fix: add index to row history table (baserow#4991)

Author: silvestrid

backend/src/baserow/contrib/database/migrations/0206_rowhistory_database_ro_action__6ea699_idx.py

@@ -0,0 +1,19 @@
+# Generated by Django 5.2.12 on 2026-03-17 09:16
+
+from django.db import migrations, models
+from django.contrib.postgres.operations import AddIndexConcurrently
+
+
+class Migration(migrations.Migration):
+    atomic = False
+
+    dependencies = [
+        ('database', '0205_formvieweditrowfield'),
+    ]
+
+    operations = [
+        AddIndexConcurrently(
+            model_name='rowhistory',
+            index=models.Index(fields=['action_timestamp'], name='database_ro_action__6ea699_idx'),
+        ),
+    ]

backend/src/baserow/contrib/database/rows/history.py

@@ -2,7 +2,7 @@
 from itertools import groupby
 
 from django.conf import settings
-from django.db import router
+from django.db import connection
 from django.db.models import QuerySet
 from django.dispatch import receiver
 
@@ -18,6 +18,7 @@
 from baserow.contrib.database.rows.types import ActionData
 from baserow.core.action.signals import action_done
 from baserow.core.models import Workspace
+from baserow.core.psycopg import sql
 from baserow.core.telemetry.utils import baserow_trace
 from baserow.core.types import AnyUser
 
@@ -68,15 +69,33 @@ def list_row_history(
         return queryset
 
     @classmethod
-    def delete_entries_older_than(cls, cutoff: datetime):
+    def delete_entries_older_than(cls, cutoff: datetime, batch_size: int = 20_000):
         """
-        Deletes all row history entries that are older than the given cutoff date.
+        Deletes all row history entries that are older than the given cutoff date
+        in batches to avoid long-running transactions.
 
         :param cutoff: The date and time before which all entries will be deleted.
+        :param batch_size: The number of rows to delete per batch.
         """
 
-        delete_qs = RowHistory.objects.filter(action_timestamp__lt=cutoff)
-        delete_qs._raw_delete(using=router.db_for_write(delete_qs.model))
+        table = sql.Identifier(RowHistory._meta.db_table)
+        query = sql.SQL(
+            """
+            WITH to_delete AS (
+                SELECT id FROM {table}
+                WHERE action_timestamp < %s
+                LIMIT %s
+            )
+            DELETE FROM {table}
+            USING to_delete
+            WHERE {table}.id = to_delete.id
+            """
+        ).format(table=table)
+        while True:
+            with connection.cursor() as cursor:
+                cursor.execute(query, [cutoff, batch_size])
+                if cursor.rowcount == 0:
+                    break
 
 
 @receiver(action_done)

backend/src/baserow/contrib/database/rows/models.py

@@ -59,4 +59,9 @@ class RowHistory(models.Model):
 
     class Meta:
         ordering = ("-action_timestamp", "-id")
-        indexes = [models.Index(fields=["table", "row_id", "-action_timestamp", "-id"])]
+        indexes = [
+            # For deleting history entries by action timestamp.
+            models.Index(fields=["action_timestamp"]),
+            # For listing the history of a row.
+            models.Index(fields=["table", "row_id", "-action_timestamp", "-id"]),
+        ]

changelog/entries/unreleased/bug/add_index_to_resolve_slow_cleanup_of_row_history_table.json

@@ -0,0 +1,9 @@
+{
+    "type": "bug",
+    "message": "Fixed slow cleanup of row history table by adding a database index.",
+    "issue_origin": "github",
+    "issue_number": null,
+    "domain": "database",
+    "bullet_points": [],
+    "created_at": "2026-03-17"
+}