Metrics: time to launch, queue time, run time, runner use issue #35 open andymccall opened this · 1 comment
andymccall set labels flow-next, feature ·
referenced by 2 changes on 21 Oct 2026
andymccall
2026-10-21 (loop 18): built and released.
- The hub's /metrics (OxygenHub 0.7.0): oxyhub_flow_job_queue_seconds (ready → started) and oxyhub_flow_job_run_seconds histograms by runner, finished jobs by runner and result, reused passes (#265), over the last day.
- A runner's own /metrics (OxygenFlow 0.3.0, off unless
metrics = ADDRESSunder [runner], optional metrics_token): claims, declines, finished jobs by result, job seconds, jobs running, version, uptime. The runners are not rebuilt yet, so nothing serves it until they are and the option is set.
Left: the launch time split (image fetch, rootfs, VM boot, command start) reported by the runner per job, and the charts on Admin › Runners and a repository's Flow tab (Insights already shows queue and run times).
andymccall referenced by change 1215f05ff8d0 on main ·
Andy (2026-10-02): measure Flow. Per job: queued → claimed (queue time), claimed → VM booted → command started (launch time, split into fetch, rootfs, boot), run time, upload time; per runner: utilisation, jobs, errors; per repo: run time trends. Kept in the hub (time series in SQLite at first), shown on Admin → Runners and a repo's Flow tab, exported as Prometheus /metrics for real monitoring.