Metrics
Tragwerk surfaces metrics at two scopes: server metrics for the server as a whole, and environment/app metrics for a specific deployed application. Both offer a live view and historical graphs.
Server metrics
Server metrics cover the health of the server itself:
- CPU usage
- Memory usage
- Disk usage
Tragwerk samples each server over SSH in the background and stores the readings, so live values and historical graphs populate automatically — no setup required on your part.
Environment / app metrics
Application metrics describe the running app within an environment — aggregated FrankenPHP and Caddy figures summed across the app's containers:
- Worker pool — total / busy / ready workers and queue depth
- Threads — total and busy
- Requests — total served, plus 4xx and 5xx counts
- Latency — average request duration (derived from sample deltas)
- In-flight — requests currently being processed
Gauges (workers, threads, in-flight) are point-in-time values; counters (requests, durations) are cumulative since container start, so rates and average latency are derived from the difference between consecutive samples.
The worker pool figures are most meaningful for apps running in worker mode. Apps in the default classic mode (the PHP-FPM drop-in) still report request, latency, and thread metrics.
Live updates and graphs
- Live tiles update in real time via Mercure (server-sent events). The page subscribes to a live endpoint and re-renders tiles as new samples arrive.
- Historical graphs are drawn with uPlot from data fetched on a data endpoint, letting you inspect trends over time.
Example
Open Metrics, pick the active project and environment, and watch the live worker and request tiles update while the graphs fill in history.