No artifact run found in ${requested}. Expected a results.js

Error message

No artifact run found in ${requested}. Expected a results.json file.

What it means

Thrown by resolveArtifactRunDir (analyze-profiles.ts:112) when neither the requested path itself contains results.json nor does any of its subdirectories. The resolver first checks <requested>/results.json; if missing, it scans child dirs for ones containing results.json and picks the most recently modified; if none found, it errors. So this means there are no benchmark run artifacts at the given location.

Source

Thrown at bench/render-pipeline/analyze-profiles.ts:112

  const requestedResults = resolve(requested, 'results.json')
  if (await exists(requestedResults)) {
    return requested
  }

  const entries = await readdir(requested, { withFileTypes: true })
  const dirs = entries.filter((entry) => entry.isDirectory())
  const runs: Array<{ dir: string; mtimeMs: number }> = []

  for (const dirent of dirs) {
    const dir = resolve(requested, dirent.name)
    const resultsPath = resolve(dir, 'results.json')
    if (!(await exists(resultsPath))) continue
    const stats = await stat(resultsPath)
    runs.push({ dir, mtimeMs: stats.mtimeMs })
  }

  if (runs.length === 0) {
    throw new Error(
      `No artifact run found in ${requested}. Expected a results.json file.`
    )
  }

  runs.sort((a, b) => b.mtimeMs - a.mtimeMs)
  return runs[0].dir
}

function toPercent(part: number, total: number): string {
  if (total <= 0) return '0.00%'
  return `${((part / total) * 100).toFixed(2)}%`
}

function toMs(us: number): string {
  return `${(us / 1000).toFixed(1)}ms`
}

function sortTop(

View on GitHub (pinned to 0ae8c72462)

Solutions

Run the benchmark first to produce artifacts: `pnpm bench:render-pipeline --scenario=e2e --stream-mode=node`, which writes results.json under bench/render-pipeline/artifacts/<timestamp>/.
Omit --artifact-dir so the resolver auto-picks the latest run under the default artifacts root.
If passing --artifact-dir, give an absolute path or a path relative to the repo root (not cwd), and confirm it contains (or is a parent of dirs containing) results.json.
Check that previous runs weren't deleted; re-run if artifacts dir is empty.

Example fix

// before
//   pnpm bench:render-pipeline:analyze --artifact-dir=./out

// after — run the benchmark to create artifacts, then analyze with no arg (auto-latest)
//   pnpm bench:render-pipeline --scenario=e2e --stream-mode=node
//   pnpm bench:render-pipeline:analyze

Defensive patterns

Strategy: validation

Validate before calling

import { existsSync } from 'node:fs'
import { resolve } from 'node:path'

function hasResultsJsonsomewhere(root: string): boolean {
  if (existsSync(resolve(root, 'results.json'))) return true
  // shallow scan of immediate subdirs
  return false
}
// cheaper: just confirm the benchmark has been run once:
// ls bench/render-pipeline/artifacts/*/results.json

Type guard

function isArtifactRunPath(p: string): boolean {
  return existsSync(resolve(p, 'results.json'))
}

Try / catch

try {
  const runDir = await resolveArtifactRunDir(artifactDirArg)
} catch (err) {
  if (err instanceof Error && err.message.startsWith('No artifact run found')) {
    console.error('Run the benchmark first: pnpm bench:render-pipeline --scenario=e2e --stream-mode=node')
    process.exit(2)
  }
  throw err
}

Prevention

Run the benchmark at least once before analyzing; artifacts are produced under bench/render-pipeline/artifacts/<timestamp>/.
Pass --artifact-dir as an absolute path or relative-to-repo-root (it's resolve()'d against REPO_ROOT).
Omit --artifact-dir to auto-pick the latest run.
Don't delete artifacts/ between runs if you want history.

When it happens

Trigger: Running analyze with --artifact-dir pointing at a path that has no results.json directly and no run subdirectories containing results.json; running analyze before ever running the benchmark; passing a relative path that resolves somewhere unexpected (it is resolve()'d against REPO_ROOT, not cwd).

Common situations: First-time use: no benchmark has been run yet, so bench/render-pipeline/artifacts is empty/absent; pointing --artifact-dir at the artifacts root but the runs were written elsewhere; a typo in the path; runs were cleaned up; relative path misinterpretation because the resolver joins against REPO_ROOT.

Related errors

AI-assisted analysis of vercel/next.js@0ae8c72462 (2026-08-06). Data as JSON: /api/errors/100c8fea6624c8d0. Report an issue: GitHub.