ci: run the CPU-independent scale PR tier in FE-GATE-005 (11-03)
No CI job ran any scale tier, and every tier past pr compares timings only on the baseline CPU. The pr tier checks counts, bounds and growth shape; it now includes S2 composition, since S0/S1 stay under the noise floor and let a quadratic accumulation through. The ungated-test check reads tier paths. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
co-authored by
Claude Opus 5.5
parent
7f889a6d83
commit
45ee231bb0
@@ -1,11 +1,11 @@
|
||||
{
|
||||
"schemaVersion": 1,
|
||||
"owner": "frontend-platform",
|
||||
"reason": "11-01, 13-03 (D-14): FE-GATE-016 retired; it ran the same FE-RB-005 drill as FE-GATE-025",
|
||||
"reason": "11-03: FE-GATE-005 runs the CPU-independent scale PR tier (check:scale:pr)",
|
||||
"gates": 27,
|
||||
"commands": 94,
|
||||
"commandReferences": 104,
|
||||
"commands": 95,
|
||||
"commandReferences": 105,
|
||||
"artifacts": 109,
|
||||
"stages": 5,
|
||||
"gateShapeSha256": "3ad2407c5929a7ec43dadebc68a3dc69e265793bf1c80c355b6d2fdeae019821"
|
||||
"gateShapeSha256": "58e5be2bdf92c9ee355fa93ced794b6cc948a960905e80919d95087e56262d94"
|
||||
}
|
||||
|
||||
@@ -215,6 +215,11 @@
|
||||
"expectedExitCode": 1,
|
||||
"expectedDiagnosticId": "Risk coverage failed:"
|
||||
},
|
||||
{
|
||||
"id": "check-scale-pr",
|
||||
"script": "check:scale:pr",
|
||||
"expect": "pass"
|
||||
},
|
||||
{
|
||||
"id": "test-component",
|
||||
"script": "test:component",
|
||||
@@ -1791,7 +1796,8 @@
|
||||
"test-unit",
|
||||
"test-contract",
|
||||
"test-coverage",
|
||||
"check-coverage-fixture"
|
||||
"check-coverage-fixture",
|
||||
"check-scale-pr"
|
||||
],
|
||||
"logArtifactId": "artifact-artifacts-quality-gates-FE-GATE-005-txt",
|
||||
"evidenceArtifactIds": [
|
||||
|
||||
@@ -1,57 +1,29 @@
|
||||
{
|
||||
"schemaVersion": 1,
|
||||
"entries": [
|
||||
{
|
||||
"path": "tests/scale/benchmark.test.ts",
|
||||
"reason": "Scale suite. It runs through check:scale:* (scripts/scale/run-scale-profile.ts) and verify:extreme-scale, which no CI gate calls, because its timing baseline passes only on the CPU that recorded it. Review 11-03 moves it into a PR tier; drop the entry when a gate runs it."
|
||||
},
|
||||
{
|
||||
"path": "tests/scale/composition-scale.test.ts",
|
||||
"reason": "Scale suite. It runs through check:scale:* (scripts/scale/run-scale-profile.ts) and verify:extreme-scale, which no CI gate calls, because its timing baseline passes only on the CPU that recorded it. Review 11-03 moves it into a PR tier; drop the entry when a gate runs it."
|
||||
"reason": "Scale suite in the main or nightly tier (scripts/scale/run-scale-profile.ts). No CI job runs those tiers: both include baseline-regression, which compares timings only on the CPU model that recorded config/scale/benchmark-baseline.json and reports incomplete elsewhere (review 11-03). Run check:scale:main or check:scale:nightly on that machine. The CPU-independent pr tier runs in FE-GATE-005."
|
||||
},
|
||||
{
|
||||
"path": "tests/scale/global-local-independence.test.ts",
|
||||
"reason": "Scale suite. It runs through check:scale:* (scripts/scale/run-scale-profile.ts) and verify:extreme-scale, which no CI gate calls, because its timing baseline passes only on the CPU that recorded it. Review 11-03 moves it into a PR tier; drop the entry when a gate runs it."
|
||||
},
|
||||
{
|
||||
"path": "tests/scale/growth-regression.test.ts",
|
||||
"reason": "Scale suite. It runs through check:scale:* (scripts/scale/run-scale-profile.ts) and verify:extreme-scale, which no CI gate calls, because its timing baseline passes only on the CPU that recorded it. Review 11-03 moves it into a PR tier; drop the entry when a gate runs it."
|
||||
},
|
||||
{
|
||||
"path": "tests/scale/scale-baseline.test.ts",
|
||||
"reason": "Scale suite. It runs through check:scale:* (scripts/scale/run-scale-profile.ts) and verify:extreme-scale, which no CI gate calls, because its timing baseline passes only on the CPU that recorded it. Review 11-03 moves it into a PR tier; drop the entry when a gate runs it."
|
||||
"reason": "Scale suite in the main or nightly tier (scripts/scale/run-scale-profile.ts). No CI job runs those tiers: both include baseline-regression, which compares timings only on the CPU model that recorded config/scale/benchmark-baseline.json and reports incomplete elsewhere (review 11-03). Run check:scale:main or check:scale:nightly on that machine. The CPU-independent pr tier runs in FE-GATE-005."
|
||||
},
|
||||
{
|
||||
"path": "tests/scale/session/long-session.test.ts",
|
||||
"reason": "Scale suite. It runs through check:scale:* (scripts/scale/run-scale-profile.ts) and verify:extreme-scale, which no CI gate calls, because its timing baseline passes only on the CPU that recorded it. Review 11-03 moves it into a PR tier; drop the entry when a gate runs it."
|
||||
"reason": "Scale suite in the main or nightly tier (scripts/scale/run-scale-profile.ts). No CI job runs those tiers: both include baseline-regression, which compares timings only on the CPU model that recorded config/scale/benchmark-baseline.json and reports incomplete elsewhere (review 11-03). Run check:scale:main or check:scale:nightly on that machine. The CPU-independent pr tier runs in FE-GATE-005."
|
||||
},
|
||||
{
|
||||
"path": "tests/scale/session/network-flap.test.ts",
|
||||
"reason": "Scale suite. It runs through check:scale:* (scripts/scale/run-scale-profile.ts) and verify:extreme-scale, which no CI gate calls, because its timing baseline passes only on the CPU that recorded it. Review 11-03 moves it into a PR tier; drop the entry when a gate runs it."
|
||||
},
|
||||
{
|
||||
"path": "tests/scale/synthetic-topology.test.ts",
|
||||
"reason": "Scale suite. It runs through check:scale:* (scripts/scale/run-scale-profile.ts) and verify:extreme-scale, which no CI gate calls, because its timing baseline passes only on the CPU that recorded it. Review 11-03 moves it into a PR tier; drop the entry when a gate runs it."
|
||||
"reason": "Scale suite in the main or nightly tier (scripts/scale/run-scale-profile.ts). No CI job runs those tiers: both include baseline-regression, which compares timings only on the CPU model that recorded config/scale/benchmark-baseline.json and reports incomplete elsewhere (review 11-03). Run check:scale:main or check:scale:nightly on that machine. The CPU-independent pr tier runs in FE-GATE-005."
|
||||
},
|
||||
{
|
||||
"path": "tests/scale/telemetry/telemetry-storm.test.ts",
|
||||
"reason": "Scale suite. It runs through check:scale:* (scripts/scale/run-scale-profile.ts) and verify:extreme-scale, which no CI gate calls, because its timing baseline passes only on the CPU that recorded it. Review 11-03 moves it into a PR tier; drop the entry when a gate runs it."
|
||||
},
|
||||
{
|
||||
"path": "tests/scale/traffic/amplification-metrics.test.ts",
|
||||
"reason": "Scale suite. It runs through check:scale:* (scripts/scale/run-scale-profile.ts) and verify:extreme-scale, which no CI gate calls, because its timing baseline passes only on the CPU that recorded it. Review 11-03 moves it into a PR tier; drop the entry when a gate runs it."
|
||||
},
|
||||
{
|
||||
"path": "tests/scale/traffic/failure-scenarios.test.ts",
|
||||
"reason": "Scale suite. It runs through check:scale:* (scripts/scale/run-scale-profile.ts) and verify:extreme-scale, which no CI gate calls, because its timing baseline passes only on the CPU that recorded it. Review 11-03 moves it into a PR tier; drop the entry when a gate runs it."
|
||||
},
|
||||
{
|
||||
"path": "tests/scale/traffic/retry-after-jitter.test.ts",
|
||||
"reason": "Scale suite. It runs through check:scale:* (scripts/scale/run-scale-profile.ts) and verify:extreme-scale, which no CI gate calls, because its timing baseline passes only on the CPU that recorded it. Review 11-03 moves it into a PR tier; drop the entry when a gate runs it."
|
||||
"reason": "Scale suite in the main or nightly tier (scripts/scale/run-scale-profile.ts). No CI job runs those tiers: both include baseline-regression, which compares timings only on the CPU model that recorded config/scale/benchmark-baseline.json and reports incomplete elsewhere (review 11-03). Run check:scale:main or check:scale:nightly on that machine. The CPU-independent pr tier runs in FE-GATE-005."
|
||||
},
|
||||
{
|
||||
"path": "tests/scale/tsc-diagnostics-parser.test.ts",
|
||||
"reason": "Scale suite. It runs through check:scale:* (scripts/scale/run-scale-profile.ts) and verify:extreme-scale, which no CI gate calls, because its timing baseline passes only on the CPU that recorded it. Review 11-03 moves it into a PR tier; drop the entry when a gate runs it."
|
||||
"reason": "Scale suite in the main or nightly tier (scripts/scale/run-scale-profile.ts). No CI job runs those tiers: both include baseline-regression, which compares timings only on the CPU model that recorded config/scale/benchmark-baseline.json and reports incomplete elsewhere (review 11-03). Run check:scale:main or check:scale:nightly on that machine. The CPU-independent pr tier runs in FE-GATE-005."
|
||||
}
|
||||
]
|
||||
}
|
||||
|
||||
@@ -35,9 +35,20 @@ canonical `check:ci`를 돌린다(`scripts/lib/removal-fixture.ts`의
|
||||
`check:ci`는 `tests/` 아래 테스트 파일(`tests/fixtures` 제외)이 모두 CI 게이트
|
||||
명령에서 실행되는지도 확인한다. 게이트 명령이 닿는 package script의 Vitest 경로와
|
||||
Playwright 설정의 `testDir`를 모아 비교한다(`scripts/lib/ungated-tests.ts`).
|
||||
`run-scale-profile.ts <tier>`를 부르는 명령은 그 tier의 Vitest 경로로 센다.
|
||||
게이트가 돌리지 않는 테스트는 `config/ci/ungated-tests.json`에 이유와 함께 적어야
|
||||
한다. 지금은 `tests/scale`의 규모 테스트 13개만 있다(리뷰 11-03에서 PR tier로
|
||||
옮길 때 뺀다). 게이트가 이미 돌리는 파일을 이 목록에 남겨 두면 실패한다.
|
||||
한다. 게이트가 이미 돌리는 파일을 이 목록에 남겨 두면 실패한다.
|
||||
|
||||
규모 테스트는 이렇게 나뉜다(리뷰 11-03, `scripts/scale/run-scale-profile.ts`).
|
||||
|
||||
- `check:scale:pr`은 merge gate `FE-GATE-005`에서 돈다. 개수·상한·증가 모양만
|
||||
검사하므로 CPU가 달라도 결과가 같다. S0~S2 조합 증가 검사가 여기 있다. S0/S1만
|
||||
재면 모든 workload가 noise floor 아래라 이차 누적을 놓친다. 조합 timing은
|
||||
산출물(`artifacts/scale/pr/composition-s0-s2.json`)에 기록만 한다.
|
||||
- `main`·`nightly`·`release` tier는 CI에서 돌지 않는다. 이 tier들은
|
||||
`baseline-regression`을 포함한다. 이 단계는 `config/scale/benchmark-baseline.json`을
|
||||
기록한 CPU 모델에서만 timing을 비교하고, 다른 CPU에서는 `incomplete`로 끝난다.
|
||||
이 tier에만 있는 테스트 6개는 `ungated-tests.json`에 이유와 함께 남아 있다.
|
||||
|
||||
템플릿을 복사해 새 프로젝트를 시작하면 아래 기준선은 템플릿의 값이다. 샘플을
|
||||
지우거나 구조를 바꾸면 프로젝트 값으로 다시 찍는다.
|
||||
|
||||
@@ -3,12 +3,14 @@ import path from "node:path";
|
||||
|
||||
import { z } from "zod";
|
||||
|
||||
import { SCALE_TIERS, type ScaleTier } from "../scale/run-scale-profile.ts";
|
||||
import { reachablePackageScripts } from "./package-script-graph.ts";
|
||||
|
||||
/**
|
||||
* 12-04. A test file protects nothing unless a CI gate runs it. The gated set
|
||||
* is read from the package scripts the CI contract's commands reach: Vitest
|
||||
* path filters (`run-vitest.ts run <paths>`) and Playwright config `testDir`s.
|
||||
* path filters (`run-vitest.ts run <paths>`), the Vitest steps of a scale tier
|
||||
* (`run-scale-profile.ts <tier>`) and Playwright config `testDir`s.
|
||||
* A test outside that set needs a reasoned entry in config/ci/ungated-tests.json.
|
||||
*/
|
||||
export const UNGATED_TESTS_PATH = "config/ci/ungated-tests.json";
|
||||
@@ -53,17 +55,22 @@ export function gatedTestPaths(
|
||||
readConfig: (file: string) => string,
|
||||
): string[] {
|
||||
const paths = new Set<string>();
|
||||
const addVitestPaths = (tokens: readonly string[]) => {
|
||||
const vitestRun = tokens.findIndex(
|
||||
(token, index) => token === "run" && /(?:run-vitest\.ts|vitest)$/u.test(tokens[index - 1] ?? ""),
|
||||
);
|
||||
if (vitestRun < 0) return;
|
||||
for (const token of tokens.slice(vitestRun + 1)) {
|
||||
if (!token.startsWith("-")) paths.add(token.replace(/\/$/u, ""));
|
||||
}
|
||||
};
|
||||
for (const script of reachablePackageScripts(scripts, entryScripts)) {
|
||||
for (const segment of scripts[script]!.split("&&")) {
|
||||
const tokens = segment.trim().split(/\s+/u);
|
||||
const vitestRun = tokens.findIndex(
|
||||
(token, index) => token === "run" && /(?:run-vitest\.ts|vitest)$/u.test(tokens[index - 1] ?? ""),
|
||||
);
|
||||
if (vitestRun >= 0) {
|
||||
for (const token of tokens.slice(vitestRun + 1)) {
|
||||
if (token.startsWith("-")) break;
|
||||
paths.add(token.replace(/\/$/u, ""));
|
||||
}
|
||||
addVitestPaths(tokens);
|
||||
const scaleProfile = tokens.findIndex((token) => /run-scale-profile\.ts$/u.test(token));
|
||||
if (scaleProfile >= 0) {
|
||||
for (const step of SCALE_TIERS[tokens[scaleProfile + 1] as ScaleTier] ?? []) addVitestPaths(step.command);
|
||||
}
|
||||
const playwrightTest = tokens.findIndex(
|
||||
(token, index) => token === "test" && tokens[index - 1] === "playwright",
|
||||
|
||||
@@ -3,14 +3,21 @@ import { mkdir, writeFile } from "node:fs/promises";
|
||||
import path from "node:path";
|
||||
|
||||
/**
|
||||
* Spec §47. CI execution tiers for the scale/chaos/soak program.
|
||||
* Spec §47. Execution tiers for the scale/chaos/soak program.
|
||||
*
|
||||
* pr S0/S1 architecture, composition and traffic invariants
|
||||
* main pr + S2 composition, TypeScript diagnostics and the stored
|
||||
* baseline regression gate (config/scale/benchmark-baseline.json)
|
||||
* pr architecture and traffic invariants, S0-S2 composition growth
|
||||
* main pr + TypeScript diagnostics and the stored baseline regression
|
||||
* gate (config/scale/benchmark-baseline.json)
|
||||
* nightly main + S3/S4 topology, telemetry storm, session cycles, browser soak
|
||||
* release release-coherence matrix + the stable main/nightly subset
|
||||
*
|
||||
* Only `pr` runs in CI (merge gate FE-GATE-005, 11-03). Its checks are counts,
|
||||
* bounds and growth shape, so they hold on any CPU; the composition timings it
|
||||
* writes are advisory. `main`, `nightly` and `release` include
|
||||
* `baseline-regression`, which compares timings only on the CPU model that
|
||||
* recorded config/scale/benchmark-baseline.json and is `incomplete` elsewhere,
|
||||
* so no CI job runs them; they are local commands for that machine.
|
||||
*
|
||||
* Every step's output goes to artifacts/scale/<tier>/<step>.log and the tier
|
||||
* writes one machine-readable summary.json. A step that could not run (e.g. a
|
||||
* browser soak without a build or browser) is `skipped`, and a tier with any
|
||||
@@ -49,18 +56,16 @@ const PR_STEPS: readonly Step[] = [
|
||||
"tests/scale/traffic",
|
||||
],
|
||||
},
|
||||
// 11-03: S2 is included because below it every workload stays under the
|
||||
// noise floor, so a quadratic accumulation passed S0/S1 unnoticed.
|
||||
{
|
||||
id: "composition-s0-s1",
|
||||
command: ["scripts/scale/run-composition-scale.ts", "--profiles=S0,S1"],
|
||||
id: "composition-s0-s2",
|
||||
command: ["scripts/scale/run-composition-scale.ts", "--profiles=S0,S1,S2"],
|
||||
},
|
||||
];
|
||||
|
||||
const MAIN_STEPS: readonly Step[] = [
|
||||
...PR_STEPS,
|
||||
{
|
||||
id: "composition-s0-s2",
|
||||
command: ["scripts/scale/run-composition-scale.ts", "--profiles=S0,S1,S2"],
|
||||
},
|
||||
{
|
||||
id: "composition-scale-suite",
|
||||
command: [...VITEST, "tests/scale/composition-scale.test.ts", "tests/scale/tsc-diagnostics-parser.test.ts"],
|
||||
|
||||
@@ -194,8 +194,8 @@ describe("CI gate contract", () => {
|
||||
).filter((gateId) => gateId !== "FE-GATE-016"),
|
||||
);
|
||||
expect(contract.jobs).toHaveLength(9);
|
||||
expect(contract.commands).toHaveLength(94);
|
||||
expect(contract.gates.reduce((total, gate) => total + gate.commandIds.length, 0)).toBe(104);
|
||||
expect(contract.commands).toHaveLength(95);
|
||||
expect(contract.gates.reduce((total, gate) => total + gate.commandIds.length, 0)).toBe(105);
|
||||
expect(contract.commands.filter(({ expect }) => expect === "fail")).toHaveLength(31);
|
||||
expect(contract.gates.reduce((total, gate) => total + gate.evidenceArtifactIds.length, 0)).toBe(87);
|
||||
expect(contract.artifacts).toHaveLength(109);
|
||||
@@ -221,6 +221,14 @@ describe("CI gate contract", () => {
|
||||
expect(stageOf("FE-GATE-028")).toBe("release");
|
||||
});
|
||||
|
||||
it("runs the CPU-independent scale PR tier in a merge gate (11-03)", async () => {
|
||||
const contract = await loadCiGateContract(process.cwd());
|
||||
const index = indexCiGateContract(contract);
|
||||
const scripts = index.gates.get("FE-GATE-005")?.commandIds.map((id) => index.commands.get(id)?.script);
|
||||
expect(scripts).toContain("check:scale:pr");
|
||||
expect(contract.stages.find(({ id }) => id === "merge")?.gateIds).toContain("FE-GATE-005");
|
||||
});
|
||||
|
||||
it("keeps the action registry recursively immutable and resolves only known actions", () => {
|
||||
expect(Object.isFrozen(CI_ACTION_REGISTRY)).toBe(true);
|
||||
expect(Object.values(CI_ACTION_REGISTRY).every((action) => Object.isFrozen(action))).toBe(true);
|
||||
|
||||
@@ -46,12 +46,24 @@ describe("ungated test files", () => {
|
||||
"tests/scripts/verify-extreme-scale.test.ts",
|
||||
"tests/scale/browser-soak.contract.test.ts",
|
||||
"tests/e2e/app-shell.spec.ts",
|
||||
// 11-03: the CPU-independent scale PR tier runs in FE-GATE-005.
|
||||
"tests/scale/growth-regression.test.ts",
|
||||
"tests/scale/traffic/amplification-metrics.test.ts",
|
||||
])("gates %s", async (file) => {
|
||||
const inputs = await repositoryInputs();
|
||||
expect(ungatedTestFailures({ ...inputs, testFiles: [file] })).toEqual([]);
|
||||
expect(inputs.exclusions.some(({ path }) => path === file)).toBe(false);
|
||||
});
|
||||
|
||||
it("reads the Vitest paths of a scale tier the gate command runs (11-03)", () => {
|
||||
expect(
|
||||
gatedTestPaths({ gate: "node scripts/scale/run-scale-profile.ts pr" }, ["gate"], () => ""),
|
||||
).toEqual(expect.arrayContaining(["tests/scale/growth-regression.test.ts", "tests/scale/traffic"]));
|
||||
expect(
|
||||
gatedTestPaths({ gate: "node scripts/scale/run-scale-profile.ts pr" }, ["gate"], () => ""),
|
||||
).not.toContain("tests/scale/session");
|
||||
});
|
||||
|
||||
it("rejects a new test directory that no gate runs", async () => {
|
||||
const inputs = await repositoryInputs();
|
||||
expect(
|
||||
|
||||
Reference in New Issue
Block a user