"Vovchansk (2024-06-02) 1513" by National Police of Ukraine (Liut Brigade) is licensed under CC BY 4.0.

This is Vovchansk, Ukraine, the city where the father of this library’s author was born. This is how it looks now, after the Russian invasion. If you find this library useful and would like to thank the author, please consider donating any amount via one of the following links:
・Armed Forces of Ukraine・"The Come Back Alive" foundation・
Thanks for your support!

NestJS-Pino

Platform agnostic logger for NestJS based on Pino with REQUEST CONTEXT IN EVERY LOG

---

This is the documentation for v5. Compatibility with earlier majors:

nestjs-pino	NestJS	pino	pino-http	Node.js
v5	11.0.8+, 12.0.2+	10	11	>=22.12
v4	8, 9, 10, 11	7.5+, 8, 9, 10	6.4+, 7, 8, 9, 10, 11	>=14
v1	< 8

---

Install

npm i nestjs-pino pino-http

Example

Firstly, import module with LoggerModule.forRoot(...) or LoggerModule.forRootAsync(...) only once in root module (check out module configuration docs below):

import { LoggerModule } from 'nestjs-pino';

@Module({
  imports: [LoggerModule.forRoot()],
})
class AppModule {}

Secondly, set up app logger:

import { Logger } from 'nestjs-pino';

const app = await NestFactory.create(AppModule, { bufferLogs: true });
app.useLogger(app.get(Logger));

Now you can use one of the loggers:

// NestJS standard built-in logger.
// Logs will be produced by pino internally
import { Logger } from '@nestjs/common';

export class MyService {
  private readonly logger = new Logger(MyService.name);
  foo() {
    // All logger methods have args format the same as pino, but pino methods
    // `trace` and `info` are mapped to `verbose` and `log` to satisfy
    // `LoggerService` interface of NestJS:
    this.logger.verbose({ foo: 'bar' }, 'baz %s', 'qux');
    this.logger.debug('foo %s %o', 'bar', { baz: 'qux' });
    this.logger.log('foo');
  }
}

Usage of the standard logger is recommended and idiomatic for NestJS. But there is one more option to use:

import { PinoLogger, InjectPinoLogger } from 'nestjs-pino';

export class MyService {
  constructor(
    private readonly logger: PinoLogger
  ) {
    // Optionally you can set context for logger in constructor or ...
    this.logger.setContext(MyService.name);
  }

  constructor(
    // ... set context via special decorator
    @InjectPinoLogger(MyService.name)
    private readonly logger: PinoLogger
  ) {}

  foo() {
    // PinoLogger has same methods as pino instance
    this.logger.trace({ foo: 'bar' }, 'baz %s', 'qux');
    this.logger.debug('foo %s %o', 'bar', { baz: 'qux' });
    this.logger.info('foo');
  }
}

Register LoggerModule only via forRoot(...) / forRootAsync(...), and only once, in the root module. Never add the bare LoggerModule class to a feature module's imports, not even just to inject PinoLogger. Because LoggerModule is @Global(), both Logger and PinoLogger are already available everywhere after the single root registration, so you never need to re-import it. A bare import instantiates the module a second time, which registers the pino-http middleware again and makes every request log twice. The failure is completely silent: no compile error, no injection failure, no warning (#3074).

Drop-in replacement: NativeLogger

NativeLogger is a drop-in replacement for NestJS's built-in ConsoleLogger. It produces identical JSON output — same field names, same argument handling, same error format — but powered by pino with request context in every log.

If you're already using ConsoleLogger with { json: true } and want to switch to pino without changing any of your logging code, this is for you:

import { NativeLogger, nativeLoggerOptions } from 'nestjs-pino';

@Module({
  imports: [LoggerModule.forRoot({ pinoHttp: nativeLoggerOptions })],
})
class AppModule {}

const app = await NestFactory.create(AppModule, { bufferLogs: true });
app.useLogger(app.get(NativeLogger));

That's it. Your existing new Logger(MyService.name) calls throughout the codebase will work exactly as before — same message, context, level, timestamp, pid, and stack fields — but now with pino's performance and automatic request context binding.

ConsoleLogger JSON output:

{"level":"log","pid":17580,"timestamp":1765305000999,"message":"Hello World","context":"AppService"}

NativeLogger + nativeLoggerOptions output:

{"level":"log","pid":17580,"timestamp":1765305000999,"message":"Hello World","context":"AppService"}

How it differs from Logger

• Logger (pino-native): treats extra arguments as pino interpolation values. this.logger.log('foo %s', 'bar') → {"msg":"foo bar"}
• NativeLogger (NestJS-native): parses arguments the way ConsoleLogger does. this.logger.log('foo', 'bar', 'Ctx') → two logs, {"message":"foo","context":"Ctx"} and {"message":"bar","context":"Ctx"}

What matches ConsoleLogger exactly

• Argument parsing: last string = context, rest = separate log entries
• Structured params: on NestJS 12, plain objects after the message are merged into a single params field on one entry (ConsoleLoggerOptions.structuredParams, on by default) — this.logger.log('foo', { a: 1 }, { b: 2 }) → {"message":"foo","params":{"a":1,"b":2}}. On NestJS 11 each of them is a separate entry. NativeLogger follows the ConsoleLogger of the NestJS version you actually have, so out of the box there is nothing to configure — see below to override it
• Error handling: this.logger.error('msg', stackTrace, 'Ctx') → {"message":"msg","stack":"Error: ...","context":"Ctx"}
• Error objects: this.logger.log(new Error('oops')) → full error+stack as message string
• Exception handler: thrown errors logged with full stack in message field
• Object messages: this.logger.log({ foo: 'bar' }) → {"message":{"foo":"bar"}}
• Field names (with nativeLoggerOptions): message, timestamp, pid, level, context, stack

Keeping your ConsoleLogger options

If your application configures ConsoleLogger rather than relying on its defaults, pass the same values to keep the output identical after the switch:

LoggerModule.forRoot({
  pinoHttp: nativeLoggerOptions,
  nativeLogger: {
    // NestJS 12 default is `true`, NestJS 11 has no such option and behaves
    // as `false`. Omit it to follow the ConsoleLogger you actually have.
    structuredParams: true,
    // Spread params into the root of the record instead of nesting them
    // under `params`. NestJS default is `false`.
    flattenParams: true,
  },
});

{"level":"log","message":"foo","context":"AppService","a":1,"b":2}

Unlike NestJS, both options are honoured on every supported NestJS version — the collecting is implemented by this library, and the installed ConsoleLogger only decides the default of structuredParams. That also lets you switch to the NestJS 12 output while still on NestJS 11, so upgrading NestJS later does not change your logs.

A flattened param named context or stack is used as such when the call does not set them itself, and loses to the explicit form when it does — so log('msg', { context: 'A' }) logs the context A, while log('msg', { context: 'A' }, 'B') logs B.

Collisions with the fields pino adds — the level, the timestamp, the message key, the base bindings — are not resolved at all: pino does not deduplicate keys either, and their names depend on your pino options, so naming a param level is your call to make.

Output:

// Logs by app itself
{"level":30,"time":1629823318326,"pid":14727,"hostname":"my-host","context":"NestFactory","msg":"Starting Nest application..."}
{"level":30,"time":1629823318326,"pid":14727,"hostname":"my-host","context":"InstanceLoader","msg":"LoggerModule dependencies initialized"}
{"level":30,"time":1629823318327,"pid":14727,"hostname":"my-host","context":"InstanceLoader","msg":"AppModule dependencies initialized"}
{"level":30,"time":1629823318327,"pid":14727,"hostname":"my-host","context":"RoutesResolver","msg":"AppController {/}:"}
{"level":30,"time":1629823318327,"pid":14727,"hostname":"my-host","context":"RouterExplorer","msg":"Mapped {/, GET} route"}
{"level":30,"time":1629823318327,"pid":14727,"hostname":"my-host","context":"NestApplication","msg":"Nest application successfully started"}

// Logs by injected Logger and PinoLogger in Services/Controllers. Every log
// has it's request data and unique `req.id` (by default id is unique per
// process, but you can set function to generate it from request context and
// for example pass here incoming `X-Request-ID` header or generate UUID)
{"level":10,"time":1629823792023,"pid":15067,"hostname":"my-host","req":{"id":1,"method":"GET","url":"/","query":{},"params":{"0":""},"headers":{"host":"localhost:3000","user-agent":"curl/7.64.1","accept":"*/*"},"remoteAddress":"::1","remotePort":63822},"context":"MyService","foo":"bar","msg":"baz qux"}
{"level":20,"time":1629823792023,"pid":15067,"hostname":"my-host","req":{"id":1,"method":"GET","url":"/","query":{},"params":{"0":""},"headers":{"host":"localhost:3000","user-agent":"curl/7.64.1","accept":"*/*"},"remoteAddress":"::1","remotePort":63822},"context":"MyService","msg":"foo bar {\"baz\":\"qux\"}"}
{"level":30,"time":1629823792023,"pid":15067,"hostname":"my-host","req":{"id":1,"method":"GET","url":"/","query":{},"params":{"0":""},"headers":{"host":"localhost:3000","user-agent":"curl/7.64.1","accept":"*/*"},"remoteAddress":"::1","remotePort":63822},"context":"MyService","msg":"foo"}

// Automatic logs of every request/response
{"level":30,"time":1629823792029,"pid":15067,"hostname":"my-host","req":{"id":1,"method":"GET","url":"/","query":{},"params":{"0":""},"headers":{"host":"localhost:3000","user-agent":"curl/7.64.1","accept":"*/*"},"remoteAddress":"::1","remotePort":63822},"res":{"statusCode":200,"headers":{"x-powered-by":"Express","content-type":"text/html; charset=utf-8","content-length":"12","etag":"W/\"c-Lve95gjOVATpfV8EL5X4nxwjKHE\""}},"responseTime":7,"msg":"request completed"}

Comparison with others

There are other Nestjs loggers. Key purposes of this module are:

• to be idiomatic NestJS logger
• to log in JSON format (thanks to pino - super fast logger) why you should use JSON
• to log every request/response automatically (thanks to pino-http)
• to bind request data to the logs automatically from any service on any application layer without passing request context (thanks to AsyncLocalStorage)
• to have another alternative logger with same API as pino instance (PinoLogger) for experienced pino users to make more comfortable usage.

Logger	Nest App logger	Logger service	Auto-bind request data to logs
nest-winston	+	+	-
nestjs-pino-logger	+	+	-
nestjs-pino	+	+	+

Configuration

Zero configuration

Just import LoggerModule to your module:

import { LoggerModule } from 'nestjs-pino';

@Module({
  imports: [LoggerModule.forRoot()],
  ...
})
class MyModule {}

Configuration params

The following interface is using for the configuration:

interface Params<
  IM = IncomingMessage,
  SR = ServerResponse,
  CustomLevels extends string = never,
> {
  /**
   * Optional parameters for `pino-http` module
   * @see https://github.com/pinojs/pino-http#api
   */
  pinoHttp?:
    | pinoHttp.Options<IM, SR, CustomLevels>
    | DestinationStream
    | [pinoHttp.Options<IM, SR, CustomLevels>, DestinationStream];

  /**
   * Optional parameter for routing. It should implement interface of
   * parameters of NestJS built-in `MiddlewareConfigProxy['forRoutes']`.
   * @see https://docs.nestjs.com/middleware#applying-middleware
   * It can be used for both disabling automatic req/res logs (see above) and
   * removing request context from following logs. It works for all requests by
   * default. If you only need to turn off the automatic request/response
   * logging for some specific (or all) routes but keep request context for app
   * logs use `pinoHttp.autoLogging` field.
   */
  forRoutes?: Parameters<MiddlewareConfigProxy['forRoutes']>;

  /**
   * Optional parameter for routing. It should implement interface of
   * parameters of NestJS built-in `MiddlewareConfigProxy['exclude']`.
   * @see https://docs.nestjs.com/middleware#applying-middleware
   * It can be used for both disabling automatic req/res logs (see above) and
   * removing request context from following logs. It works for all requests by
   * default. If you only need to turn off the automatic request/response
   * logging for some specific (or all) routes but keep request context for app
   * logs use `pinoHttp.autoLogging` field.
   */
  exclude?: Parameters<MiddlewareConfigProxy['exclude']>;

  /**
   * Optional parameter to skip pino configuration in case you are using
   * FastifyAdapter, and already configure logger in adapter's config. The Pros
   * and cons of this approach are described in the FAQ section of the
   * documentation:
   * @see https://github.com/iamolegga/nestjs-pino#faq.
   */
  useExisting?: true;

  /**
   * Optional parameter to change property name `context` in resulted logs,
   * so logs will be like:
   * {"level":30, ... "RENAME_CONTEXT_VALUE_HERE":"AppController" }
   */
  renameContext?: string;

  /**
   * Optional parameter to also assign the response logger during calls to
   * `PinoLogger.assign`. By default, `assign` does not impact response logs
   * (e.g.`Request completed`).
   */
  assignResponse?: boolean;

  /**
   * Optional parameters for automatic logging of microservice messages, and
   * for a logging context in message handlers. See the "Microservices"
   * section below. Pass `true` for the defaults.
   */
  microservice?: boolean | MicroserviceParams<CustomLevels>;
}

Typing the request, response and custom levels

All three type parameters are optional and default to what pino-http itself defaults to, so Params keeps working unparameterised. Pass them when you want your platform's request/response types inside the pinoHttp callbacks:

import type { Request, Response } from 'express';
import { LoggerModule, Params } from 'nestjs-pino';

const params: Params<Request, Response> = {
  pinoHttp: {
    // `req` is an express Request here, not a bare IncomingMessage
    genReqId: (req) => req.headers['x-correlation-id'] ?? randomUUID(),
    serializers: { req: (req: Request) => ({ id: req.id, url: req.url }) },
  },
};

LoggerModule.forRoot(params);

The third parameter carries pino's customLevels:

type CustomLevels = 'audit';

const params: Params<Request, Response, CustomLevels> = {
  pinoHttp: {
    customLevels: { audit: 35 },
    useLevel: 'audit',
  },
};

Custom levels are reachable through the underlying pino instance — pinoLogger.logger.audit('...'), not pinoLogger.audit('...'). PinoLogger is instantiated by the NestJS DI container, which cannot infer a type argument, so the level methods cannot be synthesised onto the class itself.

Synchronous configuration

Use LoggerModule.forRoot method with argument of Params interface:

import { LoggerModule } from 'nestjs-pino';

@Module({
  imports: [
    LoggerModule.forRoot({
      pinoHttp: [
        {
          name: 'add some name to every JSON line',
          level: process.env.NODE_ENV !== 'production' ? 'debug' : 'info',
          // install 'pino-pretty' package in order to use the following option
          transport: process.env.NODE_ENV !== 'production'
            ? { target: 'pino-pretty' }
            : undefined,


          // and all the other fields of:
          // - https://github.com/pinojs/pino-http#api
          // - https://github.com/pinojs/pino/blob/HEAD/docs/api.md#options-object


        },
        someWritableStream
      ],
      forRoutes: [MyController],
      exclude: [{ method: RequestMethod.ALL, path: 'check' }]
    })
  ],
  ...
})
class MyModule {}

Asynchronous configuration

With LoggerModule.forRootAsync you can, for example, import your ConfigModule and inject ConfigService to use it in useFactory method.

useFactory should return object with Params interface or undefined

Here's an example:

import { LoggerModule } from 'nestjs-pino';

@Injectable()
class ConfigService {
  public readonly level = 'debug';
}

@Module({
  providers: [ConfigService],
  exports: [ConfigService]
})
class ConfigModule {}

@Module({
  imports: [
    LoggerModule.forRootAsync({
      imports: [ConfigModule],
      inject: [ConfigService],
      useFactory: async (config: ConfigService) => {
        await somePromise();
        return {
          pinoHttp: { level: config.level },
        };
      }
    })
  ],
  ...
})
class TestModule {}

Asynchronous logging

In essence, asynchronous logging enables even faster performance by pino.

Please, read pino asynchronous mode docs first. There is a possibility of the most recently buffered log messages being lost in case of a system failure, e.g. a power cut.

If you know what you're doing, you can enable it like so:

import pino from 'pino';
import { LoggerModule } from 'nestjs-pino';

@Module({
  imports: [
    LoggerModule.forRoot({
      pinoHttp: {
        stream: pino.destination({
          dest: './my-file', // omit for stdout
          minLength: 4096, // Buffer before writing
          sync: false, // Asynchronous logging
        }),
      },
    }),
  ],
  ...
})
class MyModule {}

See pino.destination

Testing a class that uses @InjectPinoLogger

This package exposes a getLoggerToken() function that returns a prepared injection token based on the provided context. Using this token, you can provide a mock implementation of the logger using any of the standard custom provider techniques, including useClass, useValue and useFactory.

  const module: TestingModule = await Test.createTestingModule({
    providers: [
      MyService,
      {
        provide: getLoggerToken(MyService.name),
        useValue: mockLogger,
      },
    ],
  }).compile();

Logger/PinoLogger class extension

Logger and PinoLogger classes can be extended.

// logger.service.ts
import { Logger, PinoLogger, Params, PARAMS_PROVIDER_TOKEN } from 'nestjs-pino';

@Injectable()
class LoggerService extends Logger {
  constructor(
    logger: PinoLogger,
    @Inject(PARAMS_PROVIDER_TOKEN) params: Params
  ) {
    ...
  }
  // extended method
  myMethod(): any {}
}

import { PinoLogger, Params, PARAMS_PROVIDER_TOKEN } from 'nestjs-pino';

@Injectable()
class LoggerService extends PinoLogger {
  constructor(
    @Inject(PARAMS_PROVIDER_TOKEN) params: Params
  ) {
    // ...
  }
  // extended method
  myMethod(): any {}
}


// logger.module.ts
@Module({
  providers: [LoggerService],
  exports: [LoggerService],
  imports: [LoggerModule.forRoot()],
})
class LoggerModule {}

Notes on Logger injection in constructor

Since logger substitution has appeared in NestJS@8 the main purpose of Logger class is to be registered via app.useLogger(app.get(Logger)). But that requires some internal breaking change, because with such usage NestJS pass logger's context as the last optional argument in logging function. So in current version Logger's methods accept context as a last argument.

With such change it's not possible to detect if method was called by app internaly and the last argument is context or Logger was injected in some service via constructor(private logger: Logger) {} and the last argument is interpolation value for example.

Assign extra fields for future calls

You can enrich logs before calling log methods. It's possible by using assign method of PinoLogger instance. As Logger class is used only for NestJS built-in Logger substitution via app.useLogger(...) this feature is only limited to PinoLogger class. Example:

@Controller('/')
class TestController {
  constructor(
    private readonly logger: PinoLogger,
    private readonly service: MyService,
  ) {}

  @Get()
  get() {
    // assign extra fields in one place...
    this.logger.assign({ userID: '42' });
    return this.service.test();
  }
}

@Injectable()
class MyService {
  private readonly logger = new Logger(MyService.name);

  test() {
    // ...and it will be logged in another one
    this.logger.log('hello world');
  }
}

By default, this does not extend Request completed logs. Set the assignResponse parameter to true to also enrich response logs automatically emitted by pino-http.

Microservices

Message handlers get the same treatment HTTP requests do: every log made while handling a message carries the message's context, assign works, and a line per message is logged automatically. Add the microservice parameter:

LoggerModule.forRoot({
  pinoHttp: { level: 'debug' },  // as before: level, transport, redact, ...
  microservice: true,            // that's all
});

Nothing changes in main.ts — the module registers a pre-request hook by itself:

const app = await NestFactory.createMicroservice(AppModule, {
  transport: Transport.RMQ,
  options: { urls: ['amqp://localhost'], queue: 'users' },
  bufferLogs: true,
});
app.useLogger(app.get(Logger));
await app.listen();

@Controller()
export class UserController {
  constructor(private readonly logger: PinoLogger) {}

  @EventPattern('user.created')
  async handle(@Payload() data: UserCreated) {
    this.logger.assign({ userId: data.id });
    this.logger.info('provisioning');
    await this.provision(data);
  }
}

{"level":30,"reqId":1,"rpc":{"type":"event","pattern":"user.created","transport":"rmq","controller":"UserController","handler":"handle"},"userId":"u_42","msg":"provisioning"}
{"level":30,"reqId":1,"rpc":{"type":"event","pattern":"user.created","transport":"rmq","controller":"UserController","handler":"handle"},"userId":"u_42","responseTime":37,"msg":"event completed"}

The automatic messages are message completed/message errored for a @MessagePattern, where the caller is waiting for a reply, and event completed/event errored for an @EventPattern, where nobody is. As in pino-http, nothing is logged on arrival unless customReceivedMessage or customReceivedObject says so. Unlike pino-http, a failed message defaults to error rather than to useLevel, so that lowering useLevel to quiet down events does not also hide failures. Note that the hook sees the error before any @Catch() filter, so an RpcException thrown on purpose and mapped to a client response arrives here too — reach for customLogLevel if that should not be an error. No counterpart of LoggerErrorInterceptor is needed: the hook receives the error directly.

Requirements. Pre-request hooks arrived in NestJS 12, and 12.0.2 is the first release on which a hook leaves a handler's plain return value alone (nestjs/nest#17644) — hence the lower bound of the peer range. On NestJS 11 the parameter is ignored with a warning and everything else keeps working.

Hybrid applications

A hybrid application must be connected with inheritAppConfig, because without it NestJS gives the microservice an ApplicationConfig of its own that dependency injection cannot reach:

app.connectMicroservice(
  { transport: Transport.RMQ, options: { urls: ['amqp://localhost'], queue: 'users' } },
  { inheritAppConfig: true },
);

If that does not suit you, wire it up by hand instead:

import { registerMicroserviceLogging } from 'nestjs-pino';

const ms = app.connectMicroservice(options, { deferInitialization: true });
registerMicroserviceLogging(ms);

Microservice configuration params

Shaped after pino-http's: every parameter that does not need the request or the response keeps its name and meaning, and the ones that do take an ExecutionContext instead. Everything pino itself is configured with — level, transport, redact, serializers — stays in pinoHttp, which builds the one logger both halves of the application share.

interface MicroserviceParams<CustomLevels extends string = never> {
  /**
   * Set to `false` to stop logging a line per message. The logging *context*
   * is still established either way, so `PinoLogger.assign` and the inherited
   * fields keep working.
   * @default true
   */
  autoLogging?:
    | boolean
    | {
        /** Skip the automatic logs for the messages this returns `true` for. */
        ignore?: (context: ExecutionContext) => boolean;
      };

  /**
   * Level of the `received` and `completed` logs. Errors default to `error`
   * regardless of this — unlike `pino-http`, where a failed request is logged
   * at `useLevel` too — so that lowering this to quiet down events does not
   * also hide failures. Use `customLogLevel` to change the error level.
   * Cannot be combined with `customLogLevel`.
   * @default 'info'
   */
  useLevel?: LevelWithSilent | CustomLevels;

  /**
   * Decides the level of every automatic log, including the error one. Note
   * that the hook runs before any `@Catch()` filter, so `error` is set for
   * exceptions the application goes on to handle itself, an `RpcException`
   * thrown on purpose included.
   */
  customLogLevel?: (
    context: ExecutionContext,
    error?: Error,
  ) => LevelWithSilent | CustomLevels;

  /**
   * Generates the value bound as `reqId`. Defaults to an incrementing counter,
   * as in `pino-http`; a correlation id off the transport context is usually a
   * better choice.
   */
  genReqId?: (context: ExecutionContext) => ReqId;

  /**
   * Setting this — or `customReceivedObject` — is what enables the log on
   * arrival. There is no default text, so nothing is logged on arrival unless
   * asked for, exactly as in `pino-http`.
   */
  customReceivedMessage?: (context: ExecutionContext) => string;

  /** @default `'message completed'`, or `'event completed'` for an `@EventPattern` */
  customSuccessMessage?: (
    context: ExecutionContext,
    result: unknown,
    responseTime: number,
  ) => string;

  /** @default `'message errored'`, or `'event errored'` for an `@EventPattern` */
  customErrorMessage?: (
    context: ExecutionContext,
    error: Error,
    responseTime: number,
  ) => string;

  /**
   * Fields of the log on arrival, which it enables the same way
   * `customReceivedMessage` does. There is no default object, so what it
   * returns is logged as is.
   */
  customReceivedObject?: (context: ExecutionContext) => object;

  /**
   * Replaces the fields of the `completed` log. `value` is what would be logged
   * otherwise — `responseTime` under its configured key — and is not merged
   * back in, so spread it if the extra fields should come on top.
   */
  customSuccessObject?: (
    context: ExecutionContext,
    result: unknown,
    value: object,
  ) => object;

  /**
   * Replaces the fields of the `errored` log. `value` holds `err` and
   * `responseTime` under their configured keys and, as with
   * `customSuccessObject`, is not merged back in.
   */
  customErrorObject?: (
    context: ExecutionContext,
    error: Error,
    value: object,
  ) => object;

  /** Extra fields bound to every log made while handling the message. */
  customProps?: (context: ExecutionContext) => object;

  /**
   * Renames the keys this adds to the log record. The `rpc` key is the
   * microservice counterpart of `pino-http`'s `req`/`res`.
   */
  customAttributeKeys?: {
    /** @default 'rpc' */
    rpc?: string;
    /** @default 'err' */
    err?: string;
    /** @default 'reqId' */
    reqId?: string;
    /** @default 'responseTime' */
    responseTime?: string;
  };

  /**
   * Bind only `reqId`, leaving the `rpc` object out of the logs made while
   * handling the message. Mirrors `pino-http`'s `quietReqLogger`.
   */
  quietRpcLogger?: boolean;

  /**
   * Leave the `rpc` object out of the `completed`/`errored` log, which repeats
   * what the surrounding logs already carry. Mirrors `quietResLogger`.
   */
  quietResLogger?: boolean;

  /**
   * Log the message payload as `rpc.payload`. Off by default: payloads are
   * unbounded in size and routinely carry personal data. `redact` and a
   * `serializers.rpc` entry in `pinoHttp` apply to it as to anything else.
   * @default false
   */
  includePayload?: boolean;
}

Use getRpcInfo(ctx) inside any of these to reach the pattern, the transport and the kind of handler. It returns what is logged under rpc, and never throws:

import { getRpcInfo } from 'nestjs-pino';

LoggerModule.forRoot({
  microservice: {
    // events are noisy, replies are not
    customLogLevel: (ctx, err) =>
      err ? 'error' : getRpcInfo(ctx).type === 'event' ? 'debug' : 'info',

    autoLogging: { ignore: (ctx) => getRpcInfo(ctx).handler === 'healthcheck' },

    genReqId: (ctx) =>
      ctx.switchToRpc().getContext<RmqContext>().getMessage()
         .properties.correlationId ?? randomUUID(),
  },
});

Logging context anywhere else

Queue processors, cron jobs, CLI commands, standalone scripts, tests and consumers wired up outside of NestJS have no request pipeline to hook into. Use runInContext to open a logging context by hand, and assign and inherited fields work there too:

@Processor('emails')
export class EmailProcessor extends WorkerHost {
  constructor(private readonly logger: PinoLogger) { super(); }

  async process(job: Job) {
    return this.logger.runInContext(
      async () => {
        this.logger.assign({ jobId: job.id, attempt: job.attemptsMade });
        this.logger.info('sending');
        await this.send(job.data);
      },
      { bindings: { queue: 'emails' } },
    );
  }
}

It returns whatever the function returns, and an async function keeps the context across every await inside it.

Pass inherit: true to start from the surrounding context instead of from the root logger — useful for @OnEvent handlers and for work started during a request but finished after it. The store is a new one either way, so assign inside never reaches the log that closes the surrounding request:

@OnEvent('order.paid')
async onPaid(e: OrderPaid) {
  await this.logger.runInContext(
    async () => {
      this.logger.assign({ orderId: e.id });
      await this.fulfil(e);
    },
    { inherit: true },
  );
}

Change pino params at runtime

Pino root instance with passed via module registration params creates a separate child logger for every request. This root logger params can be changed at runtime via PinoLogger.root property which is the pointer to logger instance. Every log in the application descends from it — request logs, microservice message logs and logs made outside of any context alike — so a change reaches all of them. It is available in a microservice or standalone application too, where there is no HTTP middleware. Example:

@Controller('/')
class TestController {
  @Post('/change-loggin-level')
  setLevel() {
    PinoLogger.root.level = 'info';
    return null;
  }
}

Expose stack trace and error class in err property

By default, pino-http exposes err property with a stack trace and error details, however, this err property contains default error details, which do not tell anything about actual error. To expose actual error details you need you to use a NestJS interceptor which captures exceptions and assigns them to the response object err property which is later processed by pino-http:

import { LoggerErrorInterceptor } from 'nestjs-pino';

const app = await NestFactory.create(AppModule);
app.useGlobalInterceptors(new LoggerErrorInterceptor());

Migration

v5

• Requirements changed. NestJS 11.0.8+ or 12.0.2+ (11.0.8 is where NestJS started preserving the {/...} route syntax the default middleware route relies on; 12.0.2 is where a microservice pre-request hook stopped corrupting a handler's plain return value), pino@10, pino-http@11, Node.js >=22.12. Support for NestJS 8-10, pino 7-9 and pino-http 6-10 is dropped. @nestjs/core is now a peer dependency alongside @nestjs/common; every NestJS application already has it installed.
• The package now ships from dist/ behind an exports map. The public entry point is unchanged, but deep imports such as nestjs-pino/PinoLogger no longer resolve — import from nestjs-pino instead.

Everything else is backwards compatible. In particular the warning NestJS used to print on startup when a global prefix was set (Unsupported route path: "/v1/*") is gone, and requests hitting the prefix root itself are now logged as well.

Paths excluded from the global prefix with app.setGlobalPrefix(prefix, { exclude }) are logged too. On Fastify this needs @nestjs/platform-fastify@11.2.0 or newer: until then the adapter prepended the global prefix to every middleware path that did not already start with it, which put excluded paths out of reach. Express has no such requirement.

v1

• All parameters of v.0 are moved to pinoHttp property (except useExisting).
• useExisting now accept only true because you should already know if you want to use preconfigured fastify adapter's logger (and set true) or not (and just not define this field).

v2

Logger substitution

A new more convenient way to inject a custom logger that implements LoggerService has appeared in recent versions of NestJS (mind the bufferLogs field, it will force NestJS to wait for logger to be ready instead of using built-in logger on start):

// main.ts
import { Logger } from 'nestjs-pino';
// ...
  const app = await NestFactory.create(AppModule, { bufferLogs: true });
  app.useLogger(app.get(Logger));
// ...

Note that for standalone applications, buffering has to be flushed using app.flushLogs() manually after custom logger is ready to be used by NestJS (refer to this issue for more details):

// main.ts
import { Logger } from 'nestjs-pino';

// ... 
  const app = await NestFactory.createApplicationContext(AppModule, { bufferLogs: true });
  app.useLogger(app.get(Logger));
  app.flushLogs();
// ...

In all the other places you can use built-in Logger:

// my-service.ts
import { Logger } from '@nestjs/common';
class MyService {
  private readonly logger = new Logger(MyService.name);
}

To quote the official docs:

If we supply a custom logger via app.useLogger(), it will actually be used by Nest internally. That means that our code remains implementation agnostic, while we can easily substitute the default logger for our custom one by calling app.useLogger().

That way if we follow the steps from the previous section and call app.useLogger(app.get(MyLogger)), the following calls to this.logger.log() from MyService would result in calls to method log from MyLogger instance.

---

This is recommended to update all your existing Logger injections from nestjs-pino to @nestjs/common. And inject it only in your main.ts file as shown above. Support of injection of Logger (don't confuse with PinoLogger) from nestjs-pino directly in class constructors is dropped.

---

Since logger substitution has appeared the main purpose of Logger class is to be registered via app.useLogger(app.get(Logger)). But that requires some internal breaking change, because with such usage NestJS pass logger's context as the last optional argument in logging function. So in current version Logger's methods accept context as the last argument.

With such change it's not possible to detect if method was called by app internaly and the last argument is context or Logger was injected in some service via constructor(private logger: Logger) {} and the last argument is interpolation value for example. That's why logging with such injected class still works, but only for 1 argument.

NestJS LoggerService interface breaking change

In NestJS@8 all logging methods of built-in LoggerService now accept the same arguments without second context argument (which is set via injection, see above), for example: log(message: any, ...optionalParams: any[]): any;. That makes usage of built-in logger more convenient and compatible with pino's logging methods. So this is a breaking change in NestJS, and you should be aware of it.

In NestJS <= 7 and nestjs-pino@1 when you call this.logger.log('foo', 'bar'); there would be such log: {..."context":"bar","msg":"foo"} (second argument goes to context field by desing). In NestJS 8 and nestjs-pino@2 (with proper injection that shown above) same call will result in {..."context":"MyService","msg":"foo"}, so context is passed via injection, but second argument disappear from log, because now it treats as interpolation value and there should be placeholder for it in message argument. So if you want to get both foo and bar in log the right way to do this is: this.logger.log('foo %s', 'bar');. More info can be found in pino docs.

FAQ

Q: How to disable automatic request/response logs?

A: check out autoLogging field of pino-http that are set in pinoHttp field of Params

---

Q: How to pass X-Request-ID header or generate UUID for req.id field of log?

A: check out genReqId field of pino-http that are set in pinoHttp field of Params

---

Q: How does it work?

A: It uses pino-http under hood, so every request has it's own child-logger, and with help of AsyncLocalStorage Logger and PinoLogger can get it while calling own methods. So your logs can be grouped by req.id.

---

Q: Why use AsyncLocalStorage instead of REQUEST scope?

A: REQUEST scope can have perfomance issues. TL;DR: it will have to create an instance of the class (that injects Logger) on each request, and that will slow down your response times.

---

Q: I'm using old nodejs version, will it work for me?

A: Please check out history of this feature.

---

Q: What about pino built-in methods/levels?

A: Pino built-in methods names are not fully compatible with NestJS built-in LoggerService methods names, and there is an option which logger you use. Here is methods mapping:

pino method	PinoLogger method	NestJS built-in Logger method
trace	trace	verbose
debug	debug	debug
info	info	log
warn	warn	warn
error	error	error
fatal	fatal	fatal (since nestjs@10.2)

---

Q: Fastify already includes pino, and I want to configure it on Adapter level, and use this config for logger

A: You can do it by providing useExisting: true. But there is one caveat:

Fastify creates logger with your config per every request. And this logger is used by Logger/PinoLogger services inside that context underhood.

But Nest Application has another contexts of execution, for example lifecycle events, where you still may want to use logger. For that Logger/PinoLogger services use separate pino instance with config, that provided via forRoot/forRootAsync methods.

So, when you want to configure pino via FastifyAdapter there is no way to get back this config from fastify and pass it to that out of context logger.

And if you will not pass config via forRoot/forRootAsync out of context logger will be instantiated with default params. So if you want to configure it with the same options for consistency you have to provide the same config to LoggerModule configuration too. But if you already provide it to LoggerModule configuration you can drop useExisting field from config and drop logger configuration on FastifyAdapter, and it will work without code duplication.

So this property (useExisting: true) is not recommended, and can be useful only for cases when:

• this logger is not using for lifecycle events and application level logging in NestJS apps based on fastify
• pino is using with default params in NestJS apps based on fastify

All the other cases are lead to either code duplication or unexpected behavior.

---

Do you use this library?
Don't be shy to give it a star! ★

Also if you are into NestJS you might be interested in one of my other NestJS libs.