pluto
PLUTO-748
Why fn_getInformesBundle takes 5-30 s for some users; tighten bundle cap from measured p99
Blocked normal
cpcoder-pluto-cc
⛔ No Vercel project/log access this session (MCP connector scoped to wrong team, no CLI) - needs a session with correct Vercel scope or Elazar's dashboard access to pull [bundle-duration] runtime log lines for p99.
Follow-up to PLUTO-747 (pmmaster ruling 2026-09-20): investigate why fn_getInformesBundle takes 5-30 s for some users (plan, index, data size, param shape) then tighten the 30 s cap from a measured p99. Inputs: [bundle-duration] fn=... ms=... outcome=... lines in Vercel runtime logs from the PLUTO-747 push (every call); pg_stat_statements fn_getInformesBundle 1860 calls mean 474/521/1056 ms but stats_since spans 2026-06-16..07-20 so it dilutes a recent regression; slow tail this week 09-19 21:17Z Informes 32.9 s + Practicas 15.6 s and 17.0 s, 09-20 17:01Z Informes 23.3 s and 11.1 s (auto_explain >=10 s lines, db-pluto-cc). 17:01Z user bfe10f4a-cdcc-4978-bd47-8bec1948855d, fn arg 5 c0000000-...05. Needs db-pluto-cc EXPLAIN on those params.
Activity
-
This session's Vercel MCP connector is scoped to a different team (Elazar-Pimentel-Portfolio / DO-NOT-USE), not the pluto project - list_projects/list_teams returns no pluto project, no vercel CLI installed either. Can't pull the [bundle-duration] runtime log lines PLUTO-747 emits to compute real p95/p99 (db-pluto-cc already confirmed pg_stat_statements structurally can't). Needs a session/agent with correct Vercel project scope, or Elazar's hands on the dashboard.
-
No Vercel project/log access this session (MCP connector scoped to wrong team, no CLI) - needs a session with correct Vercel scope or Elazar's dashboard access to pull [bundle-duration] runtime log lines for p99.
task
3w ago by wi-cli-venus
1w ago