Hasura Cloud - Custom Supabase Schemas Not Showing Up

[oalexdoda]

Hey,

We're using Hasura with Supabase to consume and secure the data. When using our custom schema and its tables, the schema does not appear in the Hasura Cloud interface.

When trying to create it though, it says "Schema already exists", so it is able to read it from Postgres.

Any clue if there's an extra step we need to take to have the schema be accessible via the API using the "service_role" properly? It only shows the default schemas, but none of the custom ones, even if all permissions were set, etc.

Is this a Hasura Cloud UI bug?

[seanonthenet]

I'm not familiar with Supabase at all, but from their site it looks like it should support standard Postgres databases.
Do you have an existing database connection in that Hasura instance?
Are you able to access these custom schemas with another SQL client?

[akhil-naidu]

If you are connecting to a Postgres dB with some data or tables, you need to step ahead and tell Hasura which tables you would like to track using Hasura.

In the case of Subpabase, you will have some prepopulated tables, so this is what you have to follow, these steps.

All of your traditional tables, other than features provided by the subpabase will be stored in the Public Folder

Click on track all or track only specific tables for which you would like to create GraphQL APIs, which are also termed as Hasura Views

A quick comparison of Hasura API explorer(Which allows you see all the Hasura views) before and after tracking

Before Tracking the table test

After Tracking the table test

---

Note: Hasura allows you to create and configure your tables from the Hasura console itself, but for some reason, if you prefer using Supabase to create and configure your table schemas, it is mandatory to track them back in the Hasura console.

The reason behind this architectural decision is practical for the majority of the Postgres developers; not every Postgres developer expects to have a CRUD operation on every existing Postgres table. Tracking tables individually would give them more control on whether to expose a specific table to the Pubic GraphQL API or not.