From 421097ab7dbd592790ea702b400d680bbc2df40d Mon Sep 17 00:00:00 2001
From: Arpad Borsos <arpad.borsos@sentry.io>
Date: Mon, 27 Jan 2025 14:28:41 +0100
Subject: [PATCH] Add missing indices relevant for Repo Deletion

There are a few very slow `DELETE` queries as part of repo deletion.
The worst of those is deleting the newly created "shadow owner", which takes >1 minute.

Running all the `on_delete` hooks using django (which has since been fixes) has revealed that the `SET NULL` of `Commit.author` is the culprit, as that column did not have any index previously.
That is similar to the `SET NULL` on `Repository.fork`, except that query is by far not as slow, probably because of a lot less records that need to be scanned.

---

While at it, this also fixes a django warning related to the `default` not being a callable, and aligns the `Repository.fork` `SET NULL` constraint with the reality of the database.
---
 .../core/migrations/0064_missing_indices.py   | 46 +++++++++++++++++++
 shared/django_apps/core/models.py             |  6 ++-
 2 files changed, 50 insertions(+), 2 deletions(-)
 create mode 100644 shared/django_apps/core/migrations/0064_missing_indices.py

diff --git a/shared/django_apps/core/migrations/0064_missing_indices.py b/shared/django_apps/core/migrations/0064_missing_indices.py
new file mode 100644
index 000000000..199ce605d
--- /dev/null
+++ b/shared/django_apps/core/migrations/0064_missing_indices.py
@@ -0,0 +1,46 @@
+# Generated by Django 4.2.16 on 2025-01-27 13:18
+
+import django.db.models.deletion
+from django.contrib.postgres.operations import AddIndexConcurrently
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+    atomic = False
+
+    dependencies = [
+        ("core", "0063_increment_version"),
+    ]
+
+    operations = [
+        migrations.AlterField(
+            model_name="repository",
+            name="fork",
+            field=models.ForeignKey(
+                blank=True,
+                db_column="forkid",
+                null=True,
+                on_delete=django.db.models.deletion.SET_NULL,
+                to="core.repository",
+            ),
+        ),
+        migrations.AlterField(
+            model_name="repository",
+            name="languages",
+            field=django.contrib.postgres.fields.ArrayField(
+                base_field=models.CharField(),
+                blank=True,
+                default=list,
+                null=True,
+                size=None,
+            ),
+        ),
+        AddIndexConcurrently(
+            model_name="commit",
+            index=models.Index(fields=["author"], name="commits_author_cd2f50_idx"),
+        ),
+        AddIndexConcurrently(
+            model_name="repository",
+            index=models.Index(fields=["fork"], name="repos_forkid_4cd440_idx"),
+        ),
+    ]
diff --git a/shared/django_apps/core/models.py b/shared/django_apps/core/models.py
index eb3afddb5..69c632a81 100644
--- a/shared/django_apps/core/models.py
+++ b/shared/django_apps/core/models.py
@@ -108,12 +108,12 @@ class Languages(models.TextChoices):
     language = models.TextField(
         null=True, blank=True, choices=Languages.choices
     )  # Really an ENUM in db
-    languages = ArrayField(models.CharField(), default=[], blank=True, null=True)
+    languages = ArrayField(models.CharField(), default=list, blank=True, null=True)
     languages_last_updated = DateTimeWithoutTZField(null=True, blank=True)
     fork = models.ForeignKey(
         "core.Repository",
         db_column="forkid",
-        on_delete=models.DO_NOTHING,
+        on_delete=models.SET_NULL,
         null=True,
         blank=True,
     )
@@ -149,6 +149,7 @@ class Meta:
         app_label = CORE_APP_LABEL
         ordering = ["-repoid"]
         indexes = [
+            models.Index(fields=["fork"]),
             models.Index(
                 fields=["service_id", "author"],
                 name="repos_service_id_author",
@@ -287,6 +288,7 @@ class Meta:
             )
         ]
         indexes = [
+            models.Index(fields=["author"]),
             models.Index(
                 fields=["repository", "-timestamp"],
                 name="commits_repoid_timestamp_desc",