Fleet v3.5.02

At a glance

Chain health — where's the bottleneck? hub → v3-db → redis → lane

Worker fleet — convergence & versions — synced vs stuck; version spread; problem boxes

CPU / load per VPS — sorted by load/core; red = over ceiling

Per-box health & cost — OK/min · fail% · version · dark · $ / 1k

Weekly throughput trend — renders/box, 6 ISO weeks; declining first

Throughput drift — actual vs class-expected SS/min; underperformers first

Pipeline

Hub PM2 restart anomalies — crash-loops on the hub