Dashboard metrics
Dashboard collects HTTP traffic, package resources, network traffic, and storage usage without application instrumentation. Select the last hour, 6 hours, 24 hours, 3 days, or 7 days. The default is 24 hours.
HTTP panels show request volume, response status, latency, and top routes. They refresh every minute. Package CPU and memory history and network traffic refresh every ten seconds. Storage usage is loaded with the page; reload to update its breakdown.
CPU is a percentage of allocated VM capacity; 100% means all allocated vCPUs are busy. CPU rates need at least two samples. Memory measures each package’s main process, not its descendants or PostgreSQL. A running process is not proof of HTTP health.
History includes earlier deployments in the selected range. Current resource readings describe the active deployment and omit stale samples; missing data appears as a gap rather than zero usage. Before deployment, panels remain available with empty states. A read error means metrics could not be loaded, not that traffic or resource usage was zero.