The team now has three observability efforts. Lithos Probe (ModDB, on-demand profiling: /probe health, /probe profiler, file export). Stratum's internal instrumentation (/stratum timings and a metrics publisher that pushes a JSON snapshot once a second over a local named pipe for a GUI). And Pulse, continuous scrape-based time series.
There is no technical conflict (different transports, different consumers), but the lanes deserve an explicit agreement so effort does not double up: Probe and timings for point-in-time diagnosis, Pulse for dashboards and alerting. Worth discussing shared Grafana dashboards.
For a Pulse V2 on Stratum servers: Stratum compiles a public StratumEntityBehaviorTimings accumulator into its vsapi (thread-safe, drainable). Reflection-based feature detection with a clean fallback on vanilla and Lithos would give per-behavior tick histograms no public API can provide.
The team now has three observability efforts. Lithos Probe (ModDB, on-demand profiling: /probe health, /probe profiler, file export). Stratum's internal instrumentation (/stratum timings and a metrics publisher that pushes a JSON snapshot once a second over a local named pipe for a GUI). And Pulse, continuous scrape-based time series.
There is no technical conflict (different transports, different consumers), but the lanes deserve an explicit agreement so effort does not double up: Probe and timings for point-in-time diagnosis, Pulse for dashboards and alerting. Worth discussing shared Grafana dashboards.
For a Pulse V2 on Stratum servers: Stratum compiles a public StratumEntityBehaviorTimings accumulator into its vsapi (thread-safe, drainable). Reflection-based feature detection with a clean fallback on vanilla and Lithos would give per-behavior tick histograms no public API can provide.