ptth/issues/2020-12Dec/metrics-K5NPHQHP.md

37 lines
775 B
Markdown

# Metrics + instance data + log
(K5NPHQHP)
## ptth_server metrics
These are just like Prometheus. Either counters, or gauges, scalar or histograms.
- CPU usage counter
- RAM usage gauge
- Requests received counter
- Responses sent counter
- Tokio gauges?
- Outbound byte counter
- Uptime gauge
- Last seen UTC time
- Free disk space gauge
## ptth_server instance data
These are captured during process start and then don't change. They're kinda
like metrics, but static.
- Machine ID
- Git version (if possible)
- Server ID
- Random instance ID
- UTC time at startup
## ptth_server logs
Basically what the tracing crate puts out. See if we can route that over the
network?
- Metrics, captured every 5 minutes or whatever
- Probably "warn" level logs from `tracing`