This pull request tries to solve the issue in the provisioning development setup. The admin service needs the openFGA store ID and model ID to be passed in.
The proposed setup uses a k8s job, instead of using init containers in Admin UI deployment, to create openFGA store and model and share them with the admin service by using persistent volume. This is specifically to avoid potential race conditions that could be caused by multiple Admin UI replicas.
This pull request tries to solve the issue in the provisioning development setup. The admin service needs the openFGA store ID and model ID to be passed in.
The proposed setup uses a k8s job, instead of using init containers in Admin UI deployment, to create openFGA store and model and share them with the admin service by using persistent volume. This is specifically to avoid potential race conditions that could be caused by multiple Admin UI replicas.