feat: add analysis period logging

jetstream/analysis.py

@@ -269,6 +269,7 @@ def publish_view(
                 extra={
                     "experiment": self.config.experiment.normandy_slug,
                     "analysis_basis": analysis_basis,
+                    "analysis_period": window_period.value,
                 },
             )
             return
@@ -384,6 +385,7 @@ def publish_view(
                 extra={
                     "experiment": self.config.experiment.normandy_slug,
                     "analysis_basis": analysis_basis,
+                    "analysis_period": window_period.value,
                 },
             )
             raise
@@ -478,6 +480,7 @@ def calculate_metrics(
                     extra={
                         "experiment": self.config.experiment.normandy_slug,
                         "analysis_basis": analysis_basis,
+                        "analysis_period": f"{period.value}_{window}",
                     },
                 )
                 raise
@@ -576,6 +579,7 @@ def calculate_metric_for_ds(
                             "experiment": self.config.experiment.normandy_slug,
                             "metric": metric.name,
                             "analysis_basis": analysis_basis,
+                            "analysis_period": f"{period.value}_{window}",
                         },
                     )
                 return ""
@@ -618,6 +622,7 @@ def calculate_statistics(
                     "statistic": metric.statistic.name(),
                     "analysis_basis": analysis_basis,
                     "segment": segment,
+                    "analysis_period": period.value,
                 },
             )
             return StatisticResultCollection.model_validate([])
@@ -707,6 +712,7 @@ def subset_metric_table(
                     "metric": summary.metric.name,
                     "analysis_basis": analysis_basis,
                     "segment": segment,
+                    "analysis_period": period.value,
                 },
             )
             return None
@@ -858,6 +864,7 @@ def _create_subset_metric_table_query_covariate(
                     "metric": metric.name,
                     "analysis_basis": analysis_basis.value,
                     "segment": segment,
+                    "analysis_period": period.value,
                 },
             )
             return self._create_subset_metric_table_query_univariate(
@@ -1387,6 +1394,7 @@ def run(
                             extra={
                                 "experiment": self.config.experiment.normandy_slug,
                                 "analysis_basis": analysis_basis.value,
+                                "analysis_period": period.value,
                             },
                         )
                         continue
@@ -1524,6 +1532,7 @@ def run(
                                         "experiment": self.config.experiment.normandy_slug,
                                         "metric": metric.name,
                                         "analysis_basis": analysis_basis.value,
+                                        "analysis_period": period.value,
                                     },
                                 )
                             continue

jetstream/logging/bigquery_log_handler.py

@@ -41,6 +41,9 @@ def _buffer_to_json(self, buffer):
                     None if not hasattr(record, "analysis_basis") else record.analysis_basis
                 ),
                 "segment": None if not hasattr(record, "segment") else record.segment,
+                "analysis_period": (
+                    None if not hasattr(record, "analysis_period") else record.analysis_period
+                ),
                 "message": record.getMessage(),
                 "log_level": record.levelname,
                 "exception": str(record.exc_info),

jetstream/statistics.py

@@ -274,6 +274,7 @@ def apply(
                             "statistic": self.name(),
                             "analysis_basis": analysis_basis.value,
                             "segment": segment,
+                            "analysis_period": self.period.value if self.period else None,
                         },
                     )
 

jetstream/tests/integration/test_analysis_integration.py

@@ -1186,13 +1186,15 @@ def test_logging(
         assert error_logs[4].get("statistic") == "bootstrap_mean"
         assert error_logs[4].get("analysis_basis") == "enrollments"
         assert error_logs[4].get("segment") == "regular_user_v3"
+        assert error_logs[4].get("analysis_period") == "week"
 
         assert error_logs[5].get("log_level") == "ERROR"
         assert error_logs[5].get("experiment") == "test-experiment"
         assert error_logs[5].get("metric") == "active_hours"
         assert error_logs[5].get("statistic") == "bootstrap_mean"
         assert error_logs[5].get("analysis_basis") == "exposures"
         assert error_logs[5].get("segment") == "regular_user_v3"
+        assert error_logs[5].get("analysis_period") == "week"
 
     # wait for profiling results to land in BigQuery
     # todo: improve this test as it might lead to flakiness

jetstream/tests/integration/test_logging_integration.py

@@ -16,6 +16,7 @@ def _logging_table_setup(self, client, temporary_dataset, project_id):
             bigquery.SchemaField("statistic", "STRING"),
             bigquery.SchemaField("analysis_basis", "STRING"),
             bigquery.SchemaField("segment", "STRING"),
+            bigquery.SchemaField("analysis_period", "STRING"),
             bigquery.SchemaField("message", "STRING"),
             bigquery.SchemaField("log_level", "STRING"),
             bigquery.SchemaField("exception", "STRING"),
@@ -54,6 +55,7 @@ def test_logging_to_bigquery(self, client, temporary_dataset, project_id):
                 "statistic": "test_statistic",
                 "segment": "all",
                 "analysis_basis": "enrollments",
+                "analysis_period": "week_2",
             },
         )
         logger.exception(
@@ -85,6 +87,7 @@ def test_logging_to_bigquery(self, client, temporary_dataset, project_id):
             and r.segment == "all"
             and r.analysis_basis == "enrollments"
             and r.source == "jetstream"
+            and r.analysis_period == "week_2"
             for r in result
         )
         assert any(

jetstream/tests/test_analysis.py

@@ -25,6 +25,7 @@
     EnrollmentNotCompleteException,
     ExplicitSkipException,
     HighPopulationException,
+    StatisticComputationException,
     UnsupportedApplicationException,
 )
 from jetstream.metric import Metric
@@ -1292,6 +1293,13 @@ def test_calculate_metric_for_ds_returns_empty_on_specific_errors(
 
     assert result == ""
     assert "simulated error" in caplog.text
+    # assert logged extras are as expected
+    assert len(caplog.records) == 1
+    log = caplog.records[0]
+    assert log.experiment == experiments[0].normandy_slug
+    assert log.analysis_basis == AnalysisBasis.ENROLLMENTS
+    assert log.analysis_period == f"{AnalysisPeriod.WEEK.value}_1"
+    assert log.metric == mock_metric.name
 
 
 def test_calculate_metric_for_ds_raises_for_other_errors(experiments, monkeypatch):
@@ -1401,6 +1409,34 @@ def test_calculate_statistics_returns_empty_for_none_segment_data(experiments):
     assert result.root == []
 
 
+def test_calculate_statistics_logging_extras(experiments, monkeypatch, caplog):
+    """Ensure that the expected extras are logged."""
+    summary = MagicMock()
+    segment_data = MagicMock()
+    mocked_from_config = Mock(side_effect=StatisticComputationException("Statistic failed!"))
+    monkeypatch.setattr("jetstream.analysis.Summary.from_config", mocked_from_config)
+
+    with caplog.at_level(logging.ERROR):
+        result = (
+            _empty_analysis(experiments)
+            .calculate_statistics(
+                summary, segment_data, "all", AnalysisBasis.ENROLLMENTS, 7, AnalysisPeriod.WEEK
+            )
+            .compute(scheduler="synchronous")
+        )
+
+    assert result.root == []
+    assert "Statistic failed!" in caplog.text
+    assert len(caplog.records) == 1
+    log = caplog.records[0]
+    assert log.experiment == "normandy-test-slug"
+    assert log.segment == "all"
+    assert log.analysis_basis == AnalysisBasis.ENROLLMENTS
+    assert log.analysis_period == AnalysisPeriod.WEEK.value
+    assert log.metric == summary.metric.name
+    assert log.statistic == summary.statistic.name()
+
+
 def test_run_continues_after_google_api_error(experiments, monkeypatch, caplog):
     """Ensure that a GoogleAPICallError from one data source does not prevent
     save_statistics from being called, and no downstream BQ calls are made with