This is an automated email from the ASF dual-hosted git repository.
zhengruifeng pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/spark.git
The following commit(s) were added to refs/heads/master by this push:
new 9f772e8453d1 [MINOR][CONNECT][PYTHON][TEST] Fix mock session leakage
in test_clinet.py
9f772e8453d1 is described below
commit 9f772e8453d10c3a6f0f48d5f869e7c20b9e5371
Author: Yi Hu <[email protected]>
AuthorDate: Thu Jul 16 13:56:01 2026 +0800
[MINOR][CONNECT][PYTHON][TEST] Fix mock session leakage in test_clinet.py
### What changes were proposed in this pull request?
Close mock sessions at the end of test_session_hook and its variant
### Why are the changes needed?
`test_client.py` creates mock `SparkSession` instances that are not
stopped, leaving `SparkSession._default_session` set globally. On shared test
workers (e.g., under `pytest-xdist`), subsequent tests calling
`SparkSession.builder.getOrCreate()` reuse this leftover mock session, causing
downstream test failures such as `KeyError: 'spark.sql.session.timeZone'`.
### Does this PR introduce any user-facing change?
No. Test-only change.
### How was this patch tested?
Verified by running `test_client.py` followed by parity tests on a single
worker process:
```bash
python -m pytest -v -n 1 --dist loadscope
python/pyspark/sql/tests/connect/client/test_client.py
python/pyspark/pandas/tests/connect/diff_frames_ops/test_parity_basic.py
```
against a sparkconnect server
### Was this patch authored or co-authored using generative AI tooling?
Generated-by: Gemini
Closes #57282 from Abacn/fix-client-leak.
Authored-by: Yi Hu <[email protected]>
Signed-off-by: Ruifeng Zheng <[email protected]>
---
.../sql/tests/connect/client/test_client.py | 29 +++++++++++++---------
1 file changed, 17 insertions(+), 12 deletions(-)
diff --git a/python/pyspark/sql/tests/connect/client/test_client.py
b/python/pyspark/sql/tests/connect/client/test_client.py
index fefbad8960d0..b5bf76d86df4 100644
--- a/python/pyspark/sql/tests/connect/client/test_client.py
+++ b/python/pyspark/sql/tests/connect/client/test_client.py
@@ -425,20 +425,24 @@ class SparkConnectClientTestCase(unittest.TestCase):
session = (
RemoteSparkSession.builder.remote("sc://foo")._registerHook(TestHook).getOrCreate()
)
- self.assertEqual(inits, 1)
- self.assertEqual(calls, 0)
- session.client._stub = MockService(session.client._session_id)
- session.client.disable_reattachable_execute()
+ try:
+ self.assertEqual(inits, 1)
+ self.assertEqual(calls, 0)
+ session.client._stub = MockService(session.client._session_id)
+ session.client.disable_reattachable_execute()
- # Called from _execute_and_fetch_as_iterator
- session.range(1).collect()
- self.assertEqual(inits, 1)
- self.assertEqual(calls, 1)
+ # Called from _execute_and_fetch_as_iterator
+ session.range(1).collect()
+ self.assertEqual(inits, 1)
+ self.assertEqual(calls, 1)
- # Called from _execute
- session.udf.register("test_func", lambda x: x + 1)
- self.assertEqual(inits, 1)
- self.assertEqual(calls, 2)
+ # Called from _execute
+ session.udf.register("test_func", lambda x: x + 1)
+ self.assertEqual(inits, 1)
+ self.assertEqual(calls, 2)
+ finally:
+ # Close the session to avoid leaking dummy session inter-test
+ session.stop()
def test_session_hook_preserved_after_new_session(self):
calls = 0
@@ -473,6 +477,7 @@ class SparkConnectClientTestCase(unittest.TestCase):
# against the unreachable endpoint at interpreter shutdown.
new_session.client.close()
session.client.close()
+ session.stop()
def test_new_session_preserves_custom_channel_builder(self):
class CustomChannelBuilder(DefaultChannelBuilder):
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
