Upgrading
Supabase ships fast and we endeavor to add all new features to existing projects wherever possible. In some cases, access to new features require upgrading or migrating your Supabase project.
You can upgrade your project using pg_upgrade
or by pausing and restoring your project.
pg_upgrade
This upgrade method is currently in Beta.
For security purposes, passwords for custom roles are not backed up and, following a restore, they would need to be reset. See here for more details
pg_upgrade performs an in-place upgrade on your database. For projects larger than 1GB, pg_upgrade is generally faster than a pause+restore cycle, and the speed advantage grows with the size of the database.
- Plan for an appropriate downtime window, and ensure you have reviewed the caveats section of this document before executing the upgrade.
- Use the "Upgrade project" button on the Infrastructure section of your dashboard.
Additionally, if a pg_upgrade upgrade should fail, your original DB would be brought back up online and be able to service requests.
As a rough rule of thumb, pg_upgrade operates at ~100mbps (when executing an upgrade on your data). Using the size of your database, you can use this metric to derive an approximate sense of the downtime window necessary for the upgrade. During this window, you should plan for your DB and associated services to be unavailable.
Pause and restore
We recommend using the pg_upgrade method, as it is faster, and more reliable. Additionally, only Free-tier projects are eligible to use the Pause + Restore method.
When you pause and restore a project, the restored database includes the latest features. This method does include downtime, so be aware that your project will be inaccessible for a short period of time.
- On the General Settings page in the Dashboard, click Pause project. You will be redirected to the home screen as your project is pausing. This process can take several minutes.
- After your project is paused, click Restore project. The restoration can take several minutes depending on how much data your database has. You will receive an email once the restoration is complete.
Note that a pause + restore upgrade involves tearing down your project's resources before bringing them back up again. If the restore process should fail, manual intervention from Supabase support will be required to bring your project back online.
Caveats
Regardless of the upgrade method, a few caveats apply:
Logical replication
If you are using logical replication, the replication slots will not be preserved by the upgrade process. You will need to manually recreate them after the upgrade with the method pg_create_logical_replication_slot
. Refer to the Postgres docs on Replication Management Functions for more details about the method.
Breaking changes
Newer versions of services can break functionality or change the performance characteristics you rely on. If your project is eligible for an upgrade, you will be able to find your current service versions from within the Supabase dashboard.
Breaking changes are generally only present in major version upgrades of Postgres and PostgREST. You can find their respective release notes at:
If you are upgrading from a significantly older version, you will need to consider the release notes for any intermediary releases as well.
Time limits
Starting from 2024-06-24, when a project is paused, users then have a 90-day window to restore the project on the platform from within Supabase Studio.
The 90-day window allows Supabase to introduce platform changes that may not be backwards compatible with older backups. Unlike active projects, static backups can't be updated to accommodate such changes.
During the 90-day restore window a paused project can be restored to the platform with a single button click from Studio's dashboard page.
After the 90-day restore window, you can download your project's backup file, and Storage objects from the project dashboard. See restoring a backup locally for instructions on how to load that backup locally to recover your data.
If you upgrade to a paid plan while your project is paused, any expired one-click restore options are reenabled. Since the backup was taken outside the backwards compatibility window, it may fail to restore. If you have a problem restoring your backup after upgrading, contact Support.
Restoring a downloaded backup locally
If the 90 day project restore window has expired but you need to access data contained within your project using SQL, you can attempt to restore the project into a local Postgres instance. Supabase publishes tooling that can be used for that purpose. Be aware that this workflow does not produce a complete Supabase environment with REST/Auth/Storage. Instead, it creates a standalone Postgres instance that is maximally compatible with your project's backup file to assist with recovering your data.
Downloaded backups are plain text SQL files that can be executed in Postgres to restore your project's schema and data. If you need to access data within a backup we recommend restoring into a Postgres instance built using the Supabase open source build tooling to ensure that the local instance is as similar to the original project as possible. Here are the steps to start Supabase Postgres locally and restore the backup:
Install nix:
_10curl --proto '=https' --tlsv1.2 -sSf -L https://install.determinate.systems/nix | sh -s -- install
To identify your project's backup image version, reference the backup's download button following the PG:
prefix:
Given Postgres version 15.6.1.115
, start Postgres locally with the nix command:
_10nix run 'github:supabase/postgres?ref=15.6.1.115#start-server' 15
where 15
refers to the backup's Postgres major version.
Note that the earliest Supabase Postgres version that supports a direct reference is 15.6.1.115
. If you don't know your project's Postgres version or it is lower than 15.6.1.115
then omit the ref
option to build our most recent image:
_10nix run 'github:supabase/postgres#start-server' -- 15
Then restore the backup file from a separate shell session:
_10nix run github:supabase/postgres#start-client -- --file <BACKUP FILE>
Which loads the backup into Postgres and drops you into a psql prompt. For start-client
you do not need to provide a reference to the backup's Postgres version.
Once your project has been restored, connect to it with your preferred Postgres tooling using host=localhost
, user=postgres
, port=5432
, and an empty password. This will allow you to inspect your saved data and export whatever is necessary.
Errors encountered during the restore process are logged to the console by the start-server
process. Review those errors and edit the backup file as needed until the project restores successfully. The backup file is a plain text sequence of SQL statements that recreate your database. To resolve common error classes like relation "xxx" does not exist
, identify the line raising the exception and shuffle the create ... "xxx"
statement before it. A failed restore leaves artifacts on the Postgres server so be sure to restart the start-server
process between restore attempts.
A Postgres database started with start-server
command is not production ready and should never be used outside a local development workflow.
Disk sizing
When upgrading, the Supabase platform will "right-size" your disk based on the current size of the database. For example, if your database is 100GB in size, and you have a 200GB disk, the upgrade will reduce the disk size to 120GB (1.2x the size of your database).
Objects dependent on Postgres extensions
pg_upgrade does not support upgrading of databases containing reg* data types referencing system OIDs. If you have created any objects that depend on the following extensions, you will need to recreate them after the upgrade.
pg_cron
records
pg_cron does not automatically clean up historical records. This can lead to extremely large cron.job_run_details
tables if the records are not regularly pruned; you should clean unnecessary records from this table prior to an upgrade.
During the pg_upgrade
process, the pg_cron
extension gets dropped and recreated. Prior to this process, the cron.job_run_details
table is duplicated to avoid losing historical logs. The instantaneous disk pressure created by duplicating an extremely large details table can cause at best unnecessary performance degradation, or at worst, upgrade process failures.
Extensions
pg_upgrade does not currently support upgrading of databases using extensions older than the following versions:
- TimescaleDB 2.16.1
- plv8 3.1.10
To upgrade to a newer version of Postgres, you will need to drop the extensions before the upgrade, and recreate them after the upgrade.
Authentication method changes - deprecating md5 in favor of scram-sha-256
The md5 hashing method has known weaknesses that make it unsuitable for cryptography. As such, we are deprecating md5 in favor of scram-sha-256, which is the default and most secure authentication method used in the latest Postgres versions.
We automatically migrate Supabase-managed roles' passwords to scram-sha-256 during the upgrade process, but you will need to manually migrate the passwords of any custom roles you have created, else you won't be able to connect using them after the upgrade.
To identify roles using the md5 hashing method and migrate their passwords, you can use the following SQL statements after the upgrade:
_10-- List roles using md5 hashing method_10SELECT_10 rolname_10FROM pg_authid_10WHERE rolcanlogin = true_10 AND rolpassword LIKE 'md5%';_10_10-- Migrate a role's password to scram-sha-256_10ALTER ROLE <role_name> WITH PASSWORD '<password>';
Database size reduction
As part of the upgrade process, maintenance operations such as vacuuming are also executed. This can result in a reduction in the reported database size.
Post-upgrade validation
Supabase performs extensive pre- and post-upgrade validations to ensure that the database has been correctly upgraded. However, you should plan for your own application-level validations, as there might be changes you might not have anticipated, and this should be budgeted for when planning your downtime window.