spicedb itself can be given a schema + relationships on startup to be imported

Sorry for replying on an old thread. What would be the suggested way for production using k8s? Let say I have the schema in git and I want to rollout a release. I could add the schema to a yaml, put it into a k8s configmap and I can mount that and pass the file to the flag. The yaml would only have the schema (no relations, assertions, etc). I understand that I would need to test the new schema with the existing relations, but would that otherwise be a good approach?

the recommended approach would be to use the SpiceDB operator to run the instance

This is a bit orthogonal to deploying the code. I'll think a bit about what it would take to have a declarative schema management. Not sure if this was considered for the operator (e.g. have a crd for the schema including versioning)