| 1 | /** |
| 2 | * Throughput benchmarks against a **deployed** git-service stack |
| 3 | * (DESIGN.md §14–15): turns the estimated scaling table into measured |
| 4 | * numbers, one benchmark per named bottleneck. |
| 5 | * |
| 6 | * These are measurements, not assertions about performance: each case |
| 7 | * prints a line and asserts only *correctness* (every operation succeeded), |
| 8 | * so the suite can never fail because a datacenter had a slow minute. Read |
| 9 | * the printed table to see where the service actually stands. |
| 10 | * |
| 11 | * Run: |
| 12 | * bun run test test/GitBench.e2e.test.ts --profile testing |
| 13 | * BENCH_SCALE=4 ... to make every workload 4x bigger |
| 14 | * |
| 15 | * Skipped under `--fast`. |
| 16 | */ |
| 17 | import * as Cloudflare from "@/Cloudflare"; |
| 18 | import * as Test from "@/Test/Alchemy"; |
| 19 | import { expect } from "alchemy-test"; |
| 20 | import * as Data from "effect/Data"; |
| 21 | import * as Effect from "effect/Effect"; |
| 22 | import * as Fiber from "effect/Fiber"; |
| 23 | import * as FileSystem from "effect/FileSystem"; |
| 24 | import * as Path from "effect/Path"; |
| 25 | import { MinimumLogLevel } from "effect/References"; |
| 26 | import * as Schedule from "effect/Schedule"; |
| 27 | import * as Stream from "effect/Stream"; |
| 28 | import * as HttpClient from "effect/http/HttpClient"; |
| 29 | import * as HttpClientRequest from "effect/http/HttpClientRequest"; |
| 30 | import * as HttpApiClient from "effect/http-api/HttpApiClient"; |
| 31 | import * as ChildProcess from "effect/process/ChildProcess"; |
| 32 | import { GitApi } from "@/Git/Api.ts"; |
| 33 | import { verifyPackResponse } from "./harness/pack.ts"; |
| 34 | import { makeTestStack, TEST_SECRET } from "./fixtures/stack.ts"; |
| 35 | |
| 36 | const { test, beforeAll, afterAll, deploy, destroy } = Test.make({ |
| 37 | providers: Cloudflare.providers(), |
| 38 | }); |
| 39 | |
| 40 | const logLevel = Effect.provideService( |
| 41 | MinimumLogLevel, |
| 42 | process.env.DEBUG ? "Debug" : "Info", |
| 43 | ); |
| 44 | |
| 45 | const Stack = makeTestStack("GitBenchStack"); |
| 46 | |
| 47 | /** Scales every workload; 1 keeps a full run inside the alarm/test budget. */ |
| 48 | const SCALE = Number.parseInt(process.env.BENCH_SCALE ?? "1", 10) || 1; |
| 49 | |
| 50 | const skipBench = !!process.env.FAST; |
| 51 | |
| 52 | // ── plumbing ──────────────────────────────────────────────────────────────── |
| 53 | |
| 54 | const makeClient = (url: string, token: string) => |
| 55 | HttpApiClient.make(GitApi, { |
| 56 | baseUrl: url, |
| 57 | transformClient: HttpClient.mapRequest((request) => |
| 58 | request.pipe(HttpClientRequest.bearerToken(token)), |
| 59 | ), |
| 60 | }); |
| 61 | |
| 62 | class ShellError extends Data.TaggedError("ShellError")<{ |
| 63 | readonly script: string; |
| 64 | readonly exitCode: number; |
| 65 | readonly stdout: string; |
| 66 | readonly stderr: string; |
| 67 | }> { |
| 68 | override get message(): string { |
| 69 | return `bench shell exited ${this.exitCode}\n${this.script}\n${this.stderr}`; |
| 70 | } |
| 71 | } |
| 72 | |
| 73 | const sh = Effect.fn(function* (cwd: string, script: string) { |
| 74 | const handle = yield* ChildProcess.make("sh", ["-ec", script], { |
| 75 | cwd, |
| 76 | env: { |
| 77 | GIT_TERMINAL_PROMPT: "0", |
| 78 | GIT_CONFIG_NOSYSTEM: "1", |
| 79 | GIT_CONFIG_GLOBAL: "/dev/null", |
| 80 | GIT_AUTHOR_NAME: "Bench Bot", |
| 81 | GIT_AUTHOR_EMAIL: "bench@example.com", |
| 82 | GIT_COMMITTER_NAME: "Bench Bot", |
| 83 | GIT_COMMITTER_EMAIL: "bench@example.com", |
| 84 | GIT_AUTHOR_DATE: "2026-01-01T00:00:00Z", |
| 85 | GIT_COMMITTER_DATE: "2026-01-01T00:00:00Z", |
| 86 | }, |
| 87 | extendEnv: true, |
| 88 | }); |
| 89 | const [exitCode, stdout, stderr] = yield* Effect.all( |
| 90 | [ |
| 91 | handle.exitCode, |
| 92 | Stream.mkString(Stream.decodeText(handle.stdout)), |
| 93 | Stream.mkString(Stream.decodeText(handle.stderr)), |
| 94 | ], |
| 95 | { concurrency: 3 }, |
| 96 | ); |
| 97 | return { script, exitCode, stdout: stdout.trim(), stderr }; |
| 98 | }, Effect.timeout("240 seconds")); |
| 99 | |
| 100 | const mustSh = Effect.fn(function* (cwd: string, script: string) { |
| 101 | const result = yield* sh(cwd, script); |
| 102 | if (result.exitCode !== 0) return yield* new ShellError(result); |
| 103 | return result; |
| 104 | }); |
| 105 | |
| 106 | const tempDir = Effect.gen(function* () { |
| 107 | const fs = yield* FileSystem.FileSystem; |
| 108 | return yield* fs.makeTempDirectory({ prefix: "git-bench-" }); |
| 109 | }); |
| 110 | |
| 111 | const edgeRetry = <A, E, R>(effect: Effect.Effect<A, E, R>) => |
| 112 | effect.pipe( |
| 113 | Effect.timeout("15 seconds"), |
| 114 | Effect.retry({ schedule: Schedule.spaced("1500 millis"), times: 40 }), |
| 115 | ); |
| 116 | |
| 117 | const purgeRepo = Effect.fn(function* ( |
| 118 | url: string, |
| 119 | owner: string, |
| 120 | repo: string, |
| 121 | ) { |
| 122 | const admin = yield* makeClient(url, TEST_SECRET); |
| 123 | yield* admin.repos.delete({ params: { owner, repo } }).pipe( |
| 124 | Effect.catchTag("RepoNotFound", () => Effect.void), |
| 125 | edgeRetry, |
| 126 | ); |
| 127 | yield* admin.repos.get({ params: { owner, repo } }).pipe( |
| 128 | edgeRetry, |
| 129 | Effect.as(false), |
| 130 | Effect.catchTag("RepoNotFound", () => Effect.succeed(true)), |
| 131 | Effect.repeat({ |
| 132 | schedule: Schedule.spaced("1 second"), |
| 133 | until: (gone) => gone, |
| 134 | times: 60, |
| 135 | }), |
| 136 | ); |
| 137 | }); |
| 138 | |
| 139 | const freshRepo = Effect.fn(function* ( |
| 140 | url: string, |
| 141 | owner: string, |
| 142 | name: string, |
| 143 | options?: { readonly public?: boolean }, |
| 144 | ) { |
| 145 | const admin = yield* makeClient(url, TEST_SECRET); |
| 146 | const created = yield* Effect.gen(function* () { |
| 147 | yield* purgeRepo(url, owner, name); |
| 148 | return yield* admin.repos |
| 149 | .create({ payload: { owner, name, public: options?.public } }) |
| 150 | .pipe(edgeRetry); |
| 151 | }).pipe( |
| 152 | Effect.retry({ |
| 153 | while: (error: { readonly _tag?: string }) => |
| 154 | error._tag === "RepoAlreadyExists", |
| 155 | schedule: Schedule.spaced("1 second"), |
| 156 | times: 3, |
| 157 | }), |
| 158 | ); |
| 159 | const parsed = new URL(url); |
| 160 | return { |
| 161 | admin, |
| 162 | token: TEST_SECRET, |
| 163 | remote: `${parsed.protocol}//x:${TEST_SECRET}@${parsed.host}/${owner}/${name}.git`, |
| 164 | }; |
| 165 | }); |
| 166 | |
| 167 | // ── measurement ───────────────────────────────────────────────────────────── |
| 168 | |
| 169 | const results: Array<string> = []; |
| 170 | |
| 171 | /** Times an effect, records a bench line, and returns the value. */ |
| 172 | const measure = <A, E, R>( |
| 173 | label: string, |
| 174 | effect: Effect.Effect<A, E, R>, |
| 175 | describe: (value: A, ms: number) => string, |
| 176 | ) => |
| 177 | Effect.gen(function* () { |
| 178 | const started = yield* Effect.sync(() => performance.now()); |
| 179 | const value = yield* effect; |
| 180 | const ms = yield* Effect.sync(() => performance.now() - started); |
| 181 | const line = `${label.padEnd(46)} ${describe(value, ms)}`; |
| 182 | results.push(line); |
| 183 | yield* Effect.logInfo(`[bench] ${line}`); |
| 184 | return value; |
| 185 | }); |
| 186 | |
| 187 | const perSecond = (count: number, ms: number) => |
| 188 | `${((count / ms) * 1000).toFixed(1)}/s`; |
| 189 | |
| 190 | const stack = beforeAll( |
| 191 | deploy(Stack).pipe( |
| 192 | Effect.tap(({ url }) => |
| 193 | Effect.gen(function* () { |
| 194 | yield* Effect.logInfo(`git-service deployed at ${url}`); |
| 195 | const admin = yield* makeClient(url, TEST_SECRET); |
| 196 | yield* admin.repos.list({ query: {} }).pipe( |
| 197 | edgeRetry, |
| 198 | Effect.retry({ |
| 199 | schedule: Schedule.spaced("1500 millis"), |
| 200 | times: 40, |
| 201 | }), |
| 202 | ); |
| 203 | }), |
| 204 | ), |
| 205 | logLevel, |
| 206 | ), |
| 207 | ); |
| 208 | |
| 209 | afterAll( |
| 210 | Effect.gen(function* () { |
| 211 | if (results.length > 0) { |
| 212 | yield* Effect.logInfo( |
| 213 | ["", "═══ git-service throughput ═══", ...results, ""].join("\n"), |
| 214 | ); |
| 215 | } |
| 216 | }), |
| 217 | ); |
| 218 | afterAll.skipIf(!!process.env.NO_DESTROY)(destroy(Stack)); |
| 219 | |
| 220 | // ═══════════════════════════════════════════════════════════════════════════ |
| 221 | // Bottleneck 1 — closure computation / clone bandwidth on deep history |
| 222 | // ═══════════════════════════════════════════════════════════════════════════ |
| 223 | |
| 224 | test.skipIf(skipBench)( |
| 225 | "bench: deep history — push, cold clone, bundle clone, incremental fetch", |
| 226 | Effect.gen(function* () { |
| 227 | const { url } = yield* stack; |
| 228 | const repo = yield* freshRepo(url, "bench", "deep"); |
| 229 | const tmp = yield* tempDir; |
| 230 | const commits = 300 * SCALE; |
| 231 | |
| 232 | // Build the history locally first so the measurement is server-side. |
| 233 | yield* mustSh( |
| 234 | tmp, |
| 235 | ` |
| 236 | rm -rf work && git -c init.defaultBranch=main clone '${repo.remote}' work |
| 237 | cd work |
| 238 | i=1 |
| 239 | while [ $i -le ${commits} ]; do |
| 240 | printf 'line %s\\n' $i >> history.txt |
| 241 | printf 'file %s\\n' $i > "f$i.txt" |
| 242 | git add -A && git commit -q -m "c$i" |
| 243 | i=$((i+1)) |
| 244 | done |
| 245 | `, |
| 246 | ); |
| 247 | |
| 248 | yield* measure( |
| 249 | `push ${commits} commits`, |
| 250 | mustSh(tmp, `cd work && git push origin main`), |
| 251 | (_, ms) => |
| 252 | `${(ms / 1000).toFixed(1)}s (${perSecond(commits, ms)} commits)`, |
| 253 | ); |
| 254 | |
| 255 | // Cold clone: no bundle yet (the post-push alarm needs a moment), so |
| 256 | // this measures the dynamic closure walk + pack emission path. |
| 257 | yield* measure( |
| 258 | "clone (dynamic closure walk)", |
| 259 | mustSh(tmp, `rm -rf cold && git clone -q '${repo.remote}' cold`), |
| 260 | (_, ms) => `${(ms / 1000).toFixed(2)}s`, |
| 261 | ); |
| 262 | |
| 263 | // Give the bundle alarm time to cut, then clone again: same repo, same |
| 264 | // client, but served as bytes from R2 (DESIGN §12.2). |
| 265 | yield* Effect.sleep("6 seconds"); |
| 266 | yield* measure( |
| 267 | "clone (R2 bundle fast path)", |
| 268 | mustSh(tmp, `rm -rf warm && git clone -q '${repo.remote}' warm`), |
| 269 | (_, ms) => `${(ms / 1000).toFixed(2)}s`, |
| 270 | ); |
| 271 | |
| 272 | // Incremental fetch of a single new commit — the negotiation path. |
| 273 | yield* mustSh( |
| 274 | tmp, |
| 275 | `cd work && echo tip >> history.txt && git add -A && git commit -q -m tip && git push -q origin main`, |
| 276 | ); |
| 277 | yield* measure( |
| 278 | "incremental fetch (1 commit)", |
| 279 | mustSh(tmp, `cd cold && git fetch -q origin main`), |
| 280 | (_, ms) => `${ms.toFixed(0)}ms`, |
| 281 | ); |
| 282 | |
| 283 | yield* mustSh(tmp, `cd warm && git fsck --strict`); |
| 284 | }).pipe(logLevel), |
| 285 | { |
| 286 | tags: [ |
| 287 | "provider:cloudflare", |
| 288 | "provider:cloudflare:r2", |
| 289 | "provider:cloudflare:worker", |
| 290 | "live", |
| 291 | ], |
| 292 | timeout: 900_000, |
| 293 | }, |
| 294 | ); |
| 295 | |
| 296 | // ═══════════════════════════════════════════════════════════════════════════ |
| 297 | // Bottleneck 1/2 — clone TPS and bandwidth under concurrency |
| 298 | // ═══════════════════════════════════════════════════════════════════════════ |
| 299 | |
| 300 | test.skipIf(skipBench)( |
| 301 | "bench: concurrent clone storm (clone TPS on one repo)", |
| 302 | Effect.gen(function* () { |
| 303 | const { url } = yield* stack; |
| 304 | const repo = yield* freshRepo(url, "bench", "storm"); |
| 305 | const tmp = yield* tempDir; |
| 306 | const blobKiB = 256 * SCALE; |
| 307 | const clones = 8 * SCALE; |
| 308 | |
| 309 | yield* mustSh( |
| 310 | tmp, |
| 311 | ` |
| 312 | rm -rf work && git -c init.defaultBranch=main clone '${repo.remote}' work |
| 313 | cd work |
| 314 | head -c ${blobKiB * 1024} /dev/urandom > blob.bin |
| 315 | i=1; while [ $i -le 20 ]; do printf 'f %s\\n' $i > "f$i.txt"; i=$((i+1)); done |
| 316 | git add -A && git commit -q -m payload |
| 317 | git push -q origin main |
| 318 | `, |
| 319 | ); |
| 320 | // Let the bundle land so the storm hits the fast path. |
| 321 | yield* Effect.sleep("6 seconds"); |
| 322 | |
| 323 | yield* measure( |
| 324 | `${clones} concurrent clones (~${blobKiB} KiB each)`, |
| 325 | Effect.all( |
| 326 | Array.from({ length: clones }, (_, i) => |
| 327 | mustSh(tmp, `rm -rf c${i} && git clone -q '${repo.remote}' c${i}`), |
| 328 | ), |
| 329 | { concurrency: clones }, |
| 330 | ), |
| 331 | (_, ms) => |
| 332 | `${(ms / 1000).toFixed(2)}s total, ${perSecond(clones, ms)} clones, ` + |
| 333 | `~${((clones * blobKiB) / 1024 / (ms / 1000)).toFixed(1)} MiB/s`, |
| 334 | ); |
| 335 | }).pipe(logLevel), |
| 336 | { |
| 337 | tags: [ |
| 338 | "provider:cloudflare", |
| 339 | "provider:cloudflare:r2", |
| 340 | "provider:cloudflare:worker", |
| 341 | "live", |
| 342 | ], |
| 343 | timeout: 900_000, |
| 344 | }, |
| 345 | ); |
| 346 | |
| 347 | // ═══════════════════════════════════════════════════════════════════════════ |
| 348 | // Bottleneck 3 — push serialization (inherent: a ref is a serialization point) |
| 349 | // ═══════════════════════════════════════════════════════════════════════════ |
| 350 | |
| 351 | test.skipIf(skipBench)( |
| 352 | "bench: push TPS, serial and concurrent-to-distinct-branches", |
| 353 | Effect.gen(function* () { |
| 354 | const { url } = yield* stack; |
| 355 | const repo = yield* freshRepo(url, "bench", "push"); |
| 356 | const tmp = yield* tempDir; |
| 357 | const pushes = 10 * SCALE; |
| 358 | |
| 359 | yield* mustSh( |
| 360 | tmp, |
| 361 | ` |
| 362 | rm -rf work && git -c init.defaultBranch=main clone '${repo.remote}' work |
| 363 | cd work && echo base > base.txt && git add -A && git commit -q -m base |
| 364 | git push -q origin main |
| 365 | `, |
| 366 | ); |
| 367 | |
| 368 | yield* measure( |
| 369 | `${pushes} serial pushes (same branch)`, |
| 370 | mustSh( |
| 371 | tmp, |
| 372 | ` |
| 373 | cd work |
| 374 | i=1 |
| 375 | while [ $i -le ${pushes} ]; do |
| 376 | printf '%s\\n' $i >> base.txt |
| 377 | git add -A && git commit -q -m "p$i" |
| 378 | git push -q origin main |
| 379 | i=$((i+1)) |
| 380 | done |
| 381 | `, |
| 382 | ), |
| 383 | (_, ms) => `${(ms / 1000).toFixed(1)}s (${perSecond(pushes, ms)})`, |
| 384 | ); |
| 385 | |
| 386 | // Distinct branches: admission is by ingest MEMORY, not by count |
| 387 | // (DESIGN.md §2.1), so ordinary kilobyte-sized pushes to different |
| 388 | // branches should now overlap instead of queueing behind each other. |
| 389 | const branches = 4 * SCALE; |
| 390 | yield* mustSh( |
| 391 | tmp, |
| 392 | ` |
| 393 | cd work |
| 394 | i=1 |
| 395 | while [ $i -le ${branches} ]; do |
| 396 | git branch -f "side$i" HEAD |
| 397 | printf 'side %s\\n' $i > "side$i.txt" |
| 398 | i=$((i+1)) |
| 399 | done |
| 400 | `, |
| 401 | ); |
| 402 | yield* measure( |
| 403 | `${branches} concurrent pushes (distinct branches)`, |
| 404 | Effect.all( |
| 405 | Array.from({ length: branches }, (_, i) => |
| 406 | mustSh(tmp, `cd work && git push -q origin side${i + 1}`), |
| 407 | ), |
| 408 | { concurrency: branches }, |
| 409 | ), |
| 410 | (_, ms) => `${(ms / 1000).toFixed(2)}s (${perSecond(branches, ms)})`, |
| 411 | ); |
| 412 | }).pipe(logLevel), |
| 413 | { |
| 414 | tags: [ |
| 415 | "provider:cloudflare", |
| 416 | "provider:cloudflare:r2", |
| 417 | "provider:cloudflare:worker", |
| 418 | "live", |
| 419 | ], |
| 420 | timeout: 900_000, |
| 421 | }, |
| 422 | ); |
| 423 | |
| 424 | // ═══════════════════════════════════════════════════════════════════════════ |
| 425 | // Bottleneck 4 — Registry singleton (control-plane TPS) |
| 426 | // ═══════════════════════════════════════════════════════════════════════════ |
| 427 | |
| 428 | test.skipIf(skipBench)( |
| 429 | "bench: control plane — repo create/list/delete through the Registry", |
| 430 | Effect.gen(function* () { |
| 431 | const { url } = yield* stack; |
| 432 | const admin = yield* makeClient(url, TEST_SECRET); |
| 433 | const repos = 10 * SCALE; |
| 434 | const names = Array.from({ length: repos }, (_, i) => `ctl-${i}`); |
| 435 | |
| 436 | yield* Effect.all( |
| 437 | names.map((name) => purgeRepo(url, "bench", name)), |
| 438 | { concurrency: 5 }, |
| 439 | ); |
| 440 | |
| 441 | yield* measure( |
| 442 | `${repos} concurrent repo creates`, |
| 443 | Effect.all( |
| 444 | names.map((name) => |
| 445 | admin.repos.create({ payload: { owner: "bench", name } }), |
| 446 | ), |
| 447 | { concurrency: repos }, |
| 448 | ), |
| 449 | (created, ms) => { |
| 450 | expect(created.length).toBe(repos); |
| 451 | return `${(ms / 1000).toFixed(2)}s (${perSecond(repos, ms)})`; |
| 452 | }, |
| 453 | ); |
| 454 | |
| 455 | yield* measure( |
| 456 | "repo list (single Registry read)", |
| 457 | admin.repos.list({ query: { owner: "bench" } }), |
| 458 | (page, ms) => `${page.items.length} rows in ${ms.toFixed(0)}ms`, |
| 459 | ); |
| 460 | |
| 461 | yield* measure( |
| 462 | "resolve-cached repo reads (20 sequential GETs)", |
| 463 | Effect.all( |
| 464 | Array.from({ length: 20 }, () => |
| 465 | admin.repos.get({ params: { owner: "bench", repo: names[0]! } }), |
| 466 | ), |
| 467 | { concurrency: 1 }, |
| 468 | ), |
| 469 | (_, ms) => `${(ms / 20).toFixed(0)}ms each (${perSecond(20, ms)})`, |
| 470 | ); |
| 471 | |
| 472 | yield* Effect.all( |
| 473 | names.map((name) => |
| 474 | admin.repos |
| 475 | .delete({ params: { owner: "bench", repo: name } }) |
| 476 | .pipe(Effect.catchTag("RepoNotFound", () => Effect.void)), |
| 477 | ), |
| 478 | { concurrency: 5 }, |
| 479 | ); |
| 480 | }).pipe(logLevel), |
| 481 | { |
| 482 | tags: [ |
| 483 | "provider:cloudflare", |
| 484 | "provider:cloudflare:r2", |
| 485 | "provider:cloudflare:worker", |
| 486 | "live", |
| 487 | ], |
| 488 | timeout: 900_000, |
| 489 | }, |
| 490 | ); |
| 491 | |
| 492 | // ═══════════════════════════════════════════════════════════════════════════ |
| 493 | // Bottleneck 6 — compaction throughput (R2 write path) |
| 494 | // ═══════════════════════════════════════════════════════════════════════════ |
| 495 | |
| 496 | test.skipIf(skipBench)( |
| 497 | "bench: compaction — loose rows into an R2 pack, then packed reads", |
| 498 | Effect.gen(function* () { |
| 499 | const { url } = yield* stack; |
| 500 | const repo = yield* freshRepo(url, "bench", "compact"); |
| 501 | const tmp = yield* tempDir; |
| 502 | const files = 200 * SCALE; |
| 503 | |
| 504 | yield* mustSh( |
| 505 | tmp, |
| 506 | ` |
| 507 | rm -rf work && git -c init.defaultBranch=main clone '${repo.remote}' work |
| 508 | cd work |
| 509 | i=1 |
| 510 | while [ $i -le ${files} ]; do |
| 511 | head -c 4096 /dev/urandom | base64 > "b$i.txt" |
| 512 | i=$((i+1)) |
| 513 | done |
| 514 | git add -A && git commit -q -m payload |
| 515 | git push -q origin main |
| 516 | `, |
| 517 | ); |
| 518 | |
| 519 | const before = yield* repo.admin.repos.get({ |
| 520 | params: { owner: "bench", repo: "compact" }, |
| 521 | }); |
| 522 | |
| 523 | yield* measure( |
| 524 | `compact ${before.objects.loose} loose objects into R2`, |
| 525 | Effect.gen(function* () { |
| 526 | yield* repo.admin.repos.compact({ |
| 527 | params: { owner: "bench", repo: "compact" }, |
| 528 | }); |
| 529 | return yield* repo.admin.repos |
| 530 | .get({ params: { owner: "bench", repo: "compact" } }) |
| 531 | .pipe( |
| 532 | Effect.map((found) => found.objects), |
| 533 | Effect.repeat({ |
| 534 | schedule: Schedule.spaced("500 millis"), |
| 535 | until: (objects: { loose: number; packed: number }) => |
| 536 | objects.packed > 0 && objects.loose === 0, |
| 537 | times: 120, |
| 538 | }), |
| 539 | ); |
| 540 | }), |
| 541 | (objects, ms) => |
| 542 | `${objects.packed} packed, ${(objects.bytes / 1024 / 1024).toFixed(1)} MiB in ${(ms / 1000).toFixed(1)}s`, |
| 543 | ); |
| 544 | |
| 545 | // Every object now reads through ranged R2 GETs. |
| 546 | yield* measure( |
| 547 | "clone from a fully packed repo (ranged R2 reads)", |
| 548 | mustSh(tmp, `rm -rf packed && git clone -q '${repo.remote}' packed`), |
| 549 | (_, ms) => `${(ms / 1000).toFixed(2)}s`, |
| 550 | ); |
| 551 | yield* mustSh(tmp, `cd packed && git fsck --strict`); |
| 552 | }).pipe(logLevel), |
| 553 | { |
| 554 | tags: [ |
| 555 | "provider:cloudflare", |
| 556 | "provider:cloudflare:r2", |
| 557 | "provider:cloudflare:worker", |
| 558 | "live", |
| 559 | ], |
| 560 | timeout: 900_000, |
| 561 | }, |
| 562 | ); |
| 563 | |
| 564 | // ═══════════════════════════════════════════════════════════════════════════ |
| 565 | // Bottleneck 1/2 — SERVER-side clone capacity (no git process in the way) |
| 566 | // ═══════════════════════════════════════════════════════════════════════════ |
| 567 | |
| 568 | test.skipIf(skipBench)( |
| 569 | "bench: raw upload-pack throughput (server capacity, no git CLI)", |
| 570 | Effect.gen(function* () { |
| 571 | const { url } = yield* stack; |
| 572 | const repo = yield* freshRepo(url, "bench", "raw"); |
| 573 | const tmp = yield* tempDir; |
| 574 | const blobKiB = 512 * SCALE; |
| 575 | |
| 576 | yield* mustSh( |
| 577 | tmp, |
| 578 | ` |
| 579 | rm -rf work && git -c init.defaultBranch=main clone '${repo.remote}' work |
| 580 | cd work |
| 581 | head -c ${blobKiB * 1024} /dev/urandom | base64 > payload.txt |
| 582 | i=1; while [ $i -le 50 ]; do printf 'f %s\\n' $i > "f$i.txt"; i=$((i+1)); done |
| 583 | git add -A && git commit -q -m payload |
| 584 | git push -q origin main |
| 585 | `, |
| 586 | ); |
| 587 | const head = (yield* mustSh(tmp, `cd work && git rev-parse HEAD`)).stdout; |
| 588 | yield* Effect.sleep("6 seconds"); // let the bundle land |
| 589 | |
| 590 | const client = yield* HttpClient.HttpClient; |
| 591 | const pkt = (line: string) => |
| 592 | `${(line.length + 4).toString(16).padStart(4, "0")}${line}`; |
| 593 | const body = `${pkt(`want ${head}\n`)}0000${pkt("done\n")}`; |
| 594 | |
| 595 | /** One full clone over raw HTTP; resolves to bytes received. */ |
| 596 | const rawClone = client |
| 597 | .execute( |
| 598 | HttpClientRequest.post(`${url}/bench/raw.git/git-upload-pack`).pipe( |
| 599 | HttpClientRequest.setHeaders({ |
| 600 | authorization: `Bearer ${repo.token}`, |
| 601 | "content-type": "application/x-git-upload-pack-request", |
| 602 | }), |
| 603 | HttpClientRequest.bodyText(body), |
| 604 | ), |
| 605 | ) |
| 606 | .pipe( |
| 607 | Effect.flatMap((response) => response.arrayBuffer), |
| 608 | Effect.map((buffer) => buffer.byteLength), |
| 609 | ); |
| 610 | |
| 611 | // Warm one request so the measurements exclude cold start. |
| 612 | const packBytes = yield* rawClone; |
| 613 | expect(packBytes).toBeGreaterThan(1024); |
| 614 | |
| 615 | // The same clone, but with a `have` the server does not know. That |
| 616 | // disqualifies the bundle (bundleCovers refuses any request carrying |
| 617 | // haves) while producing the SAME pack, so the two lines below are a |
| 618 | // like-for-like v1-dynamic vs v2-bundle comparison. |
| 619 | const unknownHave = "f".repeat(40); |
| 620 | const dynamicClone = client |
| 621 | .execute( |
| 622 | HttpClientRequest.post(`${url}/bench/raw.git/git-upload-pack`).pipe( |
| 623 | HttpClientRequest.setHeaders({ |
| 624 | authorization: `Bearer ${repo.token}`, |
| 625 | "content-type": "application/x-git-upload-pack-request", |
| 626 | }), |
| 627 | HttpClientRequest.bodyText( |
| 628 | `${pkt(`want ${head}\n`)}${pkt(`have ${unknownHave}\n`)}0000${pkt("done\n")}`, |
| 629 | ), |
| 630 | ), |
| 631 | ) |
| 632 | .pipe( |
| 633 | Effect.flatMap((response) => response.arrayBuffer), |
| 634 | Effect.map((buffer) => buffer.byteLength), |
| 635 | ); |
| 636 | |
| 637 | const throughput = ( |
| 638 | sizes: ReadonlyArray<number>, |
| 639 | ms: number, |
| 640 | n: number, |
| 641 | ) => { |
| 642 | const total = sizes.reduce((sum, bytes) => sum + bytes, 0); |
| 643 | return ( |
| 644 | `${perSecond(n, ms)} clones, ` + |
| 645 | `${(total / 1024 / 1024 / (ms / 1000)).toFixed(1)} MiB/s` |
| 646 | ); |
| 647 | }; |
| 648 | |
| 649 | for (const concurrency of [1, 8, 32]) { |
| 650 | yield* measure( |
| 651 | `raw clone x${concurrency} — bundle (${(packBytes / 1024).toFixed(0)} KiB)`, |
| 652 | Effect.all( |
| 653 | Array.from({ length: concurrency }, () => rawClone), |
| 654 | { concurrency }, |
| 655 | ), |
| 656 | (sizes, ms) => throughput(sizes, ms, concurrency), |
| 657 | ); |
| 658 | yield* measure( |
| 659 | `raw clone x${concurrency} — dynamic closure walk`, |
| 660 | Effect.all( |
| 661 | Array.from({ length: concurrency }, () => dynamicClone), |
| 662 | { concurrency }, |
| 663 | ), |
| 664 | (sizes, ms) => throughput(sizes, ms, concurrency), |
| 665 | ); |
| 666 | } |
| 667 | }).pipe(logLevel), |
| 668 | { |
| 669 | tags: [ |
| 670 | "provider:cloudflare", |
| 671 | "provider:cloudflare:r2", |
| 672 | "provider:cloudflare:worker", |
| 673 | "live", |
| 674 | ], |
| 675 | timeout: 900_000, |
| 676 | }, |
| 677 | ); |
| 678 | |
| 679 | // ═══════════════════════════════════════════════════════════════════════════ |
| 680 | // Bottleneck 2 — subrequest limit vs. fragmented packs (DESIGN §15.2) |
| 681 | // ═══════════════════════════════════════════════════════════════════════════ |
| 682 | |
| 683 | test.skipIf(skipBench)( |
| 684 | "bench: dynamic fetch over a compacted repo with >1000 objects", |
| 685 | Effect.gen(function* () { |
| 686 | const { url } = yield* stack; |
| 687 | const repo = yield* freshRepo(url, "bench", "subreq"); |
| 688 | const tmp = yield* tempDir; |
| 689 | // Comfortably past the Workers 1000-subrequest cap once compacted: |
| 690 | // one ranged R2 GET per object would fail outright. |
| 691 | const files = 1200; |
| 692 | |
| 693 | yield* mustSh( |
| 694 | tmp, |
| 695 | ` |
| 696 | rm -rf work && git -c init.defaultBranch=main clone '${repo.remote}' work |
| 697 | cd work |
| 698 | i=1 |
| 699 | while [ $i -le ${files} ]; do |
| 700 | printf 'payload %s\\n' $i > "f$i.txt" |
| 701 | i=$((i+1)) |
| 702 | done |
| 703 | git add -A && git commit -q -m bulk |
| 704 | git push -q origin main |
| 705 | `, |
| 706 | ); |
| 707 | |
| 708 | // Force everything into an R2 pack. |
| 709 | yield* repo.admin.repos.compact({ |
| 710 | params: { owner: "bench", repo: "subreq" }, |
| 711 | }); |
| 712 | const packed = yield* repo.admin.repos |
| 713 | .get({ params: { owner: "bench", repo: "subreq" } }) |
| 714 | .pipe( |
| 715 | Effect.map((found) => found.objects), |
| 716 | Effect.repeat({ |
| 717 | schedule: Schedule.spaced("1 second"), |
| 718 | until: (objects: { loose: number; packed: number }) => |
| 719 | objects.packed > 1000 && objects.loose === 0, |
| 720 | times: 120, |
| 721 | }), |
| 722 | ); |
| 723 | expect(packed.packed).toBeGreaterThan(1000); |
| 724 | |
| 725 | const head = (yield* mustSh(tmp, `cd work && git rev-parse HEAD`)).stdout; |
| 726 | const client = yield* HttpClient.HttpClient; |
| 727 | const pkt = (line: string) => |
| 728 | `${(line.length + 4).toString(16).padStart(4, "0")}${line}`; |
| 729 | |
| 730 | // A `have` the server does not know bypasses the bundle, so this is the |
| 731 | // dynamic path reading every object out of the pack. |
| 732 | yield* measure( |
| 733 | `dynamic fetch, ${packed.packed} packed objects`, |
| 734 | client |
| 735 | .execute( |
| 736 | HttpClientRequest.post( |
| 737 | `${url}/bench/subreq.git/git-upload-pack`, |
| 738 | ).pipe( |
| 739 | HttpClientRequest.setHeaders({ |
| 740 | authorization: `Bearer ${repo.token}`, |
| 741 | "content-type": "application/x-git-upload-pack-request", |
| 742 | }), |
| 743 | HttpClientRequest.bodyText( |
| 744 | `${pkt(`want ${head}\n`)}${pkt(`have ${"f".repeat(40)}\n`)}0000${pkt("done\n")}`, |
| 745 | ), |
| 746 | ), |
| 747 | ) |
| 748 | .pipe( |
| 749 | Effect.flatMap((response) => |
| 750 | response.arrayBuffer.pipe( |
| 751 | Effect.map((buffer) => ({ |
| 752 | status: response.status, |
| 753 | bytes: buffer.byteLength, |
| 754 | })), |
| 755 | ), |
| 756 | ), |
| 757 | ), |
| 758 | (result, ms) => |
| 759 | `status ${result.status}, ${(result.bytes / 1024).toFixed(0)} KiB in ${(ms / 1000).toFixed(2)}s`, |
| 760 | ).pipe( |
| 761 | Effect.tap((result) => { |
| 762 | // The whole point: this must NOT fail on the subrequest cap. |
| 763 | expect(result.status).toBe(200); |
| 764 | expect(result.bytes).toBeGreaterThan(10_000); |
| 765 | return Effect.void; |
| 766 | }), |
| 767 | ); |
| 768 | |
| 769 | // And the repo still clones cleanly through the same path. |
| 770 | yield* mustSh(tmp, `rm -rf verify && git clone -q '${repo.remote}' verify`); |
| 771 | yield* mustSh(tmp, `cd verify && git fsck --strict`); |
| 772 | }).pipe(logLevel), |
| 773 | { |
| 774 | tags: [ |
| 775 | "provider:cloudflare", |
| 776 | "provider:cloudflare:r2", |
| 777 | "provider:cloudflare:worker", |
| 778 | "live", |
| 779 | ], |
| 780 | timeout: 900_000, |
| 781 | }, |
| 782 | ); |
| 783 | |
| 784 | // ═══════════════════════════════════════════════════════════════════════════ |
| 785 | // Bottleneck 1 — push admission. The Repo DO bounds ingest by MEMORY, not |
| 786 | // by push count, so a small push no longer waits out a large upload. Both |
| 787 | // arms run in ONE case so the comparison is same-repo, same-minute. |
| 788 | // ═══════════════════════════════════════════════════════════════════════════ |
| 789 | |
| 790 | test.skipIf(skipBench)( |
| 791 | "bench: small-push latency alone vs. behind a large upload", |
| 792 | Effect.gen(function* () { |
| 793 | const { url } = yield* stack; |
| 794 | const repo = yield* freshRepo(url, "bench", "admit"); |
| 795 | const tmp = yield* tempDir; |
| 796 | const bigMiB = 12 * SCALE; |
| 797 | |
| 798 | yield* mustSh( |
| 799 | tmp, |
| 800 | ` |
| 801 | rm -rf work && git -c init.defaultBranch=main clone '${repo.remote}' work |
| 802 | cd work |
| 803 | printf 'seed\n' > seed.txt |
| 804 | git add -A && git commit -q -m seed |
| 805 | git push -q origin main |
| 806 | `, |
| 807 | ); |
| 808 | |
| 809 | /** A one-file push to its own branch — kilobytes of actual work. */ |
| 810 | const smallPush = (n: number) => |
| 811 | mustSh( |
| 812 | tmp, |
| 813 | ` |
| 814 | cd work |
| 815 | git checkout -q -B "tiny${n}" main |
| 816 | printf 'tiny %s\n' ${n} > "tiny${n}.txt" |
| 817 | git add -A && git commit -q -m "tiny${n}" |
| 818 | git push -q origin "tiny${n}" |
| 819 | git checkout -q main |
| 820 | `, |
| 821 | ); |
| 822 | |
| 823 | // Uncontended baseline. |
| 824 | yield* smallPush(1); |
| 825 | yield* measure( |
| 826 | "small push, uncontended", |
| 827 | smallPush(2), |
| 828 | (_, ms) => `${ms.toFixed(0)} ms`, |
| 829 | ); |
| 830 | |
| 831 | // Now the same push while a large body is still uploading. Under |
| 832 | // count-based admission the small push waited for the whole upload; |
| 833 | // under a byte budget it only needs its own MiB. |
| 834 | yield* mustSh( |
| 835 | tmp, |
| 836 | ` |
| 837 | cd work |
| 838 | git checkout -q -B heavy main |
| 839 | head -c ${bigMiB * 1024 * 1024} /dev/urandom > heavy.bin |
| 840 | git add -A && git commit -q -m heavy |
| 841 | git checkout -q main |
| 842 | `, |
| 843 | ); |
| 844 | // Fork the heavy push so the measurement covers the SMALL push alone, |
| 845 | // not the pair — timing both together just reports the heavy one. |
| 846 | const heavy = yield* Effect.forkChild( |
| 847 | mustSh(tmp, `cd work && git push -q origin heavy`), |
| 848 | ); |
| 849 | yield* measure( |
| 850 | `small push, while a ${bigMiB} MiB upload is in flight`, |
| 851 | smallPush(3), |
| 852 | (_, ms) => `${ms.toFixed(0)} ms`, |
| 853 | ); |
| 854 | yield* Fiber.join(heavy); |
| 855 | }).pipe(logLevel), |
| 856 | { |
| 857 | tags: [ |
| 858 | "provider:cloudflare", |
| 859 | "provider:cloudflare:r2", |
| 860 | "provider:cloudflare:worker", |
| 861 | "live", |
| 862 | ], |
| 863 | timeout: 900_000, |
| 864 | }, |
| 865 | ); |
| 866 | |
| 867 | // ═══════════════════════════════════════════════════════════════════════════ |
| 868 | // Bottleneck 2 — bundle staleness. A ref moving after the bundle was cut |
| 869 | // used to drop every clone onto the dynamic closure walk; the bundle is now |
| 870 | // spliced with a small delta pack instead. Fresh and stale are measured in |
| 871 | // the same run against the same repo. |
| 872 | // ═══════════════════════════════════════════════════════════════════════════ |
| 873 | |
| 874 | test.skipIf(skipBench)( |
| 875 | "bench: clone TPS against a FRESH vs a STALE bundle", |
| 876 | Effect.gen(function* () { |
| 877 | const { url } = yield* stack; |
| 878 | const repo = yield* freshRepo(url, "bench", "stale"); |
| 879 | const tmp = yield* tempDir; |
| 880 | const blobKiB = 512 * SCALE; |
| 881 | |
| 882 | yield* mustSh( |
| 883 | tmp, |
| 884 | ` |
| 885 | rm -rf work && git -c init.defaultBranch=main clone '${repo.remote}' work |
| 886 | cd work |
| 887 | head -c ${blobKiB * 1024} /dev/urandom | base64 > payload.txt |
| 888 | i=1; while [ $i -le 50 ]; do printf 'f %s\n' $i > "f$i.txt"; i=$((i+1)); done |
| 889 | git add -A && git commit -q -m payload |
| 890 | git push -q origin main |
| 891 | `, |
| 892 | ); |
| 893 | yield* Effect.sleep("8 seconds"); // let the bundle land |
| 894 | |
| 895 | const client = yield* HttpClient.HttpClient; |
| 896 | const pkt = (line: string) => |
| 897 | `${(line.length + 4).toString(16).padStart(4, "0")}${line}`; |
| 898 | const request = (head: string) => |
| 899 | client.execute( |
| 900 | HttpClientRequest.post(`${url}/bench/stale.git/git-upload-pack`).pipe( |
| 901 | HttpClientRequest.setHeaders({ |
| 902 | authorization: `Bearer ${repo.token}`, |
| 903 | "content-type": "application/x-git-upload-pack-request", |
| 904 | }), |
| 905 | HttpClientRequest.bodyText( |
| 906 | `${pkt(`want ${head}\n`)}0000${pkt("done\n")}`, |
| 907 | ), |
| 908 | ), |
| 909 | ); |
| 910 | const cloneAt = (head: string) => |
| 911 | request(head).pipe( |
| 912 | Effect.flatMap((response) => response.arrayBuffer), |
| 913 | Effect.map((buffer) => buffer.byteLength), |
| 914 | ); |
| 915 | /** Which plane answered — `do-bundle:bundle`, `…:spliced`, or none. */ |
| 916 | const planeAt = (head: string) => |
| 917 | request(head).pipe( |
| 918 | Effect.map( |
| 919 | (response) => response.headers["x-git-served-by"] ?? "dynamic", |
| 920 | ), |
| 921 | ); |
| 922 | |
| 923 | const N = 32 * SCALE; |
| 924 | const throughput = (sizes: ReadonlyArray<number>, ms: number) => { |
| 925 | const total = sizes.reduce((sum, bytes) => sum + bytes, 0); |
| 926 | return ( |
| 927 | `${perSecond(sizes.length, ms)} clones, ` + |
| 928 | `${(total / 1024 / 1024 / (ms / 1000)).toFixed(1)} MiB/s` |
| 929 | ); |
| 930 | }; |
| 931 | |
| 932 | const fresh = (yield* mustSh(tmp, `cd work && git rev-parse HEAD`)).stdout; |
| 933 | yield* cloneAt(fresh); // warm |
| 934 | const freshPlane = yield* planeAt(fresh); |
| 935 | results.push(` fresh arm served by: ${freshPlane}`); |
| 936 | yield* measure( |
| 937 | `${N} clones, FRESH bundle, concurrency 32`, |
| 938 | Effect.all( |
| 939 | Array.from({ length: N }, () => cloneAt(fresh)), |
| 940 | { concurrency: 32 }, |
| 941 | ), |
| 942 | throughput, |
| 943 | ); |
| 944 | |
| 945 | // Move main past the bundle and clone immediately: the bundle on disk |
| 946 | // no longer names this tip. |
| 947 | yield* mustSh( |
| 948 | tmp, |
| 949 | ` |
| 950 | cd work |
| 951 | printf 'more\n' > more.txt |
| 952 | git add -A && git commit -q -m more |
| 953 | git push -q origin main |
| 954 | `, |
| 955 | ); |
| 956 | const stale = (yield* mustSh(tmp, `cd work && git rev-parse HEAD`)).stdout; |
| 957 | const warmStale = yield* cloneAt(stale); |
| 958 | expect(warmStale).toBeGreaterThan(1024); |
| 959 | const stalePlane = yield* planeAt(stale); |
| 960 | results.push(` stale arm served by: ${stalePlane}`); |
| 961 | yield* measure( |
| 962 | `${N} clones, STALE bundle (dynamic fallback), concurrency 32`, |
| 963 | Effect.all( |
| 964 | Array.from({ length: N }, () => cloneAt(stale)), |
| 965 | { concurrency: 32 }, |
| 966 | ), |
| 967 | throughput, |
| 968 | ); |
| 969 | }).pipe(logLevel), |
| 970 | { |
| 971 | tags: [ |
| 972 | "provider:cloudflare", |
| 973 | "provider:cloudflare:r2", |
| 974 | "provider:cloudflare:worker", |
| 975 | "live", |
| 976 | ], |
| 977 | timeout: 900_000, |
| 978 | }, |
| 979 | ); |
| 980 | |
| 981 | // ═══════════════════════════════════════════════════════════════════════════ |
| 982 | // Single large clone — where does the time go? Splits time-to-first-byte |
| 983 | // (planning: Worker -> DO -> R2 HEAD -> marker) from transfer (bytes on |
| 984 | // the wire), per serving path, on a real ~36 MiB repository. This is the |
| 985 | // instrument for "clones feel slow": a large TTFB is planning latency, a |
| 986 | // low MiB/s with small TTFB is the streaming path itself. |
| 987 | // ═══════════════════════════════════════════════════════════════════════════ |
| 988 | |
| 989 | test.skipIf(skipBench)( |
| 990 | "bench: single large clone — TTFB vs transfer, by serving path", |
| 991 | Effect.gen(function* () { |
| 992 | const { url } = yield* stack; |
| 993 | const tmp = yield* tempDir; |
| 994 | |
| 995 | // Same source as the real-world suite: this monorepo, depth-1, as one |
| 996 | // root commit — ~13.6k objects, ~36 MiB. |
| 997 | const root = (yield* mustSh(process.cwd(), `git rev-parse --show-toplevel`)) |
| 998 | .stdout; |
| 999 | yield* mustSh( |
| 1000 | tmp, |
| 1001 | ` |
| 1002 | rm -rf src |
| 1003 | git clone -q --depth 1 "file://${root}" src |
| 1004 | cd src && rm -rf .git && git init -q -b main && git add -A |
| 1005 | git commit -q -m "alchemy snapshot" |
| 1006 | `, |
| 1007 | ); |
| 1008 | const repo = yield* freshRepo(url, "bench", "large", { public: true }); |
| 1009 | yield* mustSh( |
| 1010 | tmp, |
| 1011 | `cd src && git remote add origin '${repo.remote}' && git push -q origin main`, |
| 1012 | ); |
| 1013 | const head = (yield* mustSh(tmp, `cd src && git rev-parse HEAD`)).stdout; |
| 1014 | |
| 1015 | const client = yield* HttpClient.HttpClient; |
| 1016 | const pkt = (line: string) => |
| 1017 | `${(line.length + 4).toString(16).padStart(4, "0")}${line}`; |
| 1018 | const request = (body: string) => |
| 1019 | client.execute( |
| 1020 | HttpClientRequest.post(`${url}/bench/large.git/git-upload-pack`).pipe( |
| 1021 | HttpClientRequest.setHeaders({ |
| 1022 | authorization: `Bearer ${repo.token}`, |
| 1023 | "content-type": "application/x-git-upload-pack-request", |
| 1024 | }), |
| 1025 | HttpClientRequest.bodyText(body), |
| 1026 | ), |
| 1027 | ); |
| 1028 | const bodies = { |
| 1029 | passthrough: `${pkt(`want ${head}\n`)}0000${pkt("done\n")}`, |
| 1030 | sideband: `${pkt(`want ${head} side-band-64k\n`)}0000${pkt("done\n")}`, |
| 1031 | dynamic: `${pkt(`want ${head}\n`)}${pkt(`have ${"f".repeat(40)}\n`)}0000${pkt("done\n")}`, |
| 1032 | }; |
| 1033 | |
| 1034 | /** |
| 1035 | * One clone, timed in two halves: headers (TTFB) and body drain — and |
| 1036 | * VERIFIED: the pack is de-framed and its SHA-1 trailer checked. A |
| 1037 | * stream that aborts mid-body can reach a lenient HTTP client as a |
| 1038 | * clean end (seen once: 0.5 MiB of a 40.6 MiB pack, test green), so |
| 1039 | * bytes alone prove nothing. |
| 1040 | */ |
| 1041 | const timedClone = (body: string) => |
| 1042 | Effect.gen(function* () { |
| 1043 | const t0 = performance.now(); |
| 1044 | const response = yield* request(body); |
| 1045 | const ttfbMs = performance.now() - t0; |
| 1046 | const raw = new Uint8Array(yield* response.arrayBuffer); |
| 1047 | const totalMs = performance.now() - t0; |
| 1048 | const pack = yield* Effect.sync(() => |
| 1049 | verifyPackResponse(raw, body.includes("side-band-64k")), |
| 1050 | ); |
| 1051 | expect(pack.error, `pack response: ${pack.error}`).toBeUndefined(); |
| 1052 | return { |
| 1053 | ttfbMs, |
| 1054 | totalMs, |
| 1055 | bytes: raw.byteLength, |
| 1056 | objects: pack.objects, |
| 1057 | via: response.headers["x-git-served-by"] ?? "dynamic", |
| 1058 | }; |
| 1059 | }); |
| 1060 | const describe = (r: { |
| 1061 | ttfbMs: number; |
| 1062 | totalMs: number; |
| 1063 | bytes: number; |
| 1064 | objects: number; |
| 1065 | via: string; |
| 1066 | }) => |
| 1067 | `ttfb ${r.ttfbMs.toFixed(0)}ms, total ${(r.totalMs / 1000).toFixed(2)}s, ` + |
| 1068 | `${(r.bytes / 1048576).toFixed(1)} MiB (${r.objects} objects, trailer ok) @ ` + |
| 1069 | `${(r.bytes / 1048576 / ((r.totalMs - r.ttfbMs) / 1000)).toFixed(1)} MiB/s transfer ` + |
| 1070 | `[${r.via}]`; |
| 1071 | |
| 1072 | // Wait for the bundle to land (a 36 MiB cut takes a while), bounded. |
| 1073 | yield* timedClone(bodies.passthrough).pipe( |
| 1074 | Effect.repeat({ |
| 1075 | schedule: Schedule.spaced("5 seconds"), |
| 1076 | until: (r) => r.via.startsWith("do-bundle"), |
| 1077 | times: 36, |
| 1078 | }), |
| 1079 | ); |
| 1080 | |
| 1081 | for (const [name, body] of Object.entries(bodies)) { |
| 1082 | yield* timedClone(body); // warm |
| 1083 | yield* measure(`large clone x1 — ${name}`, timedClone(body), (r) => |
| 1084 | describe(r), |
| 1085 | ); |
| 1086 | } |
| 1087 | |
| 1088 | // Does the streaming path scale with parallel streams, or is it a |
| 1089 | // per-stream ceiling? 4 passthrough clones at once. |
| 1090 | yield* measure( |
| 1091 | "large clone x4 — passthrough, concurrent", |
| 1092 | Effect.all( |
| 1093 | Array.from({ length: 4 }, () => timedClone(bodies.passthrough)), |
| 1094 | { concurrency: 4 }, |
| 1095 | ), |
| 1096 | (rs, ms) => { |
| 1097 | const total = rs.reduce((n, r) => n + r.bytes, 0); |
| 1098 | return `${(total / 1048576 / (ms / 1000)).toFixed(1)} MiB/s aggregate over ${(ms / 1000).toFixed(2)}s`; |
| 1099 | }, |
| 1100 | ); |
| 1101 | }).pipe(logLevel), |
| 1102 | { |
| 1103 | tags: [ |
| 1104 | "provider:cloudflare", |
| 1105 | "provider:cloudflare:r2", |
| 1106 | "provider:cloudflare:worker", |
| 1107 | "live", |
| 1108 | ], |
| 1109 | timeout: 1_200_000, |
| 1110 | }, |
| 1111 | ); |
| 1112 | |
| 1113 | // ═══════════════════════════════════════════════════════════════════════════ |
| 1114 | // Bottleneck 8 — the DO on the anonymous read path (Continuity learnings, |
| 1115 | // DESIGN.md §21): every advertisement and bundle-planning hop wakes the |
| 1116 | // repo DO today, so anonymous public reads share one single-threaded |
| 1117 | // object. Baseline for the DO-less head-object fast path. |
| 1118 | // ═══════════════════════════════════════════════════════════════════════════ |
| 1119 | |
| 1120 | test.skipIf(skipBench)( |
| 1121 | "bench: anonymous reads on a PUBLIC repo (advertisement + raw clone TPS)", |
| 1122 | Effect.gen(function* () { |
| 1123 | const { url } = yield* stack; |
| 1124 | const repo = yield* freshRepo(url, "bench", "pubread", { public: true }); |
| 1125 | const tmp = yield* tempDir; |
| 1126 | const blobKiB = 256 * SCALE; |
| 1127 | |
| 1128 | yield* mustSh( |
| 1129 | tmp, |
| 1130 | ` |
| 1131 | rm -rf work && git -c init.defaultBranch=main clone '${repo.remote}' work |
| 1132 | cd work |
| 1133 | head -c ${blobKiB * 1024} /dev/urandom > blob.bin |
| 1134 | i=1; while [ $i -le 20 ]; do printf 'f %s\n' $i > "f$i.txt"; i=$((i+1)); done |
| 1135 | git add -A && git commit -q -m payload |
| 1136 | git push -q origin main |
| 1137 | `, |
| 1138 | ); |
| 1139 | const head = (yield* mustSh(tmp, `cd work && git rev-parse HEAD`)).stdout; |
| 1140 | yield* Effect.sleep("6 seconds"); // let the bundle land |
| 1141 | |
| 1142 | const client = yield* HttpClient.HttpClient; |
| 1143 | const pkt = (line: string) => |
| 1144 | `${(line.length + 4).toString(16).padStart(4, "0")}${line}`; |
| 1145 | |
| 1146 | // TOKENLESS advertisement — GET info/refs with no Authorization. |
| 1147 | const advertise = client |
| 1148 | .get(`${url}/bench/pubread.git/info/refs?service=git-upload-pack`) |
| 1149 | .pipe( |
| 1150 | Effect.flatMap((response) => response.arrayBuffer), |
| 1151 | Effect.map((buffer) => buffer.byteLength), |
| 1152 | ); |
| 1153 | // warm + correctness: the advertisement carries the tip AND proves |
| 1154 | // the DO-less path served it (not an accidental DO fall-through). |
| 1155 | const warmAdv = yield* client.get( |
| 1156 | `${url}/bench/pubread.git/info/refs?service=git-upload-pack`, |
| 1157 | ); |
| 1158 | expect(warmAdv.headers["x-git-served-by"]).toBe("head-snapshot"); |
| 1159 | const advBytes = (yield* warmAdv.arrayBuffer).byteLength; |
| 1160 | expect(advBytes).toBeGreaterThan(40); |
| 1161 | |
| 1162 | const SERIAL_N = 20; |
| 1163 | yield* measure( |
| 1164 | `${SERIAL_N} serial anonymous advertisements`, |
| 1165 | Effect.forEach(Array.from({ length: SERIAL_N }), () => advertise), |
| 1166 | (_, ms) => `${(ms / SERIAL_N).toFixed(0)} ms each`, |
| 1167 | ); |
| 1168 | |
| 1169 | const ADV_N = 128 * SCALE; |
| 1170 | yield* measure( |
| 1171 | `${ADV_N} anonymous advertisements, concurrency 32`, |
| 1172 | Effect.all( |
| 1173 | Array.from({ length: ADV_N }, () => advertise), |
| 1174 | { concurrency: 32 }, |
| 1175 | ), |
| 1176 | (sizes, ms) => `${perSecond(sizes.length, ms)} advertisements`, |
| 1177 | ); |
| 1178 | |
| 1179 | // TOKENLESS full clone over raw HTTP (want tip, done — bundle-eligible). |
| 1180 | const rawClone = client |
| 1181 | .execute( |
| 1182 | HttpClientRequest.post(`${url}/bench/pubread.git/git-upload-pack`).pipe( |
| 1183 | HttpClientRequest.setHeaders({ |
| 1184 | "content-type": "application/x-git-upload-pack-request", |
| 1185 | }), |
| 1186 | HttpClientRequest.bodyText( |
| 1187 | `${pkt(`want ${head}\n`)}0000${pkt("done\n")}`, |
| 1188 | ), |
| 1189 | ), |
| 1190 | ) |
| 1191 | .pipe( |
| 1192 | Effect.flatMap((response) => response.arrayBuffer), |
| 1193 | Effect.map((buffer) => buffer.byteLength), |
| 1194 | ); |
| 1195 | // warm + prove the bundle was served DO-lessly from the snapshot |
| 1196 | const warmClone = yield* client.execute( |
| 1197 | HttpClientRequest.post(`${url}/bench/pubread.git/git-upload-pack`).pipe( |
| 1198 | HttpClientRequest.setHeaders({ |
| 1199 | "content-type": "application/x-git-upload-pack-request", |
| 1200 | }), |
| 1201 | HttpClientRequest.bodyText( |
| 1202 | `${pkt(`want ${head}\n`)}0000${pkt("done\n")}`, |
| 1203 | ), |
| 1204 | ), |
| 1205 | ); |
| 1206 | expect(warmClone.headers["x-git-served-by"]).toBe("head-snapshot:bundle"); |
| 1207 | const first = (yield* warmClone.arrayBuffer).byteLength; |
| 1208 | expect(first).toBeGreaterThan(blobKiB * 1024 * 0.5); |
| 1209 | |
| 1210 | const CLONE_N = 64 * SCALE; |
| 1211 | yield* measure( |
| 1212 | `${CLONE_N} anonymous raw clones, concurrency 32`, |
| 1213 | Effect.all( |
| 1214 | Array.from({ length: CLONE_N }, () => rawClone), |
| 1215 | { concurrency: 32 }, |
| 1216 | ), |
| 1217 | (sizes, ms) => { |
| 1218 | const total = sizes.reduce((sum, bytes) => sum + bytes, 0); |
| 1219 | return ( |
| 1220 | `${perSecond(sizes.length, ms)} clones, ` + |
| 1221 | `${(total / 1024 / 1024 / (ms / 1000)).toFixed(1)} MiB/s` |
| 1222 | ); |
| 1223 | }, |
| 1224 | ); |
| 1225 | |
| 1226 | // and a real git clone with NO credentials still works |
| 1227 | const parsed = new URL(url); |
| 1228 | yield* mustSh( |
| 1229 | tmp, |
| 1230 | `rm -rf anon && git clone -q '${parsed.protocol}//${parsed.host}/bench/pubread.git' anon && cd anon && git fsck --strict`, |
| 1231 | ); |
| 1232 | }).pipe(logLevel), |
| 1233 | { |
| 1234 | tags: [ |
| 1235 | "provider:cloudflare", |
| 1236 | "provider:cloudflare:r2", |
| 1237 | "provider:cloudflare:worker", |
| 1238 | "live", |
| 1239 | ], |
| 1240 | timeout: 900_000, |
| 1241 | }, |
| 1242 | ); |
| 1243 | |
| 1244 | // ═══════════════════════════════════════════════════════════════════════════ |
| 1245 | // Bottleneck 9 — pack-count growth (Continuity learnings, DESIGN.md §21): |
| 1246 | // each compaction run writes a NEW pack and never merges, so reads over an |
| 1247 | // old repo fan out across ever more packs. Baseline for geometric merging. |
| 1248 | // ═══════════════════════════════════════════════════════════════════════════ |
| 1249 | |
| 1250 | test.skipIf(skipBench)( |
| 1251 | "bench: dynamic fetch over a repo compacted in 6 increments (pack fan-out)", |
| 1252 | Effect.gen(function* () { |
| 1253 | const { url } = yield* stack; |
| 1254 | const repo = yield* freshRepo(url, "bench", "npack"); |
| 1255 | const tmp = yield* tempDir; |
| 1256 | const ROUNDS = 6; |
| 1257 | |
| 1258 | yield* mustSh( |
| 1259 | tmp, |
| 1260 | `rm -rf work && git -c init.defaultBranch=main clone '${repo.remote}' work`, |
| 1261 | ); |
| 1262 | |
| 1263 | // ROUNDS x (push a batch, compact it) => ROUNDS separate packs today. |
| 1264 | for (let round = 1; round <= ROUNDS; round++) { |
| 1265 | yield* mustSh( |
| 1266 | tmp, |
| 1267 | ` |
| 1268 | cd work |
| 1269 | i=1; while [ $i -le 12 ]; do head -c ${32 * 1024} /dev/urandom > "r${round}-f$i.bin"; i=$((i+1)); done |
| 1270 | git add -A && git commit -q -m "round ${round}" |
| 1271 | git push -q origin main |
| 1272 | `, |
| 1273 | ); |
| 1274 | yield* repo.admin.repos |
| 1275 | .compact({ params: { owner: "bench", repo: "npack" } }) |
| 1276 | .pipe(edgeRetry); |
| 1277 | // wait until the loose rows have drained into the pack |
| 1278 | yield* repo.admin.repos |
| 1279 | .get({ params: { owner: "bench", repo: "npack" } }) |
| 1280 | .pipe( |
| 1281 | edgeRetry, |
| 1282 | Effect.repeat({ |
| 1283 | schedule: Schedule.spaced("2 seconds"), |
| 1284 | until: (meta) => meta.objects.loose === 0, |
| 1285 | times: 45, |
| 1286 | }), |
| 1287 | ); |
| 1288 | } |
| 1289 | |
| 1290 | const after = yield* repo.admin.repos |
| 1291 | .get({ params: { owner: "bench", repo: "npack" } }) |
| 1292 | .pipe(edgeRetry); |
| 1293 | results.push( |
| 1294 | `npack repo: ${after.objects.packed} packed objects after ${ROUNDS} compactions`.padEnd( |
| 1295 | 46, |
| 1296 | ), |
| 1297 | ); |
| 1298 | |
| 1299 | const head = (yield* mustSh(tmp, `cd work && git rev-parse HEAD`)).stdout; |
| 1300 | const client = yield* HttpClient.HttpClient; |
| 1301 | const pkt = (line: string) => |
| 1302 | `${(line.length + 4).toString(16).padStart(4, "0")}${line}`; |
| 1303 | |
| 1304 | // Unknown have forces the dynamic path (bundle refused): the closure |
| 1305 | // walk + object reads must range-read across every pack. |
| 1306 | const unknownHave = "f".repeat(40); |
| 1307 | const dynamicFetch = client |
| 1308 | .execute( |
| 1309 | HttpClientRequest.post(`${url}/bench/npack.git/git-upload-pack`).pipe( |
| 1310 | HttpClientRequest.setHeaders({ |
| 1311 | authorization: `Bearer ${repo.token}`, |
| 1312 | "content-type": "application/x-git-upload-pack-request", |
| 1313 | }), |
| 1314 | HttpClientRequest.bodyText( |
| 1315 | `${pkt(`want ${head}\n`)}${pkt(`have ${unknownHave}\n`)}0000${pkt("done\n")}`, |
| 1316 | ), |
| 1317 | ), |
| 1318 | ) |
| 1319 | .pipe( |
| 1320 | Effect.flatMap((response) => response.arrayBuffer), |
| 1321 | Effect.map((buffer) => buffer.byteLength), |
| 1322 | ); |
| 1323 | |
| 1324 | const warm = yield* dynamicFetch; |
| 1325 | expect(warm).toBeGreaterThan(100_000); |
| 1326 | |
| 1327 | const N = 8; |
| 1328 | yield* measure( |
| 1329 | `${N} serial dynamic fetches over ${ROUNDS}-increment repo`, |
| 1330 | Effect.forEach(Array.from({ length: N }), () => dynamicFetch), |
| 1331 | (sizes, ms) => |
| 1332 | `${(ms / N).toFixed(0)} ms/fetch, ` + |
| 1333 | `${((sizes[0] ?? 0) / 1024 / 1024).toFixed(1)} MiB pack`, |
| 1334 | ); |
| 1335 | |
| 1336 | // correctness: clone back clean through the same packs |
| 1337 | yield* mustSh(tmp, `rm -rf verify && git clone -q '${repo.remote}' verify`); |
| 1338 | yield* mustSh(tmp, `cd verify && git fsck --strict`); |
| 1339 | }).pipe(logLevel), |
| 1340 | { |
| 1341 | tags: [ |
| 1342 | "provider:cloudflare", |
| 1343 | "provider:cloudflare:r2", |
| 1344 | "provider:cloudflare:worker", |
| 1345 | "live", |
| 1346 | ], |
| 1347 | timeout: 900_000, |
| 1348 | }, |
| 1349 | ); |