fix(quick-start): update data for freshness anomaly

Author: aarthy-dk

testgen/__main__.py

@@ -428,7 +428,7 @@ def quick_start(
     click.echo("loading initial data")
     run_quick_start_increment(0)
     now_date = datetime.now(UTC)
-    time_delta = timedelta(days=-30) # 1 month ago
+    time_delta = timedelta(days=-35) # before the first monitor iteration (~34 days back)
     table_group_id = "0ea85e17-acbe-47fe-8394-9970725ad37d"
     test_suite_id = "9df7489d-92b3-49f9-95ca-512160d7896f"
 
@@ -449,16 +449,19 @@ def quick_start(
         run_quick_start_increment(iteration)
         run_test_execution(test_suite_id, run_date=run_date)
 
-    monitor_iterations = 42  # 3 weeks
+    monitor_iterations = 68  # ~5 weeks
     monitor_interval = timedelta(hours=12)
     monitor_test_suite_id = "823a1fef-9b6d-48d5-9d0f-2db9812cc318"
     # Round down to nearest 12-hour mark (12:00 AM or 12:00 PM UTC)
     now = datetime.now(UTC)
     nearest_12h_mark = now.replace(hour=12 if now.hour >= 12 else 0, minute=0, second=0, microsecond=0)
     monitor_run_date = nearest_12h_mark - monitor_interval * (monitor_iterations - 1)
+    weekday_morning_count = 0
     for iteration in range(1, monitor_iterations + 1):
         click.echo(f"Running monitor iteration: {iteration} / {monitor_iterations}")
-        run_monitor_increment(monitor_run_date, iteration)
+        if monitor_run_date.weekday() < 5 and monitor_run_date.hour < 12:
+            weekday_morning_count += 1
+        run_monitor_increment(monitor_run_date, iteration, weekday_morning_count)
         run_test_execution(monitor_test_suite_id, run_date=monitor_run_date)
         monitor_run_date += monitor_interval
 

testgen/commands/run_quick_start.py

@@ -130,15 +130,18 @@ def _metric_cumulative_shift(iteration: int) -> tuple[float, float]:
     return discount, price
 
 
-def _get_monitor_params_mapping(run_date: datetime, iteration: int = 0) -> dict:
+def _get_monitor_params_mapping(run_date: datetime, iteration: int = 0, weekday_morning_count: int = 0) -> dict:
     # Volume: linear growth with jitter, spike at specific iteration for anomaly
-    if iteration == 37:
+    if iteration == 60:
         new_sales = 100
     else:
-        new_sales = random.randint(8, 12)  # noqa: S311
+        new_sales = random.randint(5, 15)  # noqa: S311
 
-    # Freshness: update every other iteration, late update for anomaly
-    is_update_suppliers_iter = (iteration % 2 == 0 and iteration != 38) or iteration == 39
+    # Freshness: weekday morning updates with 1-day outage after schedule goes active
+    is_weekday = run_date.weekday() < 5
+    is_morning = run_date.hour < 12
+    is_outage = weekday_morning_count == 21
+    is_update_suppliers_iter = is_weekday and is_morning and not is_outage
 
     # Metrics: compute deltas for discount and price shifts
     curr_discount, curr_price = _metric_cumulative_shift(iteration)
@@ -151,11 +154,11 @@ def _get_monitor_params_mapping(run_date: datetime, iteration: int = 0) -> dict:
         "ITERATION_NUMBER": iteration,
         "RUN_DATE": run_date,
         "NEW_SALES": new_sales,
-        "IS_ADD_CUSTOMER_COL_ITER": iteration == 29,
-        "IS_DELETE_CUSTOMER_COL_ITER": iteration == 36,
-        "IS_UPDATE_PRODUCT_ITER": not 14 < iteration < 18,
-        "IS_CREATE_RETURNS_TABLE_ITER": iteration == 32,
-        "IS_DELETE_CUSTOMER_ITER": iteration in (18, 22, 34),
+        "IS_ADD_CUSTOMER_COL_ITER": iteration == 47,
+        "IS_DELETE_CUSTOMER_COL_ITER": iteration == 58,
+        "IS_UPDATE_PRODUCT_ITER": not 24 < iteration < 28,
+        "IS_CREATE_RETURNS_TABLE_ITER": iteration == 52,
+        "IS_DELETE_CUSTOMER_ITER": iteration in (29, 36, 55),
         "IS_UPDATE_SUPPLIERS_ITER": is_update_suppliers_iter,
         "DISCOUNT_DELTA": discount_delta,
         "PRICE_DELTA": price_delta,
@@ -234,8 +237,8 @@ def run_quick_start_increment(iteration):
     setup_cat_tests(iteration)
 
 
-def run_monitor_increment(run_date, iteration):
-    params_mapping = _get_monitor_params_mapping(run_date, iteration)
+def run_monitor_increment(run_date, iteration, weekday_morning_count=0):
+    params_mapping = _get_monitor_params_mapping(run_date, iteration, weekday_morning_count)
     _prepare_connection_to_target_database(params_mapping)
 
     target_db_name = params_mapping["PROJECT_DB"]

testgen/template/quick_start/initial_data_seeding.sql

@@ -54,7 +54,7 @@ SELECT '823a1fef-9b6d-48d5-9d0f-2db9812cc318'::UUID AS id,
        '{TABLE_GROUPS_NAME} Monitor Suite'          AS test_suite_description,
        'N'                                          AS export_to_observability,
        TRUE                                         AS is_monitor,
-       14                                           AS monitor_lookback,
+       28                                           AS monitor_lookback,
        30                                           AS predict_min_lookback;
 
 INSERT INTO job_schedules

testgen/template/quick_start/run_monitor_iteration.sql

@@ -42,7 +42,8 @@ SELECT * FROM new_sales;
 
 
 UPDATE demo.d_ebike_customers
-SET last_contact = :RUN_DATE
+SET last_contact = :RUN_DATE,
+    customer_decile = customer_decile + 1
 WHERE ctid IN (
     SELECT ctid
     FROM demo.d_ebike_customers