Cold access screen render p99 exceeds 1 s at 300 ms RTT

kayg commented

2026-10-07 08:07:04 +00:00

Owner

Finding from the D130 latency run on branch latency-gate at base cecc7ecd7b. Reproduce with TRACER_TEST_PORT=4621 bun x playwright test test/terminal.spec.ts --grep "access screens". The production tracer build used the loopback stream proxy at configured RTT 300 ms and jitter ±30 ms; its self-test measured p50 299.05 ms and p99 301.81 ms over 5 samples.

Setup screen, 30 cold local Chromium documents: render upper bound p50 1071.80 ms / p99 1664.10 ms; first-contentful-paint p50 998.60 ms / p99 1582.30 ms. The existing local test budget is 1000 ms and failed at p99 1664.10 ms. Host: infraternal-dev, Linux x86_64, 8 cores, load average 6.29 / 7.97 / 7.09. This run is a desktop local measurement, not a mid-range phone result.

This is a separate access-screen finding from reconnect issue #61. Product changes are outside #55; track the cold access waterfall and decide the high-RTT budget here.

Finding from the D130 latency run on branch latency-gate at base cecc7ecd7bb452229fcf292bebde0917c0568522. Reproduce with TRACER_TEST_PORT=4621 bun x playwright test test/terminal.spec.ts --grep "access screens". The production tracer build used the loopback stream proxy at configured RTT 300 ms and jitter ±30 ms; its self-test measured p50 299.05 ms and p99 301.81 ms over 5 samples. Setup screen, 30 cold local Chromium documents: render upper bound p50 1071.80 ms / p99 1664.10 ms; first-contentful-paint p50 998.60 ms / p99 1582.30 ms. The existing local test budget is 1000 ms and failed at p99 1664.10 ms. Host: infraternal-dev, Linux x86_64, 8 cores, load average 6.29 / 7.97 / 7.09. This run is a desktop local measurement, not a mid-range phone result. This is a separate access-screen finding from reconnect issue #61. Product changes are outside #55; track the cold access waterfall and decide the high-RTT budget here.

kayg referenced this issue

2026-10-07 08:07:09 +00:00

Tests: high latency by default + input-to-photon gate (D130) #55

kayg commented

2026-10-07 08:17:47 +00:00

Author

Owner

Final burst-jitter relay confirmation: setup screen render p50 1442.90 ms / p99 1955.50 ms and first-contentful-paint p50 1221.40 ms / p99 1733.80 ms (n=30) at configured RTT 300 ms / jitter ±30. Host load during the run was 19.18/11.15/8.71 on 8 cores. The original 60 s test timeout expired before the sign-in group; the harness timeout is now 120 s.

kayg commented

2026-10-07 09:14:50 +00:00

Author

Owner

Updated #55 production tracer measurement on infraternal-dev (x86_64, 8 cores; tracer-browser load 14.31/16.38/15.76), configured RTT 300 ms ±30 ms: setup access render upper-bound n=30 p50 1074.20 ms, p99 1682.30 ms versus the existing 1000 ms gate, FAIL. FCP p50 992.70 ms/p99 1630.30 ms, report-only. Measured by scripts/verify latency with TRACER_TEST_PORT=4621; source apps/tracer/test/terminal.spec.ts; log latency-profile.log in the latency-gate scratch directory.

kayg commented

2026-10-07 09:15:33 +00:00

Author

Owner

The same #55 tracer-browser run also measured the returning sign-in access screen at n=30, p50 render upper-bound 1795.30 ms/p99 1848.10 ms versus budget 1000 ms (FAIL); first-contentful-paint p50 1751.90 ms/p99 1803.70 ms (report-only).

kayg commented

2026-10-07 09:43:30 +00:00

Author

Owner

Latest production-build profile on infraternal-dev (x86_64, 8 cores; tracer-browser phase load 17.73 10.48 10.96): setup render n=30 p50=1051.30 ms, p99=1509.70 ms, 1000 ms budget FAIL; setup FCP p50=984.80 ms, p99=1454.30 ms (report-only). Sign-in render p50=1788.60 ms, p99=1839.10 ms, budget FAIL; sign-in FCP p50=1753.50 ms, p99=1800.40 ms (report-only). Full run and screenshot paths are in #55.

kayg referenced this issue

2026-10-07 10:32:58 +00:00

Tests: high latency by default + input-to-photon gate (D130) #55

kayg commented

2026-10-07 10:32:58 +00:00

Author

Owner

Latest access-screen render measurements on infraternal-dev (x86_64, 8 cores; tracer-browser load 8.84/9.29/7.19), production build, 300 ms RTT ±30 ms:
METRIC name=access_screen_render_upper_bound_setup source=tracer-production-build n=30 configured_rtt_ms=300 p50_ms=1341.60 p99_ms=1861.70 budget_ms=1000 result=FAIL
METRIC name=access_screen_first_contentful_paint_setup source=tracer-production-build n=30 configured_rtt_ms=300 p50_ms=1172.20 p99_ms=1614.20 budget_ms=1000 result=REPORT
METRIC name=access_screen_render_upper_bound_signin source=tracer-production-build n=30 configured_rtt_ms=300 p50_ms=1621.60 p99_ms=2054.40 budget_ms=1000 result=FAIL
METRIC name=access_screen_first_contentful_paint_signin source=tracer-production-build n=30 configured_rtt_ms=300 p50_ms=1454.40 p99_ms=1964.30 budget_ms=1000 result=REPORT
Full profile log was $HOME/.cache/codex-scratch/latency-gate/latency-final.log.

Latest access-screen render measurements on infraternal-dev (x86_64, 8 cores; tracer-browser load 8.84/9.29/7.19), production build, 300 ms RTT ±30 ms: `METRIC name=access_screen_render_upper_bound_setup source=tracer-production-build n=30 configured_rtt_ms=300 p50_ms=1341.60 p99_ms=1861.70 budget_ms=1000 result=FAIL` `METRIC name=access_screen_first_contentful_paint_setup source=tracer-production-build n=30 configured_rtt_ms=300 p50_ms=1172.20 p99_ms=1614.20 budget_ms=1000 result=REPORT` `METRIC name=access_screen_render_upper_bound_signin source=tracer-production-build n=30 configured_rtt_ms=300 p50_ms=1621.60 p99_ms=2054.40 budget_ms=1000 result=FAIL` `METRIC name=access_screen_first_contentful_paint_signin source=tracer-production-build n=30 configured_rtt_ms=300 p50_ms=1454.40 p99_ms=1964.30 budget_ms=1000 result=REPORT` Full profile log was `$HOME/.cache/codex-scratch/latency-gate/latency-final.log`.

kayg referenced this issue

2026-10-07 10:55:40 +00:00

Tests: high latency by default + input-to-photon gate (D130) #55

kayg referenced this issue from a commit

2026-10-10 09:20:35 +00:00

Merge latency-gate: 300 ms RTT test profile and input-to-photon gate (#55, D130)

kayg referenced this issue

2026-10-28 17:52:29 +00:00

Tests: high latency by default + input-to-photon gate (D130) #55

~~kayg referenced this issue 2026-10-28 17:53:55 +00:00~~

Orchestrator handover 2026-10-28: moving to a new machine #254

~~kayg referenced this issue 2026-10-28 17:53:56 +00:00~~