M03 · Data-centre telemetry · infra health scoring
Environment diagnostics
Establish whether the environment can actually run the system before anything else.
$ python -m fde_toolkit diagnose
Workflow
the order an FDE actually runs it in
- 1Collect the normalised infrastructure vector: CPU, memory, network, latency, error, loss, temperature, GPU, IO, pods.
- 2Score health and rank the dominant degradation term.
- 3Correlate paired signals (memory pressure with pod restarts, latency with packet loss).
- 4Emit the recovery workaround with a timebox.
- 5Fall back to the hosted environment when local recovery exceeds the box.
Function output
deterministic trace
CPU 72% MEM 81% NET 63% LAT 410ms LOSS 2.1% PODS restarts 7 TEMP 71C AI infrastructure health = 0.61 nudge: memory pressure + pod restart correlation — inspect eviction and container limits.
