Problem: there is not an easy way to load initial data into the system

[jraddaoui]

Is your feature request related to a problem? Please describe.

The current workflow requires a default location in the database in order to store and/or proxy package downloads. In the development environment we work around this issue by using a Kubernetes job that makes sure a default location is created in the enduro_storage.location table.

• For a3m (MinIO locations): https://github.com/artefactual-sdps/enduro/blob/main/hack/kube/base/mysql-create-locations-job.yaml
• For AM (AMSS location): https://github.com/artefactual-sdps/enduro/blob/main/hack/kube/overlays/dev-am/mysql-create-amss-location-job.yaml

However, in other environments that location needs to be created manually in the DB.

Describe the solution you'd like

Just an idea to investigate and make sure it's feasible for production environments.

I'd like a way to indicate an initial data file/folder that's read by Enduro and makes sure that data is added to the persistence layer in an initial installation.

Describe alternatives you've considered

• Data migrations (non configurable)
• Load through application config (similar approach but not as versatile)

Additional context

Some examples I recently came up with:

• SFTPGo allows loading initial users, admins, etc. with a JSON file and an environment variable (SFTPGO_LOADDATA_FROM) (I used it in this Gist to connect Enduro and AM locally).
• Keycloak allows to configure a realm, clients, users, etc. with a similar JSON file located on an specific path and a flag in the start command, --import-realm (among other options).