Gorm automigrate and the db-operations pod were a good stop-gap for our budding service. As we mature we need better, repeatable processes, that can be used across all envs with backwards compatibility and rollback functionality.

Ideally this will be a process that lives in the inventory repo where engineers can write their migrations as individual files and commit them with their feature changes.

See https://gitlab.cee.redhat.com/service/uhc-account-manager/-/blob/master/pkg/db/README.md?ref_type=heads for an example on how OCM accomplishes this