From 47025a1060eb18123d49bc6ce7cd546a0e93ee75 Mon Sep 17 00:00:00 2001
From: Tomas Kopecek <tkopecek@redhat.com>
Date: Mon, 2 Nov 2020 11:12:52 +0100
Subject: [PATCH] hub: use CTE for build_references

Fixes: https://pagure.io/koji/issue/2535
---
 hub/kojihub.py | 30 +++++++++++++++++-------------
 1 file changed, 17 insertions(+), 13 deletions(-)

diff --git a/hub/kojihub.py b/hub/kojihub.py
index 2df0fb0b..c4e0611f 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
-               )"""
-        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)
+               )
+               SELECT MAX(create_event)
                FROM standard_buildroot
                WHERE buildroot_id IN (
-                 SELECT buildroot_id
+                  SELECT buildroot_id FROM buildroot_ids
+               )"""
+        event_id = (_fetchSingle(q, {'rpm_ids': build_rpm_ids}) or (0,))[0] or 0
+    if build_archive_ids:
+        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))"""