GrowthVerdict

@forge/monorepo

---

@forge/monorepo / backend/src / GrowthVerdict

Type Alias: GrowthVerdict

GrowthVerdict = object

Defined in: backend/src/loadtest/metrics.ts:103

Properties

bytesPerHour

readonly bytesPerHour: number

Defined in: backend/src/loadtest/metrics.ts:105

Bytes per hour, from a least-squares fit over the samples.

---

firstQuartileMean

readonly firstQuartileMean: number

Defined in: backend/src/loadtest/metrics.ts:106

---

lastQuartileMean

readonly lastQuartileMean: number

Defined in: backend/src/loadtest/metrics.ts:107

---

leaking

readonly leaking: boolean

Defined in: backend/src/loadtest/metrics.ts:110

---

observedMs

readonly observedMs: number

Defined in: backend/src/loadtest/metrics.ts:109

---

reason

readonly reason: "leaking" | "stable" | "insufficient-samples" | "insufficient-duration"

Defined in: backend/src/loadtest/metrics.ts:118

Why the verdict is what it is — including "the run was too short to tell".

The important arm. A short run producing "no leak" is the single most misleading output this harness could have, because it looks exactly like a passing result. AC-2 asks for a multi-hour soak precisely because a short burst hides slow growth, so anything shorter must say so rather than reporting a clean bill of health.

---

samples

readonly samples: number

Defined in: backend/src/loadtest/metrics.ts:108