Grafana zero-downtime data source migration: file-based provisioning reload or HTTP API partial updates?
I'm planning a migration of a small Grafana instance (single node, Grafana 10.x assumed) to a new back-end data source endpoint, and the goal is to keep all dashboards queryable throughout the switch with no restart and no visible gap for viewers. Two documented approaches seem viable, and I'm unsure which fits this constraint better. File-based provisioning