Blog: Add Timescale to pg_partman migration guide (#40037)

## I have read the
[CONTRIBUTING.md](https://github.com/supabase/supabase/blob/master/CONTRIBUTING.md)
file.

YES

## What kind of change does this PR introduce?

Blog post on using pg_partman instead of TimescaleDB to prepare for the
upcoming deprecation

## What is the current behavior?

## What is the new behavior?

Blog post to include migration information for those using Timescale

## Additional context

Not to be merged until pg_partman is released in 15 and 17 images


<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

* **Documentation**
* Added a comprehensive pg_partman guide covering setup, time- and
integer-based partitioning, maintenance, automation, and resources.
* Added a migration guide for moving from TimescaleDB hypertables to
native PostgreSQL partitioning using pg_partman.
  * Updated TimescaleDB docs with migration notes and support guidance.
* **New Features**
* Listed pg_partman in the public extensions reference and added
navigation entries linking to the pg_partman guide and migration guide.

<sub>✏️ Tip: You can customize this high-level summary in your review
settings.</sub>
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

---------

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
This commit is contained in:
Chris Gwilliamsandgithub-actions[bot] authored and GitHub committed 2026-03-03 11:47:50 +02:00
1 parent 732180608a
commit 81415c6053
5 files changed
+229

No files matched your search

@@ -1049,6 +1049,10 @@ export const database: NavMenuConstant = {
name: 'Partitioning your tables',
url: '/guides/database/partitions' as `/${string}`,
},
{
name: 'Migrating to pg_partman',
url: '/guides/database/migrating-to-pg-partman' as `/${string}`,
},
{
name: 'Managing connections',
url: '/guides/database/connection-management' as `/${string}`,
@@ -1249,6 +1253,10 @@ export const database: NavMenuConstant = {
name: 'pg_net: Async Networking',
url: '/guides/database/extensions/pg_net' as `/${string}`,
},
{
name: 'pg_partman: Partition management',
url: '/guides/database/extensions/pg_partman' as `/${string}`,
},
{
name: 'pg_plan_filter: Restrict Total Cost',
url: '/guides/database/extensions/pg_plan_filter' as `/${string}`,
@@ -0,0 +1,95 @@
---
id: 'pg_partman'
title: 'pg_partman: partition management'
description: 'Automated partition management'
---
[`pg_partman`](https://github.com/pgpartman/pg_partman) is a Postgres extension that automates the creation and maintenance of partitions for tables using Postgres native partitioning.
## Enable the extension
To enable `pg_partman`, create a dedicated schema for it and enable the extension there.
{/* prettier-ignore */}
```sql
create schema if not exists partman;
create extension if not exists pg_partman with schema partman;
```
## Create a partitioned table
`pg_partman` requires your parent table to already be declared as a partitioned table.
{/* prettier-ignore */}
```sql
create table public.messages (
id bigint generated by default as identity,
sent_at timestamptz not null,
sender_id uuid,
recipient_id uuid,
body text,
primary key (sent_at, id)
)
partition by range (sent_at);
```
## Set up partitioning
You configure the parent table using `partman.create_parent()`. The function takes an `ACCESS EXCLUSIVE` lock briefly while it creates the initial partitions.
### Time-based partitions
{/* prettier-ignore */}
```sql
select partman.create_parent(
p_parent_table := 'public.messages',
p_control := 'sent_at',
p_type := 'range',
p_interval := '7 days',
p_premake := 7,
p_start_partition := '2025-01-01 00:00:00'
);
```
### Integer-based partitions
{/* prettier-ignore */}
```sql
create table public.events (
id bigint generated by default as identity,
inserted_at timestamptz not null default now(),
payload jsonb,
primary key (id)
)
partition by range (id);
select partman.create_parent(
p_parent_table := 'public.events',
p_control := 'id',
p_type := 'range',
p_interval := '100000'
);
```
## Running maintenance
It’s important to call `pg_partman` maintenance regularly so future partitions are pre-created and retention policies are applied.
{/* prettier-ignore */}
```sql
call partman.run_maintenance_proc();
```
To automate this, schedule it using `pg_cron`.
{/* prettier-ignore */}
```sql
create extension if not exists pg_cron;
select
cron.schedule('@hourly', $$call partman.run_maintenance_proc()$$);
```
## Resources
- Official [pg_partman documentation](https://github.com/pgpartman/pg_partman/blob/development/doc/pg_partman.md)
@@ -8,6 +8,10 @@ description: 'Scalable time-series data storage and analysis'
The `timescaledb` extension is deprecated in projects using Postgres 17. It continues to be supported in projects using Postgres 15, but will need to dropped before those projects are upgraded to Postgres 17. See the [Upgrading to Postgres 17 notes](/docs/guides/platform/upgrading#upgrading-to-postgres-17) for more information.
If you are using hypertables, follow the [migration guide](/docs/guides/database/migrating-to-pg-partman) to convert to native partitioning managed by `pg_partman`.
For additional support, contact our Success team by creating a support ticket in the Supabase Dashboard.
</Admonition>
[`timescaledb`](https://docs.timescale.com/timescaledb/latest/) is a Postgres extension designed for improved handling of time-series data. It provides a scalable, high-performance solution for storing and querying time-series data on top of a standard Postgres database.
@@ -0,0 +1,113 @@
---
id: 'migrating-from-timescaledb-to-pg-partman'
title: 'Migrate from TimescaleDB to pg_partman'
description: 'Convert TimescaleDB hypertables to Postgres native partitions managed by pg_partman.'
---
Starting from Postgres 17, Supabase projects do not have the `timescaledb` extension available. If your project relies on TimescaleDB hypertables, you will need to migrate to standard Postgres tables before upgrading.
This guide shows one approach to migrate a hypertable to a native Postgres partitioned table and optionally configure `pg_partman` to automate ongoing partition maintenance.
The approach outlined in this guide can also be used for traditional partitioned tables.
## Before you begin
- Test the migration path in a staging environment (for example by creating a copy of your production project or using branching).
- Review your application for TimescaleDB-specific SQL usage (for example `time_bucket()`, compression policies). Those features are not provided by `pg_partman`.
## Migration overview
1. Create a new partitioned table.
2. Copy data from the hypertable to the new table.
3. Swap over and drop the hypertable.
4. Configure `pg_partman` (optional) and schedule maintenance.
## Example: Migrate `messages` from hypertable to native partitions
This example assumes a `messages` hypertable partitioned by `sent_at`.
### 1. Rename the existing hypertable
This keeps the original data in place while you create a new partitioned table with the original name.
{/* prettier-ignore */}
```sql
alter table public.messages rename to ht_messages;
```
### 2. Create a new partitioned table
When using native partitioning, the partitioning column must be included in any unique index (including the primary key).
{/* prettier-ignore */}
```sql
create table public.messages (
like public.ht_messages including all,
primary key (sent_at, id)
)
partition by range (sent_at);
```
### 3. Copy data into the new table
For large tables, consider copying in batches (for example by time range) during a maintenance window.
{/* prettier-ignore */}
```sql
insert into public.messages
select *
from public.ht_messages;
```
### 4. Drop the old hypertable (and TimescaleDB)
Only drop the extension once you’ve migrated all hypertables and no other objects depend on it.
{/* prettier-ignore */}
```sql
drop table public.ht_messages;
drop extension if exists timescaledb;
```
### 5. Configure `pg_partman` (optional)
Enable `pg_partman` and register your table so partitions are created ahead of time.
{/* prettier-ignore */}
```sql
create schema if not exists partman;
create extension if not exists pg_partman with schema partman;
select partman.create_parent(
p_parent_table := 'public.messages',
p_control := 'sent_at',
p_type := 'range',
p_interval := '7 days',
p_premake := 7,
p_start_partition := '2025-01-01 00:00:00'
);
```
## Keep partitions up to date
`pg_partman` requires running maintenance to pre-make partitions and apply retention policies.
{/* prettier-ignore */}
```sql
call partman.run_maintenance_proc();
```
To automate this, schedule it with `pg_cron`.
{/* prettier-ignore */}
```sql
create extension if not exists pg_cron;
select cron.schedule('@daily', $$call partman.run_maintenance_proc()$$);
```
## Additional resources
- [Partitioning your tables](/docs/guides/database/partitions).
- [`pg_partman` documentation](/docs/guides/database/extensions/pg_partman)
- [`pg_partman` migration guides](https://github.com/pgpartman/pg_partman/blob/development/doc/migrate_to_partman.md)
+9
View File
@@ -278,6 +278,15 @@
"product": "Database Webhooks",
"product_url": "/project/{ref}/integrations/webhooks"
},
{
"name": "pg_partman",
"comment": "Automated partition management",
"tags": ["Utility"],
"link": "/guides/database/extensions/pg_partman",
"github_url": "https://github.com/pgpartman/pg_partman",
"product": null,
"product_url": null
},
{
"name": "pg_prewarm",
"comment": "prewarm relation data",