The virtual accelerator's archive The recorder samples the running virtual accelerator every ten seconds and writes what answered into a MongoDB collection; the project's archiver connector reads history back out of that same collection, so what the agent plots is what the deployment recorded. The first deploy seeds a base series thirty days back, and the seeded past joins the recorded present on the same timestamps. first deploy seeds 30 days back virtual accelerator the simulated machine recorder every 10 seconds archiver-mongodb one collection archiver connector mongodb_archiver Osprey agent plots the history written by the recorder, read back by the connector SEEDED PAST 30 days — the last 48 h every 10 s, the rest every 60 RECORDED PRESENT every 10 s from deploy on the join is meant to be invisible — same timestamps, same baselines, noise rather than a step on control_system: mock the recorder idles by design — history stays readable and the archive reports stale, which is a warning, not an error