The ring's 31-day test holds its 5 s budget scaled to this machine, measured against plain inserts of the same points #48

"31 days of 15-s points keep at most 30 days of 5-min roll-ups" timed 178,560 points through record() against a fixed 5,000 ms. On the check runner (half a core, 512 MiB) it took 6,012 ms and failed gate (1) of PR #44's check (run 903, boards-sampler.test.ts:192); main's own push gate has failed on it too. Where the budget was set it takes ~500 ms.

The test now first times plain INSERTs of the same points in the same 5-min batches (~250 ms there) and allows 5,000 ms times max(1, plain / 250 ms). The budget is unchanged on a machine at least that fast, and scales on a slower one. It still catches what it is there for: with the roll-up rescanning every window (from = ""), the run is 40,702 ms against a budget of 5,481 ms and fails.

<!-- pipeline:validation -->

Validation

pipeline/validate on b424e9eb: pass in 8s — bun 1.3.14, CI-shaped env (scrubbed vars, runner PATH), exact check.yml shard + forge.json commands.

• ⚠️ node v20.19.2 here, v22 on CI (tsc runs under node)

▶ bun install --frozen-lockfile
  ok  exit 0  0s
▶ forge.json check.fast
Ran 99 tests across 8 files in 2.0s (8 shards).
  ok  exit 0  4s
▶ check.yml gate (1) pull_request
Ran 17 tests across 2 files in 1.2s (2 shards).
  ok  exit 0  1s
▶ check.yml gate (2) pull_request
Ran 14 tests across 3 files in 0.1s (3 shards).
  ok  exit 0  1s
▶ check.yml gate (3) pull_request
Ran 68 tests across 3 files in 2.3s (3 shards).
  ok  exit 0  2s