Troubleshooting

This page collects the symptoms people hit most often and the fix for each. If none of these match, open the raw profile JSON at /_profiler/{token}/data to see exactly what was captured.

ERR_REQUIRE_ESM on NestJS 12Copy Anchor Link

The app crashes on startup with Error [ERR_REQUIRE_ESM]: require() of ES Module …/@nestjs/common/index.js from …/@eleven-labs/nest-profiler/dist/….js not supported.

NestJS 12 ships its packages as ESM and keeps supporting CommonJS applications, which load them through Node's require() of ES modules. A CommonJS application loads the profiler's CommonJS build, which loads NestJS the same way; an ESM application imports the profiler's ESM build, which never does. Node.js only allows require() of ES modules by default from 22.12 on.

• Node.js 22.0 to 22.11. Upgrade to 22.12 or newer, the minimum NestJS 12 itself requires on the 22.x line.
• require() of ES modules disabled. Some runtimes turn it off whatever the Node.js version, such as AWS Lambda on its Node.js 20, 22 and 24 runtimes. Turn it back on with NODE_OPTIONS=--experimental-require-module, as the NestJS migration guide recommends: a CommonJS NestJS 12 application needs it there anyway.

No profiles appear at allCopy Anchor Link

The list at /_profiler stays empty even after making requests.

• The app booted without the profiler. With the dev-dependency install, only the dev entrypoint loads it: start the app with main-dev.ts (nest start --entryFile main-dev -p tsconfig.dev.json), not main.ts or node dist/main. With the runtime gate, the active ProfilerModule is only loaded when isProfilerEnabled is true (e.g. PROFILER_ENABLED=true) — check the environment the app actually booted with.
• The path is ignored. Requests matching ignorePaths or the built-in defaults (favicon, robots.txt, the Chrome DevTools probe, apple-touch-icon) are skipped. Set useDefaultIgnorePaths: false to profile those too, or verify your ignorePaths / ignoreRequest predicate isn't matching your route.
• Sampling is on. A sampleRate below 1.0 profiles only a fraction of requests. Set it to 1.0 while debugging.
• In-memory storage was cleared. The default memory backend is emptied on every restart. Use storageType: 'file' (or the SQLite adapter) to keep profiles across restarts. See Storage backends.

The Logs tab is emptyCopy Anchor Link

Logs are written to the console but never show up in the profile.

• Wrap the logger in the entrypoint that loads the profiler (main-dev.ts with the dev-dependency install): app.useLogger(createProfilerLogger(new ConsoleLogger())) (import createProfilerLogger from @eleven-labs/nest-profiler), and create the app with { bufferLogs: true }. See Set up the profiler.
• app.useLogger() only captures logs that flow through NestJS's Logger: in services, use new Logger(MyService.name) rather than a ConsoleLogger instance. A logger injected directly into a service (e.g. nestjs-pino's PinoLogger) must be wrapped explicitly — through an optional LOGGER_WRAP token filled by the dev bundle, or with createProfilerLogger(...) at the injection point under the runtime gate. See Log capture.
• A CLI (nest-commander) app passes the wrapped logger to CommandFactory.run(CliDevModule, { logger: createProfilerLogger(...) }).

GraphQL operations are not profiledCopy Anchor Link

REST requests appear but GraphQL operations don't.

• The GraphQL package is missing. The core ProfilerModule is HTTP-only. Install and register @eleven-labs/nest-profiler-graphql.
• The context factory is missing. The profiler needs access to the underlying HTTP request. Configure context: ({ req }) => ({ req }) (Apollo / graphql-yoga) or context: ({ request }) => ({ request }) (Mercurius uses request, not req). Without it the profiler silently falls back to passthrough.
• Subscriptions. WebSocket-transport subscriptions are not profiled at this time.

See GraphQL support.

A collector is installed but its panel is missingCopy Anchor Link

The package is added but no tab shows up in the UI.

• The collector isn't loaded. Each collector must be imported in the feature module it instruments and gated with the same isProfilerEnabled predicate. If profiling is off, it isn't loaded.
• Wrong placement. Infra-scoped collectors (database, HTTP client, cache, messaging, GraphQL transport) must sit next to the HttpModule / ORM / broker they instrument - they cannot join a root-level ProfilingModule bundle. See Configuration.
• Nothing was captured. A panel only renders when the collector recorded at least one entry during the request. Exercise the code it instruments (run a query, hit the cache, consume a message).

Database queries are missingCopy Anchor Link

The Database (or MongoDB) tab is empty for a request that clearly hit the database.

• Queries executed outside a request context - during module initialization, in a scheduled job with no active profile, or before the middleware ran - are silently ignored because there is no active CLS profile to append to. Only queries triggered within a profiled request are captured.

/_profiler returns 401 UnauthorizedCopy Anchor Link

Access to the UI is protected.

• A security strategy (authorize predicate and/or guards) is configured and denied the request. Send the credential it expects (e.g. Authorization: Bearer <token>, a session cookie, or Basic auth), or remove the security option for local development. See Securing the UI.

A nest-commander command exits immediately and prints nothingCopy Anchor Link

After installing the profiler, a CLI command logs Starting Nest application... (or nothing at all) and the process exits 0 without ever running — no result, no error.

• The CLI root module doesn't import ConfigModule.forRoot() (runtime gate only — a dev-dependency install has no gate). ConditionalModule.registerWhen awaits ConfigModule.envVariablesLoaded from @nestjs/config, and that promise only resolves once @nestjs/config's ConfigModule.forRoot() has run. A CLI bootstrapped with CommandFactory whose root module never imports it never resolves the promise, so registration hangs — and because the internal 5s timeout timer is unref'd, the event loop just drains and the process exits 0 silently instead of throwing. Fix: import ConfigModule.forRoot() from @nestjs/config in your CLI root module. This applies to every module the CLI loads that gates something with registerWhen (core, database, RabbitMQ collectors…). HTTP apps rarely hit this because their root module already imports ConfigModule.forRoot(). See Command profiling and Enabling and disabling the profiler.

CLI command profiles don't appear next to HTTP requestsCopy Anchor Link

You profiled a nest-commander command but can't find it in the web UI.

• The CLI process and the HTTP app are separate processes with separate in-memory stores. Use file (or SQLite) storage with the same storagePath for both, so the command runs written by the CLI show up at /_profiler next to the HTTP profiles. See Command profiling and the example app.
• The command failed while commander parsed the argv, and commander exited the process itself. An unknown option, a missing required option, an invalid choices value, or an @Option() parser throwing commander's InvalidArgumentError all end in commander printing a CLI error and calling process.exit() — nothing is left running to persist a profile. An option parser that throws a plain Error is profiled (failed, with the raw value of the rejected flag). See Command profiling.