increase limit, introduce randomness

Author: nora-shap

src/sentry/tasks/llm_issue_detection/trace_data.py

@@ -36,19 +36,26 @@ def get_project_top_transaction_traces_for_llm_detection(
     random_offset = random.randint(1, 8)
     end_time = datetime.now(UTC)
     start_time = end_time - timedelta(minutes=start_time_delta_minutes)
-
-    # use for both queries to ensure they are searching the same time window
-    snuba_params = SnubaParams(
-        start=start_time,
-        end=end_time,
-        projects=[project],
-        organization=project.organization,
-    )
     config = SearchResolverConfig(auto_fields=True)
 
+    def _build_snuba_params(start: datetime) -> SnubaParams:
+        """
+        Both queries have different start times and the same end time.
+        """
+        return SnubaParams(
+            start=start,
+            end=end_time,
+            projects=[project],
+            organization=project.organization,
+        )
+
+    transaction_snuba_params = _build_snuba_params(start_time)
+    random_offset = random.randint(1, 8)
+    trace_snuba_params = _build_snuba_params(end_time - timedelta(minutes=30 - random_offset))
+
     # Step 1: Get top transactions by total time in time window
     transactions_result = Spans.run_table_query(
-        params=snuba_params,
+        params=transaction_snuba_params,
         query_string="is_transaction:true",
         selected_columns=[
             "transaction",
@@ -77,7 +84,7 @@ def get_project_top_transaction_traces_for_llm_detection(
         # Step 2: Get ONE trace for this transaction from THE SAME time window
         escaped_transaction_name = UNESCAPED_QUOTE_RE.sub('\\"', transaction_name)
         trace_result = Spans.run_table_query(
-            params=snuba_params,
+            params=trace_snuba_params,
             query_string=f'is_transaction:true transaction:"{escaped_transaction_name}"',
             selected_columns=["trace", "precise.start_ts"],
             orderby=["precise.start_ts"],  # First trace in the window

tests/sentry/tasks/test_llm_issue_detection.py

@@ -308,7 +308,7 @@ def test_returns_deduped_transaction_traces(self) -> None:
         self.store_spans([span1, span2, span3], is_eap=True)
 
         evidence_traces = get_project_top_transaction_traces_for_llm_detection(
-            self.project.id, limit=50, start_time_delta_minutes=30
+            self.project.id, limit=100, start_time_delta_minutes=30
         )
 
         assert len(evidence_traces) == 2