)
```
This reports an orderly shutdown as an unexpected daemon-thread failure. The
same listener implementation is used by the executor-side lifecycle, where
interruption can produce `ClosedByInterruptException`.
The patch makes both expected shutdown paths explicit. It interrupts before
closing, handles the relevant Java NIO exception hierarchy, restores the
interrupt flag only for interrupt-driven shutdown, and transitions the
processor handle to its terminal `CLOSED` state.
### Does this PR introduce _any_ user-facing change?
<!--
Note that it means *any* user-facing change including all aspects such as
new features, bug fixes, or other behavior changes. Documentation-only updates
are not considered user-facing changes.
If yes, please clarify the previous behavior and the change this PR proposes
- provide the console output, description and/or an example to show the
behavior difference if possible.
If possible, please also clarify if this is a user-facing change compared to
the released Spark versions or within the unreleased branches such as master.
If no, write 'No'.
-->
Yes.
Previously, stopping a TransformWithState in PySpark query before its Python
worker connected could report the expected state server listener shutdown as an
unexpected daemon-thread crash.
After this change, the state server listener terminates cleanly during this
shutdown window. Expected interruption or channel-closure exceptions no longer
surface as a spurious `SparkException`.
This is a bug fix relative to released Spark versions as well as the
behavior on the current `master` branch. It does not change the
TransformWithState API or successful query execution behavior.
### How was this patch tested?
<!--
If tests were added, say they were added here. Please make sure to add some
test cases that check the changes thoroughly including negative and positive
cases if possible.
If it was tested in a way different from regular unit tests, please clarify
how you tested step by step, ideally copy and paste-able, so that other
reviewers can test and check, and descendants can verify in the future.
If tests were not added, please describe why they were not added and/or why
it was difficult to add.
If benchmark tests were added, please run the benchmarks in GitHub Actions
for the consistent environment, and the instructions could accord to:
https://spark.apache.org/developer-tools.html#github-workflow-benchmarks.
-->
Added tests to `TransformWithStateInPySparkStateServerSuite` covering:
1. `InterruptedException`, `InterruptedIOException`, and
`ClosedByInterruptException` while waiting for the Python worker, including
preservation of the listener thread's interrupt status.
2. `AsynchronousCloseException` and `ClosedChannelException` while waiting
for the Python worker, including verification that the listener thread is not
marked interrupted.
3. A real loopback `ServerSocketChannel` with the listener thread already
interrupted before entering `accept()`.
4. A real loopback `ServerSocketChannel` with the listener confirmed to be
blocked inside `accept()` before applying the production-equivalent
interrupt-then-close shutdown sequence.
5. Bounded listener termination, no escaped throwable, server channel
closure, transition to `StatefulProcessorHandleState.CLOSED`, and no protocol
output for the real-channel shutdown paths.
The following local static checks passed:
```bash
git diff --check
dev/lint-scala
```
`dev/lint-scala` completed both the Scalastyle and Scalafmt checks
successfully.
The focused suite can be run with:
```bash
build/sbt -Phive -Phive-thriftserver \
"sql/core/testOnly
org.apache.spark.sql.execution.python.streaming.TransformWithStateInPySparkStateServerSuite"
```
The focused SBT suite could not be executed in the local development
environment because Maven Central and the SBT repository hosts failed DNS
resolution while retrieving the required SBT launcher and dependencies. The
generated partial launcher file was removed from the worktree. The suite is
expected to run in GitHub Actions, where the build dependencies are available.
### Was this patch authored or co-authored using generative AI tooling?
<!--
If generative AI tooling has been used in the process of authoring this
patch, please include the
phrase: 'Generated-by: ' followed by the name of the tool and its version.
If no, write 'No'.
Please refer to the [ASF Generative Tooling
Guidance](https://www.apache.org/legal/generative-tooling.html) for details.
-->
Generated-by: databricks-gpt-5-6-sol on KiroCrew 0.2.0-customapi.5
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
