All configuration loading and persistence is currently focused entirely on Kubernetes deployments (ConfigMaps all the way down). This is great for a certain chunk of Prometheus users, but... suboptimal, we'll say... for others.
Config loading / persistence should be abstracted away behind an interface to allow varying implementations for alternate orchestration schemes and bare metal deployments.
All configuration loading and persistence is currently focused entirely on Kubernetes deployments (ConfigMaps all the way down). This is great for a certain chunk of Prometheus users, but... suboptimal, we'll say... for others.
Config loading / persistence should be abstracted away behind an interface to allow varying implementations for alternate orchestration schemes and bare metal deployments.