Files
supabase/apps/docs/content/guides/self-hosting/postgres-upgrade-17.mdx
T
Miranda Limonczenko 7ce4ee53ae chore(docs) Retire supa-mdx-lint (#50602)
Closes
[DOCS-1289](https://linear.app/supabase/issue/DOCS-1289/get-the-linter-to-fix-what-it-flags-or-retirereplace-the-linter)

Stacked on #50600, which points contributors at the authoring skills.
Merge that one first.

## Problem

Contributors experienced friction with the linter. They felt nickle and
dimed for tiny nits and felt detracted from the work itself. PRs would
become noisy with tiny one-word suggestions.

Additionally, our homegrown linter is not very intelligent, causing
frequent overrides.

## Solution

This removes the linter entirely in favor of directing contributors to
use SKILLS instead.

The removal entails...

- **CI.** Delete the three `docs_lint` workflows: the PR check, the
external-PR comment companion, and the nightly `--fix` bot. Drop the
stale `zizmor.yml` ignore entry for the deleted workflow.
- **Tooling.** Delete `supa-mdx-lint.config.toml` and the 14 rule files.
Drop the `lint:mdx` script and the `@supabase/supa-mdx-lint` dependency
from docs, learn, and ui-library, and regenerate the lockfile.
- **Content.** Remove the 181 directives. A separate commit carries
Prettier's reformatting of the tables and blank lines those comments had
suppressed, so the deletion commit stays readable. No prose changes.
- **Style guide.** The word list states each rule directly instead of
describing what the linter flagged. Every term survives, including the
phrase groups that mirrored `Rule004ExcludeWords`.
- **Skills.** `write-the-docs`, `edit-the-docs`, and `review-the-docs`
drop `pnpm lint:mdx` from their self-review commands and check the word
list directly. `ask-the-docs`'s CI reference drops both workflows.

## Manual testing

1. Run `git grep -i supa-mdx-lint -- . ':!pnpm-lock.yaml'`. No matches.
2. Run `pnpm install --frozen-lockfile --lockfile-only`. It passes, so
the lockfile matches the three trimmed manifests.
3. Run `git diff master...HEAD --name-only --diff-filter=ACMR | grep -E
'\.(md|mdx)$' | xargs npx prettier --config prettier.config.mjs
--check`. All changed markdown passes.
4. Open the [reformatted filter
table](https://docs-git-docs-retire-mdx-linter-supabase.vercel.app/docs/guides/observability/logs#filter-events)
on the preview and compare it with
[production](https://supabase.com/docs/guides/observability/logs#filter-events).
The table renders the same.

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

* **Documentation**
* Documentation guidance now uses manual prose and terminology review
with the shared word list.
* Clarified storage configuration and common Realtime channel mistakes.
* Improved table formatting, text wrapping, and selected reference
links.
  * Updated documentation authoring and review guidance.

* **Chores**
* Retired automated MDX linting from workflows and local validation
commands.
* Removed lint-suppression markers throughout documentation without
changing instructions.
  * Added targeted documentation review guidance for pull requests.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
2026-09-22 10:00:41 -07:00

298 lines
12 KiB
Plaintext

---
title: 'Upgrade to Postgres 17'
description: 'Start a new self-hosted Supabase deployment with Postgres 17, or upgrade an existing Postgres 15 installation.'
subtitle: 'Start a new self-hosted Supabase deployment with Postgres 17, or upgrade an existing Postgres 15 installation.'
---
Self-hosted Supabase ships with Postgres 17 by default. This guide covers two scenarios:
- **New deployment** - start fresh with Postgres 17 (no existing data)
- **Upgrade existing deployment** - migrate from Postgres 15 to Postgres 17 using `pg_upgrade`
## New deployment with Postgres 17
Postgres 17 is the default, so a new self-hosted instance with **no existing data** starts on Postgres 17 - no override needed:
```sh
sh run.sh start
```
The rest of the setup is unchanged (see the Docker install [guide](/docs/guides/self-hosting/docker)).
<Admonition type="note" title="GraphQL is off by default">
On a fresh Postgres 17 deployment, the `pg_graphql` extension is **disabled by default**. Enable it from Studio (**Database** > **Extensions**) or with `create extension pg_graphql;`. Databases that already use GraphQL keep it after an upgrade.
</Admonition>
If the new Postgres 17 container fails to start, make sure to check for an old `db-config` Docker volume. See [Postgres 17 fails to start with a leftover db-config volume](#postgres-17-fails-to-start-with-a-leftover-db-config-volume) for details.
## Upgrade an existing Postgres 15 deployment
Upgrading an existing deployment uses `pg_upgrade` to migrate data in place. The included upgrade scripts automates the full process.
### What the upgrade does
1. Pulls a specific Postgres 17 image and extracts upgrade binaries
2. Pulls supplemental upgrade scripts from Supabase's [Postgres](https://github.com/supabase/postgres) repository
3. Stops all self-hosted Supabase containers
4. Runs `pg_upgrade` inside a temporary Postgres 15 container
5. Runs additional tasks inside a temporary Postgres 17 container (re-enables extensions, applies patches, runs `VACUUM ANALYZE`)
6. Swaps data directories (the original is kept as a backup)
7. Starts self-hosted Supabase with Postgres 17
8. Applies post-upgrade migrations and reconciles extension versions to the target image
### Create a backup
<Admonition type="danger" title="Back up your data before upgrading">
You should create your own independent backup in case of disk failure or other issues.
</Admonition>
The upgrade script automatically preserves the pgsodium key and original data directory as `./volumes/db/data.bak.pg15` as the final step. However, it is recommended to **always** create your own independent backup before starting:
Back up the database data directory:
```sh
cp -a ./volumes/db/data ./volumes/db/data-manual-backup
```
Back up the pgsodium encryption key (stored in a Docker named volume):
```sh
docker compose run --rm db cat /etc/postgresql-custom/pgsodium_root.key > ./pgsodium_root.key.backup
```
<Admonition type="caution">
The `db-config` Docker named volume contains the pgsodium root encryption key. If you lose this key and have vault secrets, they become unrecoverable. The `cp -a` above backs up the data directory but NOT the named volume.
</Admonition>
Optionally, take a logical backup too:
```sh
docker exec supabase-db pg_dumpall -h localhost -U supabase_admin > ./pg15_dump.sql
```
### Requirements
- At least **2x your current database size + 5 GB** of free disk space (`pg_upgrade` copies the data directory; the upgrade tarball is ~1.2 GB compressed)
- The script prompts for confirmation at each major step (use `--yes` to skip prompts)
- All self-hosted Supabase containers must be running before starting the upgrade
- Requires `bash`
- Must be run as root or using `sudo`
### Extensions removed in Postgres 17
The following extensions are **not available** in Postgres 17 builds. The upgrade script will prompt you to drop them if any of these are found:
| Extension | Notes |
| ------------- | ------------------------- |
| `timescaledb` | Not built for Postgres 17 |
| `plv8` | Not built for Postgres 17 |
| `plcoffee` | Companion to plv8 |
| `plls` | Companion to plv8 |
None of the above extensions are installed by default in the self-hosted Supabase setup. If you have installed any of them manually and need to keep them, **do not proceed** with the upgrade.
### Run the upgrade
```sh
sudo bash utils/upgrade-pg17.sh
```
The script might require your confirmation at some steps (e.g., while checking for disk space, or whether to disable extensions, or remove previous backups).
### After the upgrade
To verify that Postgres 17 is running:
```sh
docker compose exec db psql -U postgres -c "SELECT version();"
```
The original Postgres 15 data is preserved at `./volumes/db/data.bak.pg15`. The pgsodium root key is saved as `./volumes/db/pgsodium_root.key.bak.pg15`. The upgrade binaries tarball is cached at `./volumes/db/pg17_upgrade_bin_*.tar.gz`. Once you have verified that everything works, you can reclaim disk space:
```sh
rm -rf ./volumes/db/data.bak.pg15 \
./volumes/db/pgsodium_root.key.bak.pg15 \
./volumes/db/pg17_upgrade_bin_*.tar.gz
```
<Admonition type="caution">
Do not delete `data.bak.pg15` until you have verified the upgrade. Rollback is only possible while the backup exists.
</Admonition>
### Rollback
If you need to revert to Postgres 15 (run the following commands as root):
```sh
docker compose down && \
rm -rf ./volumes/db/data && \
mv ./volumes/db/data.bak.pg15 ./volumes/db/data && \
sh run.sh config add pg15 && \
docker compose run --rm db chown -R postgres:postgres /etc/postgresql-custom/ && \
sh run.sh start
```
This restores the original data directory, fixes file ownership on the `db-config` volume, and configures Supabase to use the old Postgres 15 image. Supabase's Postgres 15 and 17 images use different user IDs, so the `chown` and startup must run as Postgres 15.
### Custom Postgres configuration
The Postgres 17 image loads any `.conf` files from `/etc/postgresql-custom/conf.d/` on startup. This directory is on the `db-config` named volume, so changes persist across restarts.
<Admonition type="note">
This is a Supabase Postgres 17 image feature. The Postgres 15 image does not load files from `conf.d/`.
</Admonition>
To add custom Postgres settings, create a `.conf` file in the volume. Since `conf.d/` is on a Docker named volume (not a bind mount), you need to write through the container:
```sh
docker exec supabase-db bash -c 'cat > /etc/postgresql-custom/conf.d/custom.conf << EOF
max_connections = 200
EOF'
```
Restart to apply (`max_connections` requires a full restart):
```sh
sh run.sh restart db
```
Verify the new settings:
```sh
docker compose exec db psql -U postgres -c "SHOW max_connections;"
```
### Upgrade process details
The upgrade script delegates the core migration work to two scripts from the [supabase/postgres](https://github.com/supabase/postgres) repository (`ansible/files/admin_api_scripts/pg_upgrade_scripts/`).
**Phase 1 - Migrate data (Postgres 15 container):**
1. Disables extensions that are incompatible with `pg_upgrade` (`pg_graphql`, `pg_stat_monitor`, `pg_backtrace`) and generates SQL to re-enable them after the upgrade
2. Temporarily grants superuser to the `postgres` role (required by `pg_upgrade`)
3. Extracts the previously saved Postgres 17 binaries tarball and runs `initdb` to create a new empty database
4. Runs `pg_upgrade --check` to verify the upgrade can succeed before making changes
5. Stops Postgres 15 and runs `pg_upgrade` to migrate all data to the new database
6. Copies Postgres configuration and the SQL scripts generated by `pg_upgrade` to a staging directory for the next phase
**Phase 2 - Finalize (Postgres 17 container):**
1. Moves the upgraded data directory into place and starts Postgres 17
2. Applies extension compatibility patches for Wrappers, `pg_net`, `pg_cron`, and Vault (fixes ownership, grants, and foreign server options)
3. Runs the SQL scripts generated by `pg_upgrade` to update system catalogs and extension versions
4. Re-enables the extensions that were disabled in phase 1
5. Grants predefined roles (`pg_monitor`, `pg_read_all_data`, `pg_signal_backend`, and on Postgres 16+ also `pg_create_subscription`) and revokes the temporary superuser grant
6. Restarts Postgres and runs `vacuumdb --all --analyze-in-stages` to rebuild optimizer statistics
After both phases, the upgrade script applies migrations that normally run only on a fresh install, and reconciles installed extensions to the target image's versions. It then preserves the original Postgres 15 data directory as a backup and starts the full Supabase stack with Postgres 17.
## Troubleshooting
### pg_upgrade fails with replication slot errors
`pg_upgrade` cannot proceed if there are active replication slots. Default self-hosted installs don't have any, but if you set up logical replication or have custom replication configurations, drop the slots before upgrading:
```sh
docker exec supabase-db psql -h localhost -U supabase_admin -d postgres -c "
SELECT pg_drop_replication_slot(slot_name)
FROM pg_replication_slots;
"
```
Then re-run the upgrade script. Replication slots will need to be manually recreated after the upgrade.
### "Permission denied" on the data directory
The upgrade script fixes file ownership automatically (Postgres 15 and 17 use different UIDs). If you still see permission errors, run:
```sh
docker compose run --rm db \
chown -R postgres:postgres /var/lib/postgresql/data
```
### pgsodium / Supabase Vault errors
The `db-config` named volume contains the pgsodium root encryption key at `/etc/postgresql-custom/pgsodium_root.key`. This volume is preserved during the upgrade. Never run `docker compose down -v` as this destroys named volumes and makes vault secrets unrecoverable.
### Services fail to connect after upgrade
Restart all services to pick up the new database:
```sh
sh run.sh recreate
```
### Disk space issues during upgrade
The upgrade needs space for:
- The upgrade tarball (~1.2 GB compressed, cached for re-runs)
- A full copy of your database (created by `pg_upgrade`)
- The original data (kept as backup)
The script uses `/tmp` (or `TMPDIR` if set) for its staging directory, which holds the downloaded tarball and upgrade scripts. If your `/tmp` filesystem is small or mounted with limited space, you can point it to a different location, e.g.:
```sh
sudo TMPDIR=/mnt/my-tmp bash utils/upgrade-pg17.sh
```
If you run out of space mid-upgrade, the safest path is to roll back and free up disk space before retrying.
### Postgres 17 fails to start with a leftover db-config volume
If you are starting a **fresh** Postgres 17 deployment (not using the upgrade script) and the container fails to start, the most likely cause is a leftover `db-config` volume from a previous Postgres 15 installation. Start the containers without the `-d` option or check the logs for errors about `postgresql.conf` or other configuration mismatch.
To fix, remove the old volume and let Postgres 17 initialize a clean configuration:
```sh
sh run.sh stop && \
docker volume rm $(docker volume ls --filter "name=db-config" --format '{{.Name}}') && \
sh run.sh start
```
<Admonition type="caution">
Removing the `db-config` volume destroys any custom Postgres configuration and the pgsodium root key. Only do this for fresh installations with no existing data or vault secrets.
</Admonition>
### Restoring from a manual backup
If the upgrade fails and the script's built-in rollback isn't sufficient, restore from the manual backups created in the [Create a backup](#create-a-backup) step:
Restore the data:
```sh
docker compose down && \
rm -rf ./volumes/db/data && \
cp -a ./volumes/db/data-manual-backup ./volumes/db/data && \
sh run.sh config add pg15
```
Restore the pgsodium key to the `db-config` volume:
```sh
docker compose run --rm db \
sh -c 'cat > /etc/postgresql-custom/pgsodium_root.key' < ./pgsodium_root.key.backup && \
docker compose run --rm db \
chown -R postgres:postgres /etc/postgresql-custom/ && \
docker compose run --rm db \
chmod 600 /etc/postgresql-custom/pgsodium_root.key
```
Start self-hosted Supabase with Postgres 15:
```sh
sh run.sh start
```