Public: anyone can read this; the crew can change it.
GET /api/v1/telemetry/flow never blocks the forge: it held the main thread 19% of the time #8978
Landed in pull #8981
2 earlier comments
Reviewed pull #8981 at its head: ticked The answer is computed at most once every few seconds, in the background or off the thread, however many readers ask; cached copies answer in under 5 ms with a test on a production-sized fixture.; A named reader polling a route too often is visible on the metrics page (#7559).; If foundry is the caller, it reads the shared stream from #8955 instead, by an issue on trevato/foundry.. left open: Before and after numbers from the pod log on the pull. — The numbers on the pull come from the bigForge fixture, not from the pod log..
Left open after pull #8981 landed: 1 unticked item remains (Before and after numbers from the pod log on the pull.). The planner files each as its own issue and closes this one; a person may close it, tick what is done, or say what is left. Until then the crew does not pick it again.
Split: no builder can read the pod log, so the before/after box becomes #9016. It shows stall time by route on the metrics page, so the owner can read telemetry/flow's figure from the forge itself.
Measured 2026-11-02 23:39Z to 23:45Z on 0.7.349: 96
event loop blockedlines running/api/v1/telemetry/flow, median 662 ms, max 1,577 ms, 69.6 s of 6 minutes. It's the top stall source in the last 11 minutes, ahead of everything else. It started this evening, so name the caller: trevato/foundry's new read of the forge (#8955) is my suspect, not proved.Done when: