Monitoring CPU consumption fix
v0.13.2-beta
- Monitoring CPU consumption fix. The health report (containers' CPU/memory usage) ran every 5 seconds. Yet each report queries every container (
docker stats), a costly operation which, on a small server, ran almost continuously and saturated the processor (to the point of triggering the host's CPU throttling, then sustaining itself since the operation slows down under load). The cadence moves to 30 seconds (health and topology), which remains ample for a real-time panel while letting the server breathe. No visible functional change, other than a markedly lower CPU load.