[Question] How to handle/init/persist migrations in Docker Image

If someone is using Hasura (cli-migrations) docker image, what is the recommended way to init migrations and share metadata and migrations with host?

I could not find anything related to docker management in the documentation. All information is about when running docker locally.

[GavinRay97]

The workflow you can use with Docker/Kubernetes and the .cli-migrations image is to make a custom Dockerfile where you copy the migrations folder from the host to the container and let it apply them.

If you are using the V2 migrations format:

FROM hasura/graphql-engine:v1.2.0-beta.3.cli-migrations-v2
COPY migrations /hasura-migrations
COPY metadata /hasura-metadata

If you are using the V1 format:

FROM hasura/graphql-engine:v1.2.0-beta.3.cli-migrations
COPY migrations /hasura-migrations

Assuming that your migrations/metadata are stored in those directories. You can also customize the location from which Hasura reads migrations/metadata by setting environment variables:

HASURA_GRAPHQL_MIGRATIONS_DIR=/custom-path-for-migrations
# V2 Only
HASURA_GRAPHQL_METADATA_DIR=/custom-path-for-metadata

The workflow looks like this:

• Dockerfile copies metadata/migrations automatically
• .cli-migrations image applies them on start
• Use the Hasura CLI to make changes to your schema/metadata
  • They will be applied now on your instance but will also have the migrations generated for them
  • So creating a fresh container instance will replicate the same state as your current one

Do you think the documentation would be improved by putting more information about the specifics of the entire workflow, outside of what pertains to CI/CD here?

https://hasura.io/docs/1.0/graphql/manual/migrations/auto-apply-migrations.html#

[mubaidr]

Got it running. I need to use 0.0.0.0 to make sure it is available outside the container. (localhost is not)

hasura-cli console --address 0.0.0.0

Thanks @GavinRay97