Analytics and scheduler fixes

Author: MrEricSir

deploy/cloud/setup-scheduler.sh

@@ -16,8 +16,8 @@ CHECKER_SCHEDULE="${CHECKER_SCHEDULE:-*/3 * * * *}"  # Every 3 minutes
 
 # Analytics reports job
 REPORTS_JOB="munimetro-reports"
-REPORTS_SCHEDULER="munimetro-reports-daily"
-REPORTS_SCHEDULE="${REPORTS_SCHEDULE:-0 0 * * *}"  # Daily at midnight UTC
+REPORTS_SCHEDULER="munimetro-reports-periodic"
+REPORTS_SCHEDULE="${REPORTS_SCHEDULE:-*/30 * * * *}"  # Every 30 minutes
 
 echo "=========================================="
 echo "MuniMetro Cloud Scheduler Setup"
@@ -27,7 +27,7 @@ echo "Region: $REGION"
 echo ""
 echo "Schedules:"
 echo "  Status checker: $CHECKER_SCHEDULE (every 3 min)"
-echo "  Analytics reports: $REPORTS_SCHEDULE (daily midnight UTC)"
+echo "  Analytics reports: $REPORTS_SCHEDULE (every 30 min)"
 echo "=========================================="
 echo ""
 
@@ -100,7 +100,7 @@ gcloud scheduler jobs create http "$REPORTS_SCHEDULER" \
     --max-retry-attempts=2 \
     --max-backoff=3600s \
     --min-backoff=60s \
-    --description="Generates MuniMetro analytics reports daily at midnight UTC" \
+    --description="Generates MuniMetro analytics reports every 30 minutes" \
     --project="$PROJECT_ID"
 
 echo "✓ Analytics reports scheduler created: $REPORTS_SCHEDULER"
@@ -132,7 +132,7 @@ echo "    Target: $CHECKER_JOB"
 echo ""
 echo "  Analytics Reports:"
 echo "    Name: $REPORTS_SCHEDULER"
-echo "    Schedule: $REPORTS_SCHEDULE (daily midnight UTC)"
+echo "    Schedule: $REPORTS_SCHEDULE (every 30 min)"
 echo "    Target: $REPORTS_JOB"
 echo ""
 echo "View scheduler status:"

lib/analytics.py

@@ -397,22 +397,27 @@ def get_delay_frequency(days=7):
     ''', (cutoff,))
     total_checks = cursor.fetchone()['total']
 
-    # Checks by status
+    # Status distribution uses best_status (what users actually see on dashboard)
     cursor.execute('''
-        SELECT status, COUNT(*) as count
+        SELECT best_status, COUNT(*) as count
         FROM status_checks
         WHERE timestamp >= ?
-        GROUP BY status
+        GROUP BY best_status
     ''', (cutoff,))
 
     by_status = {'green': 0, 'yellow': 0, 'red': 0}
     for row in cursor.fetchall():
-        by_status[row['status']] = row['count']
+        by_status[row['best_status']] = row['count']
+
+    # Delay tracking uses raw status (actual delay detections, regardless of smoothing)
+    cursor.execute('''
+        SELECT COUNT(*) as count FROM status_checks
+        WHERE timestamp >= ? AND status = 'yellow'
+    ''', (cutoff,))
+    delayed_checks = cursor.fetchone()['count']
 
     conn.close()
 
-    # Only count yellow as delayed - red is after-hours maintenance
-    delayed_checks = by_status['yellow']
     delay_rate = delayed_checks / total_checks if total_checks > 0 else 0.0
 
     return {
@@ -609,22 +614,20 @@ def get_analytics_report(days=7):
             except (json.JSONDecodeError, ValueError):
                 cached = None
 
-    # Check if cache is fresh
+    # Always return cached data if it exists (never return empty if we have data)
     if cached:
+        cached['from_cache'] = True
         try:
             cached_at = datetime.fromisoformat(cached.get('cached_at', ''))
             cache_age = (datetime.now() - cached_at).total_seconds()
-
-            if cache_age < REPORT_CACHE_MAX_AGE:
-                # Return cached report with cache info
-                cached['from_cache'] = True
-                cached['cache_age'] = round(cache_age, 1)
-                return cached
+            cached['cache_age'] = round(cache_age, 1)
+            cached['cache_stale'] = cache_age >= REPORT_CACHE_MAX_AGE
         except (ValueError, KeyError):
-            pass  # Cache corrupted or missing timestamp
+            cached['cache_age'] = None
+            cached['cache_stale'] = True
+        return cached
 
-    # No fresh cache - return empty report
-    # Reports are generated by scheduled task, not on-demand
+    # No cache at all - return empty report
     return _empty_report(days)