Optimize get_monthly_active_user_count by splitting up queries to avo…

…id usage of .distinct

osf/metrics/reporters/institution_summary_monthly.py

@@ -85,19 +85,22 @@ def get_monthly_logged_in_user_count(self, institution, yearmonth):
         ).count()
 
     def get_monthly_active_user_count(self, institution, yearmonth):
-        institution_users = institution.get_institution_users().filter(
+        institution_user_ids = institution.get_institution_users().filter(
             date_disabled__isnull=True
-        )
+        ).values_list('id', flat=True)
+
+        logs_active_users = institution.logs.filter(
+            created__gte=yearmonth.target_month(),
+            created__lt=yearmonth.next_month(),
+            user_id__in=institution_user_ids
+        ).values_list('user_id', flat=True)
+
+        preprint_logs_active_users = institution.preprint_logs.filter(
+            created__gte=yearmonth.target_month(),
+            created__lt=yearmonth.next_month(),
+            user_id__in=institution_user_ids
+        ).values_list('user_id', flat=True)
+
+        active_user_ids = set(logs_active_users).union(set(preprint_logs_active_users))
 
-        active_users = institution_users.filter(
-            Q(
-                logs__created__gte=yearmonth.target_month(),
-                logs__created__lt=yearmonth.next_month()
-            ) |
-            Q(
-                preprint_logs__created__gte=yearmonth.target_month(),
-                preprint_logs__created__lt=yearmonth.next_month()
-            )
-        ).distinct()
-
-        return active_users.count()
+        return len(active_user_ids)