How to configure multiple data sources on startup?

[paymog]

I'm currently running the community edition of hasura in my infrastructure and I'm using HASURA_GRAPHQL_DATABASE_URL. We're now upgrading our infra and are examining ways of configuring multiple data sources.

I found these docs but I can't quite figure out how to wire up a config file like that. I notice that the title of the section mentions using the cli and api. I've also searched for how to "point" to a config file at startup with the docker image but haven't been successful.

Is there a way to configure hasura at startup with more than 1 data source?

[SamirTalwar]

Data sources are configured at runtime by adding them using the metadata API. There's no way of pointing Hasura GraphQL Engine to a configuration file at startup because the configuration is stored in the metadata database. However, you can use the CLI to apply a configuration to a running HGE instance.

As you've been using HASURA_GRAPHQL_DATABASE_URL, Hasura is configured the first time it starts up with that database as both the metadata DB and the source DB named "default". When you start it again later, the URL will be used for the metadata database only. HGE will then read the metadata, which already contains the same URL for the "default" source DB.

In order to add a new data source to an existing HGE instance, you can use the pg_add_source API endpoint (or equivalent for another database), or use the Hasura Console to the same thing (under the "Data" tab).

Once you've done this, you can set up the Hasura CLI and run hasura metadata export to download the current metadata as a set of YAML files. You can then run hasura metadata apply to re-apply this configuration to a running HGE instance.

[cmartin616]

Additionally, you can specify any plaintext named environment variable (I'd recommend not starting HASURA_*) for each of your databases. For example:

• PG_APP_DB_X: [conn_string]
• PG_APP_DB_X_READ_REPLICA: [conn_string]
• PG_APP_DB_Y: [conn_string]
• ...

Then configure your data sources to use the environment variable rather than a hard-coded value in the metadata. This is also essential for devops/tier promotions because you need to be able to change the value of the database connection string for each tier without altering the metadata. This allows for the same metadata to be used in each environment where each environment has different environment variables.