From afca5603bd3fae9aeb8f1376eeb567ecb9792991 Mon Sep 17 00:00:00 2001
From: Tomas Kopecek <tkopecek@redhat.com>
Date: Nov 11 2020 13:31:36 +0000
Subject: hub: use CTE for build_references


Fixes: https://pagure.io/koji/issue/2535

---

diff --git a/hub/kojihub.py b/hub/kojihub.py
index 9488eaa..e0a0c46 100644
--- a/hub/kojihub.py
+++ b/hub/kojihub.py
@@ -8000,27 +8000,31 @@ def build_references(build_id, limit=None, lazy=False):
     if build_rpm_ids:
         # psql planner gots confused if buildroot table is large (>trillion)
         # and len(rpm_ids) > ~500. In such case it switched to looped sequential scans
-        _dml("SET enable_hashjoin=off", {})
-        q = """SELECT MAX(create_event)
-               FROM standard_buildroot
-               WHERE buildroot_id IN (
-                 SELECT buildroot_id
+        # using "SET enabled_hashjoin=off" improved it for some cases. CTE could be slower for
+        # simple cases but would improve complicated ones.
+        q = """WITH buildroot_ids as (
+                 SELECT DISTINCT buildroot_id
                  FROM buildroot_listing
                  WHERE rpm_id IN %(rpm_ids)s
+               )
+               SELECT MAX(create_event)
+               FROM standard_buildroot
+               WHERE buildroot_id IN (
+                  SELECT buildroot_id FROM buildroot_ids
                )"""
         event_id = (_fetchSingle(q, {'rpm_ids': build_rpm_ids}) or (0,))[0] or 0
-        _dml("SET enable_hashjoin=on", {})
     if build_archive_ids:
-        _dml("SET enable_hashjoin=off", {})
-        q = """SELECT MAX(create_event)
-               FROM standard_buildroot
-               WHERE buildroot_id IN (
-                 SELECT buildroot_id
+        q = """WITH buildroot_ids as (
+                 SELECT DISTINCT buildroot_id
                  FROM buildroot_archives
                  WHERE archive_id IN %(archive_ids)s
+               )
+               SELECT MAX(create_event)
+               FROM standard_buildroot
+               WHERE buildroot_id IN (
+                  SELECT buildroot_id FROM buildroot_ids
                )"""
         event_id2 = (_fetchSingle(q, {'archive_ids': build_archive_ids}) or (0,))[0] or 0
-        _dml("SET enable_hashjoin=on", {})
         event_id = max(event_id, event_id2)
     if event_id:
         q = """SELECT EXTRACT(EPOCH FROM get_event_time(%(event_id)i))"""