You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Prepare for a release with telemetry on by default (#717)
* Prepare for a release with telemetry on by default
Signed-off-by: samikshya-chand_data <samikshya.chand@databricks.com>
* Make edits
Signed-off-by: samikshya-chand_data <samikshya.chand@databricks.com>
* Update version
Signed-off-by: samikshya-chand_data <samikshya.chand@databricks.com>
* Fix CHANGELOG formatting to match previous style
Signed-off-by: samikshya-chand_data <samikshya.chand@databricks.com>
* Fix telemetry e2e tests for default-enabled behavior
- Update test expectations to reflect telemetry being enabled by default
- Add feature flags cache cleanup in teardown to prevent state leakage between tests
- This ensures each test runs with fresh feature flag state
* Add wait after connection close for async telemetry submission
* Remove debug logging from telemetry tests
* Mark telemetry e2e tests as serial - must not run in parallel
Root cause: Telemetry tests share host-level client across pytest-xdist workers,
causing test isolation issues with patches. Tests pass serially but fail with -n auto.
Solution: Add @pytest.mark.serial marker. CI needs to run these separately without -n auto.
* Split test execution to run serial tests separately
Telemetry e2e tests must run serially due to shared host-level
telemetry client across pytest-xdist workers. Running with -n auto
causes test isolation issues where futures aren't properly captured.
Changes:
- Run parallel tests with -m 'not serial' -n auto
- Run serial tests with -m 'serial' without parallelization
- Use --cov-append for serial tests to combine coverage
- Mark telemetry e2e tests with @pytest.mark.serial
- Update test expectations for default telemetry behavior
- Add feature flags cache cleanup in test teardown
* Mark telemetry e2e tests as serial - must not run in parallel
The concurrent telemetry e2e test globally patches telemetry methods
to capture events. When run in parallel with other tests via pytest-xdist,
it captures telemetry events from other concurrent tests, causing
assertion failures (expected 60 events, got 88).
All telemetry e2e tests must run serially to avoid cross-test
interference with the shared host-level telemetry client.
---------
Signed-off-by: samikshya-chand_data <samikshya.chand@databricks.com>
Copy file name to clipboardExpand all lines: CHANGELOG.md
+7Lines changed: 7 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -1,5 +1,12 @@
1
1
# Release History
2
2
3
+
# 4.2.2 (2025-12-01)
4
+
- Change default use_hybrid_disposition to False (databricks/databricks-sql-python#714 by @samikshya-db)
5
+
- Circuit breaker changes using pybreaker (databricks/databricks-sql-python#705 by @nikhilsuri-db)
6
+
- perf: Optimize telemetry latency logging to reduce overhead (databricks/databricks-sql-python#715 by @samikshya-db)
7
+
- basic e2e test for force telemetry verification (databricks/databricks-sql-python#708 by @nikhilsuri-db)
8
+
- Telemetry is ON by default to track connection stats. (Note : This strictly excludes PII, query text, and results) (databricks/databricks-sql-python#717 by @samikshya-db)
9
+
3
10
# 4.2.1 (2025-11-20)
4
11
- Ignore transactions by default (databricks/databricks-sql-python#711 by @jayantsing-db)
0 commit comments