Connect Supabase

DBConvert connects to Supabase as to any PostgreSQL server. For the generic workflow, see Source and destination connections.

• Which endpoint to use
• Get the database password
• Connection fields
• The errors you will hit
• Connection limits on small projects
• SSL
• Backups on the Free plan
• Coming from Lovable Cloud

Which endpoint to use

Open Connect in the top bar of your project and pick the Direct connection string tab. It offers three connection methods, and they are not interchangeable:

Use the session pooler unless your network has IPv6. All three endpoints resolve to IPv6 by default, and only the session pooler is proxied over IPv4 for free - Supabase sells a dedicated IPv4 address as an add-on for the other two. Most Windows desktops on an office or home connection have no IPv6 route, which makes the session pooler the only endpoint that answers.

Note the username. The pooler identifies your project from the user name, so it must be postgres.<ref>, where <ref> is the project reference from the host name (also shown as Project ID in Project Settings). Plain postgres only works on the direct host.

Get the database password

Supabase shows the database password once, when the project is created, and never again. From the docs page in the dashboard: "The database password isn't viewable after creation. Resetting it will break any existing connections."

If you do not have it, reset it - the button sits in the Connect dialog next to the connection string, and again under Database → Settings → Database password. Two things to know before you click:

• The reset invalidates the old password, so anything already connected with it stops working. On a project that is serving an application, plan for that.
• The new password is displayed only until you reload the page - the dialog labels it "New password shown until refresh". Copy it immediately, or you will be resetting it again.

Connection fields

For the session pooler, fill in the DBConvert connection form like this:

Click Connect before going further. A correct session pooler setup reports that it connected.

The errors you will hit

Connection timed out

could not connect to server: Connection timed out (0x0000274C/10060)
Is the server running on host "db.<ref>.supabase.co"
(2a05:...) and accepting TCP/IP connections on port 5432?

This is the direct host on a network without IPv6. The address in brackets is the giveaway: it is an IPv6 address, because that is the only kind the direct host publishes. Your machine has no route to it.

Switch to the session pooler host and remember to change the username as well.

No tenant identifier provided

FATAL: (ENOIDENTIFIER) no tenant identifier provided
(external_id or sni_hostname required)

This is the pooler host with the username left as plain postgres. Change the username to postgres.<ref>; the host, port and password stay as they are.

Password authentication failed for user "postgres"

failed SASL auth: FATAL: password authentication failed
for user "postgres" (SQLSTATE 28P01)

Read this one carefully, because it names the wrong culprit. You typed postgres.<ref>, but the pooler strips the project reference out of the username and authenticates against Postgres as plain postgres - so the name in the message says nothing about what you entered. The password is what is wrong.

Almost always it is a password that has since been reset: any reset takes effect immediately and invalidates the one saved in your tool. Copy the connection string again with the Copy button rather than selecting it by eye - the field is visually truncated, so a hand-copied password loses its tail. And if the password contains @ : / ? # [ ] %, either percent-encode it in the URI or skip the URI and fill the fields in by hand.

Connection limits on small projects

Under Database → Settings → Connection pooling, a project on the smallest compute size (Nano, which is what the Free plan gives you) defaults to a pool size of 15 connections, with max client connections fixed at 200. Both scale with compute size; the pool size is editable, the client limit is not.

This matters because DBConvert can process large tables in parallel, opening more than one connection at a time. On a Nano project the pool is shallow, so if a large job behaves oddly, lower the parallelism before looking anywhere else.

SSL

Under Database → Settings → SSL configuration, Enforce SSL on incoming connections is off by default, so a plain connection is accepted. For anything holding real data, turn it on and download the certificate from the same panel - see Secure connections for where the certificate goes.

Backups on the Free plan

Under Database → Backups, a Free project says "Free Plan does not include project backups". Paid plans get 7, 14 or 30 days of daily backups depending on tier, and those restore into a project through the dashboard rather than downloading to you.

Copying the data into a database you run yourself is the part DBConvert covers - see Supabase migration and export tools.

Coming from Lovable Cloud

Related

• Supabase migration and export tools
• Connect PostgreSQL
• Secure connections (SSH, SSL, PuTTY)
• Source and destination connections