Bunyan is a simple and fast JSON logging library for node.js services:

1var bunyan = require('bunyan');
2var log = bunyan.createLogger({name: "myapp"});
3log.info("hi");

and a bunyan CLI tool for nicely viewing those logs:

Manifesto: Server logs should be structured. JSON's a good format. Let's do that. A log record is one line of JSON.stringify'd output. Let's also specify some common names for the requisite and common fields for a log record (see below).

Table of Contents

• Current Status
• Installation
• Features
• Introduction
  • Constructor API
  • Log Method API
  • CLI Usage
  • Streams Introduction
  • log.child
  • Serializers
    • Requirements for serializers functions
    • Standard Serializers
  • src
• Levels
  • Level suggestions
• Log Record Fields
  • Core fields
  • Recommended/Best Practice Fields
  • Other fields to consider
• Streams
  • Adding a Stream
  • stream errors
  • stream type: stream
  • stream type: file
  • stream type: rotating-file
  • stream type: raw
  • raw + RingBuffer Stream
  • third-party streams
• Runtime log snooping via DTrace
  • DTrace examples
• Runtime environments
  • Browserify
  • Webpack
• Versioning
• License
• See Also

Current Status

Stable. I do my best to follow semver: i.e. you should only need to worry about code breaking for a major version bump. Bunyan currently supports node 0.10 and greater. Follow @trentmick for updates to Bunyan.

There is an email discussion list bunyan-logging@googlegroups.com, also as a forum in the browser.

Active branches:

• "1.x" is for 1.x maintenance work, if any. 1.x releases are still "latest" in npm.
• "master" is currently for coming Bunyan 2.x work. For now, 2.x releases are published to npm with the "beta" tag, meaning that npm install bunyan is still 1.x for now. To install 2.x use npm install bunyan@2 or npm install bunyan@beta.

Installation

1npm install bunyan

Tip: The bunyan CLI tool is written to be compatible (within reason) with all versions of Bunyan logs. Therefore you might want to npm install -g bunyan to get the bunyan CLI on your PATH, then use local bunyan installs for node.js library usage of bunyan in your apps.

Tip: Installing without optional dependencies can dramatically reduce bunyan's install size. dtrace-provider is used for dtrace features, mv is used for RotatingFileStream, and moment is used for local time. If you don't need these features, consider installing with the --no-optional flag.

Features

• elegant log method API
• extensible streams system for controlling where log records go (to a stream, to a file, log file rotation, etc.)
• bunyan CLI for pretty-printing and filtering of Bunyan logs
• simple include of log call source location (file, line, function) with src: true
• lightweight specialization of Logger instances with log.child
• custom rendering of logged objects with "serializers"
• Runtime log snooping via DTrace support
• Support for a few runtime environments: Node.js, Browserify, Webpack, NW.js.

Introduction

Like most logging libraries you create a Logger instance and call methods named after the logging levels:

1// hi.js
2var bunyan = require('bunyan');
3var log = bunyan.createLogger({name: 'myapp'});
4log.info('hi');
5log.warn({lang: 'fr'}, 'au revoir');

All loggers must provide a "name". This is somewhat akin to the log4j logger "name", but Bunyan doesn't do hierarchical logger names.

Bunyan log records are JSON. A few fields are added automatically: "pid", "hostname", "time" and "v".

1$ node hi.js
2{"name":"myapp","hostname":"banana.local","pid":40161,"level":30,"msg":"hi","time":"2013-01-04T18:46:23.851Z","v":0}
3{"name":"myapp","hostname":"banana.local","pid":40161,"level":40,"lang":"fr","msg":"au revoir","time":"2013-01-04T18:46:23.853Z","v":0}

Constructor API

1var bunyan = require('bunyan');
2var log = bunyan.createLogger({
3    name: <string>,                     // Required
4    level: <level name or number>,      // Optional, see "Levels" section
5    stream: <node.js stream>,           // Optional, see "Streams" section
6    streams: [<bunyan streams>, ...],   // Optional, see "Streams" section
7    serializers: <serializers mapping>, // Optional, see "Serializers" section
8    src: <boolean>,                     // Optional, see "src" section
9
10    // Any other fields are added to all log records as is.
11    foo: 'bar',
12    ...
13});

Log Method API

The example above shows two different ways to call log.info(...). The full API is:

1log.info();     // Returns a boolean: is the "info" level enabled?
2                // This is equivalent to `log.isInfoEnabled()` or
3                // `log.isEnabledFor(INFO)` in log4j.
4
5log.info('hi');                     // Log a simple string message (or number).
6log.info('hi %s', bob, anotherVar); // Uses `util.format` for msg formatting.
7
8log.info({foo: 'bar'}, 'hi');
9                // The first field can optionally be a "fields" object, which
10                // is merged into the log record.
11
12log.info(err);  // Special case to log an `Error` instance to the record.
13                // This adds an "err" field with exception details
14                // (including the stack) and sets "msg" to the exception
15                // message.
16log.info(err, 'more on this: %s', more);
17                // ... or you can specify the "msg".
18
19log.info({foo: 'bar', err: err}, 'some msg about this error');
20                // To pass in an Error *and* other fields, use the `err`
21                // field name for the Error instance **and ensure your logger
22                // has a `err` serializer.** One way to ensure the latter is:
23                //      var log = bunyan.createLogger({
24                //          ...,
25                //          serializers: bunyan.stdSerializers
26                //      });
27                // See the "Serializers" section below for details.

Note that this implies you cannot blindly pass any object as the first argument to log it because that object might include fields that collide with Bunyan's core record fields. In other words, log.info(mywidget) may not yield what you expect. Instead of a string representation of mywidget that other logging libraries may give you, Bunyan will try to JSON-ify your object. It is a Bunyan best practice to always give a field name to included objects, e.g.:

1log.info({widget: mywidget}, ...)

This will dove-tail with Bunyan serializer support, discussed later.

The same goes for all of Bunyan's log levels: log.trace, log.debug, log.info, log.warn, log.error, and log.fatal. See the levels section below for details and suggestions.

CLI Usage

Bunyan log output is a stream of JSON objects. This is great for processing, but not for reading directly. A bunyan tool is provided for pretty-printing bunyan logs and for filtering (e.g. | bunyan -c 'this.foo == "bar"'). Using our example above:

1$ node hi.js | ./node_modules/.bin/bunyan
2[2013-01-04T19:01:18.241Z]  INFO: myapp/40208 on banana.local: hi
3[2013-01-04T19:01:18.242Z]  WARN: myapp/40208 on banana.local: au revoir (lang=fr)

See the screenshot above for an example of the default coloring of rendered log output. That example also shows the nice formatting automatically done for some well-known log record fields (e.g. req is formatted like an HTTP request, res like an HTTP response, err like an error stack trace).

One interesting feature is filtering of log content, which can be useful for digging through large log files or for analysis. We can filter only records above a certain level:

1$ node hi.js | bunyan -l warn
2[2013-01-04T19:08:37.182Z]  WARN: myapp/40353 on banana.local: au revoir (lang=fr)

Or filter on the JSON fields in the records (e.g. only showing the French records in our contrived example):

1$ node hi.js | bunyan -c 'this.lang == "fr"'
2[2013-01-04T19:08:26.411Z]  WARN: myapp/40342 on banana.local: au revoir (lang=fr)

See bunyan --help for other facilities.

Streams Introduction

By default, log output is to stdout and at the "info" level. Explicitly that looks like:

1var log = bunyan.createLogger({
2    name: 'myapp',
3    stream: process.stdout,
4    level: 'info'
5});

That is an abbreviated form for a single stream. You can define multiple streams at different levels.

1var log = bunyan.createLogger({
2  name: 'myapp',
3  streams: [
4    {
5      level: 'info',
6      stream: process.stdout            // log INFO and above to stdout
7    },
8    {
9      level: 'error',
10      path: '/var/tmp/myapp-error.log'  // log ERROR and above to a file
11    }
12  ]
13});

More on streams in the Streams section below.

log.child

Bunyan has a concept of a child logger to specialize a logger for a sub-component of your application, i.e. to create a new logger with additional bound fields that will be included in its log records. A child logger is created with log.child(...).

In the following example, logging on a "Wuzzle" instance's this.log will be exactly as on the parent logger with the addition of the widget_type field:

1var bunyan = require('bunyan');
2var log = bunyan.createLogger({name: 'myapp'});
3
4function Wuzzle(options) {
5    this.log = options.log.child({widget_type: 'wuzzle'});
6    this.log.info('creating a wuzzle')
7}
8Wuzzle.prototype.woos = function () {
9    this.log.warn('This wuzzle is woosey.')
10}
11
12log.info('start');
13var wuzzle = new Wuzzle({log: log});
14wuzzle.woos();
15log.info('done');

Running that looks like (raw):

1$ node myapp.js
2{"name":"myapp","hostname":"myhost","pid":34572,"level":30,"msg":"start","time":"2013-01-04T07:47:25.814Z","v":0}
3{"name":"myapp","hostname":"myhost","pid":34572,"widget_type":"wuzzle","level":30,"msg":"creating a wuzzle","time":"2013-01-04T07:47:25.815Z","v":0}
4{"name":"myapp","hostname":"myhost","pid":34572,"widget_type":"wuzzle","level":40,"msg":"This wuzzle is woosey.","time":"2013-01-04T07:47:25.815Z","v":0}
5{"name":"myapp","hostname":"myhost","pid":34572,"level":30,"msg":"done","time":"2013-01-04T07:47:25.816Z","v":0}

And with the bunyan CLI (using the "short" output mode):

1$ node myapp.js  | bunyan -o short
207:46:42.707Z  INFO myapp: start
307:46:42.709Z  INFO myapp: creating a wuzzle (widget_type=wuzzle)
407:46:42.709Z  WARN myapp: This wuzzle is woosey. (widget_type=wuzzle)
507:46:42.709Z  INFO myapp: done

A more practical example is in the node-restify web framework. Restify uses Bunyan for its logging. One feature of its integration, is that if server.use(restify.requestLogger()) is used, each restify request handler includes a req.log logger that is:

1log.child({req_id: <unique request id>}, true)

Apps using restify can then use req.log and have all such log records include the unique request id (as "req_id"). Handy.

Serializers

Bunyan has a concept of "serializer" functions to produce a JSON-able object from a JavaScript object, so you can easily do the following:

1log.info({req: <request object>}, 'something about handling this request');

and have the req entry in the log record be just a reasonable subset of <request object> fields (or computed data about those fields).

A logger instance can have a serializers mapping of log record field name ("req" in this example) to a serializer function. When creating the log record, Bunyan will call the serializer function for top-level fields of that name. An example:

1function reqSerializer(req) {
2    return {
3        method: req.method,
4        url: req.url,
5        headers: req.headers
6    };
7}
8var log = bunyan.createLogger({
9    name: 'myapp',
10    serializers: {
11        req: reqSerializer
12    }
13});

Typically serializers are added to a logger at creation time via:

1var log = bunyan.createLogger({
2    name: 'myapp',
3    serializers: {
4        foo: function fooSerializer(foo) { ... },
5        ...
6    }
7});
8
9// or
10var log = bunyan.createLogger({
11    name: 'myapp',
12    serializers: bunyan.stdSerializers
13});

Serializers can also be added after creation via <logger>.addSerializers(...), e.g.:

1var log = bunyan.createLogger({name: 'myapp'});
2log.addSerializers({req: reqSerializer});

Requirements for serializers functions

A serializer function is passed unprotected objects that are passed to the log.info, log.debug, etc. call. This means a poorly written serializer function can cause side-effects. Logging shouldn't do that. Here are a few rules and best practices for serializer functions:

• A serializer function should never throw. The bunyan library does protect somewhat from this: if the serializer throws an error, then bunyan will (a) write an ugly message on stderr (along with the traceback), and (b) the field in the log record will be replaced with a short error message. For example:

  bunyan: ERROR: Exception thrown from the "foo" Bunyan serializer. This should never happen. This is a bug in that serializer function.
  TypeError: Cannot read property 'not' of undefined
      at Object.fooSerializer [as foo] (/Users/trentm/tm/node-bunyan/bar.js:8:26)
      at /Users/trentm/tm/node-bunyan/lib/bunyan.js:873:50
      at Array.forEach (native)
      at Logger._applySerializers (/Users/trentm/tm/node-bunyan/lib/bunyan.js:865:35)
      at mkRecord (/Users/trentm/tm/node-bunyan/lib/bunyan.js:978:17)
      at Logger.info (/Users/trentm/tm/node-bunyan/lib/bunyan.js:1044:19)
      at Object.<anonymous> (/Users/trentm/tm/node-bunyan/bar.js:13:5)
      at Module._compile (module.js:409:26)
      at Object.Module._extensions..js (module.js:416:10)
      at Module.load (module.js:343:32)
  {"name":"bar","hostname":"danger0.local","pid":47411,"level":30,"foo":"(Error in Bunyan log \"foo\" serializer broke field. See stderr for details.)","msg":"one","time":"2017-03-08T02:53:51.173Z","v":0}
  

• A serializer function should never mutate the given object. Doing so will change the object in your application.

• A serializer function should be defensive. In my experience, it is common to set a serializer in an app, say for field name "foo", and then accidentally have a log line that passes a "foo" that is undefined, or null, or of some unexpected type. A good start at defensiveness is to start with this:

  1function fooSerializer(foo) {
  2    // Guard against foo be null/undefined. Check that expected fields
  3    // are defined.
  4    if (!foo || !foo.bar)
  5        return foo;
  6    var obj = {
  7        // Create the object to be logged.
  8        bar: foo.bar
  9    }
  10    return obj;
  11};

Standard Serializers

Bunyan includes a small set of "standard serializers", exported as bunyan.stdSerializers. Their use is completely optional. An example using all of them:

1var log = bunyan.createLogger({
2    name: 'myapp',
3    serializers: bunyan.stdSerializers
4});

or particular ones:

1var log = bunyan.createLogger({
2    name: 'myapp',
3    serializers: {err: bunyan.stdSerializers.err}
4});

Standard serializers are:

Field	Description
err	Used for serializing JavaScript error objects, including traversing an error's cause chain for error objects with a .cause() -- e.g. as from verror.
req	Common fields from a node.js HTTP request object.
res	Common fields from a node.js HTTP response object.

Note that the req and res serializers intentionally do not include the request/response body, as that can be prohibitively large. If helpful, the restify framework's audit logger plugin has its own req/res serializers that include more information (optionally including the body).

src

The source file, line and function of the log call site can be added to log records by using the src: true config option:

1var log = bunyan.createLogger({src: true, ...});

This adds the call source info with the 'src' field, like this:

1{
2  "name": "src-example",
3  "hostname": "banana.local",
4  "pid": 123,
5  "component": "wuzzle",
6  "level": 4,
7  "msg": "This wuzzle is woosey.",
8  "time": "2012-02-06T04:19:35.605Z",
9  "src": {
10    "file": "/Users/trentm/tm/node-bunyan/examples/src.js",
11    "line": 20,
12    "func": "Wuzzle.woos"
13  },
14  "v": 0
15}

WARNING: Determining the call source info is slow. Never use this option in production.

Levels

The log levels in bunyan are as follows. The level descriptions are best practice opinions of the author.

• "fatal" (60): The service/app is going to stop or become unusable now. An operator should definitely look into this soon.
• "error" (50): Fatal for a particular request, but the service/app continues servicing other requests. An operator should look at this soon(ish).
• "warn" (40): A note on something that should probably be looked at by an operator eventually.
• "info" (30): Detail on regular operation.
• "debug" (20): Anything else, i.e. too verbose to be included in "info" level.
• "trace" (10): Logging from external libraries used by your app or very detailed application logging.

Setting a logger instance (or one of its streams) to a particular level implies that all log records at that level and above are logged. E.g. a logger set to level "info" will log records at level info and above (warn, error, fatal).

While using log level names is preferred, the actual level values are integers internally (10 for "trace", ..., 60 for "fatal"). Constants are defined for the levels: bunyan.TRACE ... bunyan.FATAL. The lowercase level names are aliases supported in the API, e.g. log.level("info"). There is one exception: DTrace integration uses the level names. The fired DTrace probes are named 'bunyan-$levelName'.

Here is the API for querying and changing levels on an existing logger. Recall that a logger instance has an array of output "streams":

1log.level() -> INFO   // gets current level (lowest level of all streams)
2
3log.level(INFO)       // set all streams to level INFO
4log.level("info")     // set all streams to level INFO
5
6log.levels() -> [DEBUG, INFO]   // get array of levels of all streams
7log.levels(0) -> DEBUG          // get level of stream at index 0
8log.levels("foo")               // get level of stream with name "foo"
9
10log.levels(0, INFO)             // set level of stream 0 to INFO
11log.levels(0, "info")           // can use "info" et al aliases
12log.levels("foo", WARN)         // set stream named "foo" to WARN

Level suggestions

Trent's biased suggestions for server apps: Use "debug" sparingly. Information that will be useful to debug errors post mortem should usually be included in "info" messages if it's generally relevant or else with the corresponding "error" event. Don't rely on spewing mostly irrelevant debug messages all the time and sifting through them when an error occurs.

Trent's biased suggestions for node.js libraries: IMHO, libraries should only ever log at trace-level. Fine control over log output should be up to the app using a library. Having a library that spews log output at higher levels gets in the way of a clear story in the app logs.

Log Record Fields

This section will describe rules for the Bunyan log format: field names, field meanings, required fields, etc. However, a Bunyan library doesn't strictly enforce all these rules while records are being emitted. For example, Bunyan will add a time field with the correct format to your log records, but you can specify your own. It is the caller's responsibility to specify the appropriate format.

The reason for the above leniency is because IMO logging a message should never break your app. This leads to this rule of logging: a thrown exception from log.info(...) or equivalent (other than for calling with the incorrect signature) is always a bug in Bunyan.

A typical Bunyan log record looks like this:

1{"name":"myserver","hostname":"banana.local","pid":123,"req":{"method":"GET","url":"/path?q=1#anchor","headers":{"x-hi":"Mom","connection":"close"}},"level":3,"msg":"start request","time":"2012-02-03T19:02:46.178Z","v":0}

Pretty-printed:

1{
2  "name": "myserver",
3  "hostname": "banana.local",
4  "pid": 123,
5  "req": {
6    "method": "GET",
7    "url": "/path?q=1#anchor",
8    "headers": {
9      "x-hi": "Mom",
10      "connection": "close"
11    },
12    "remoteAddress": "120.0.0.1",
13    "remotePort": 51244
14  },
15  "level": 3,
16  "msg": "start request",
17  "time": "2012-02-03T19:02:57.534Z",
18  "v": 0
19}

Core fields

• v: Required. Integer. Added by Bunyan. Cannot be overridden. This is the Bunyan log format version (require('bunyan').LOG_VERSION). The log version is a single integer. 0 is until I release a version "1.0.0" of node-bunyan. Thereafter, starting with 1, this will be incremented if there is any backward incompatible change to the log record format. Details will be in "CHANGES.md" (the change log).
• level: Required. Integer. Added by Bunyan. Cannot be overridden. See the "Levels" section.
• name: Required. String. Provided at Logger creation. You must specify a name for your logger when creating it. Typically this is the name of the service/app using Bunyan for logging.
• hostname: Required. String. Provided or determined at Logger creation. You can specify your hostname at Logger creation or it will be retrieved via os.hostname().
• pid: Required. Integer. Filled in automatically at Logger creation.
• time: Required. String. Added by Bunyan. Can be overridden. The date and time of the event in ISO 8601 Extended Format format and in UTC, as from Date.toISOString().
• msg: Required. String. Every log.debug(...) et al call must provide a log message.
• src: Optional. Object giving log call source info. This is added automatically by Bunyan if the "src: true" config option is given to the Logger. Never use in production as this is really slow.

Go ahead and add more fields, and nested ones are fine (and recommended) as well. This is why we're using JSON. Some suggestions and best practices follow (feedback from actual users welcome).

Recommended/Best Practice Fields

• err: Object. A caught JS exception. Log that thing with log.info(err) to get:

  1...
  2"err": {
  3  "message": "boom",
  4  "name": "TypeError",
  5  "stack": "TypeError: boom\n    at Object.<anonymous> ..."
  6},
  7"msg": "boom",
  8...

  Or use the bunyan.stdSerializers.err serializer in your Logger and do this log.error({err: err}, "oops"). See "examples/err.js".

• req_id: String. A request identifier. Including this field in all logging tied to handling a particular request to your server is strongly suggested. This allows post analysis of logs to easily collate all related logging for a request. This really shines when you have a SOA with multiple services and you carry a single request ID from the top API down through all APIs (as node-restify facilitates with its 'Request-Id' header).

• req: An HTTP server request. Bunyan provides bunyan.stdSerializers.req to serialize a request with a suggested set of keys. Example:

  1{
  2  "method": "GET",
  3  "url": "/path?q=1#anchor",
  4  "headers": {
  5    "x-hi": "Mom",
  6    "connection": "close"
  7  },
  8  "remoteAddress": "120.0.0.1",
  9  "remotePort": 51244
  10}

• res: An HTTP server response. Bunyan provides bunyan.stdSerializers.res to serialize a response with a suggested set of keys. Example:

  1{
  2  "statusCode": 200,
  3  "header": "HTTP/1.1 200 OK\r\nContent-Type: text/plain\r\nConnection: keep-alive\r\nTransfer-Encoding: chunked\r\n\r\n"
  4}

Other fields to consider

• req.username: Authenticated user (or for a 401, the user attempting to auth).
• Some mechanism to calculate response latency. "restify" users will have an "X-Response-Time" header. A latency custom field would be fine.
• req.body: If you know that request bodies are small (common in APIs, for example), then logging the request body is good.

Streams

A "stream" is Bunyan's name for where it outputs log messages (the equivalent to a log4j Appender). Ultimately Bunyan uses a Writable Stream interface, but there are some additional attributes used to create and manage the stream. A Bunyan Logger instance has one or more streams. In general streams are specified with the "streams" option:

1var bunyan = require('bunyan');
2var log = bunyan.createLogger({
3    name: "foo",
4    streams: [
5        {
6            stream: process.stderr,
7            level: "debug"
8        },
9        ...
10    ]
11});

For convenience, if there is only one stream, it can be specified with the "stream" and "level" options (internally converted to a Logger.streams).

1var log = bunyan.createLogger({
2    name: "foo",
3    stream: process.stderr,
4    level: "debug"
5});

Note that "file" streams do not support this shortcut (partly for historical reasons and partly to not make it difficult to add a literal "path" field on log records).

If neither "streams" nor "stream" are specified, the default is a stream of type "stream" emitting to process.stdout at the "info" level.

Adding a Stream

After a bunyan instance has been initialized, you may add additional streams by calling the addStream function.

1var bunyan = require('bunyan');
2var log = bunyan.createLogger('myLogger');
3log.addStream({
4  name: "myNewStream",
5  stream: process.stderr,
6  level: "debug"
7});

stream errors

A Bunyan logger instance can be made to re-emit "error" events from its streams. Bunyan does so by default for type === "file" streams, so you can do this:

1var log = bunyan.createLogger({name: 'mylog', streams: [{path: LOG_PATH}]});
2log.on('error', function (err, stream) {
3    // Handle stream write or create error here.
4});

As of bunyan@1.7.0, the reemitErrorEvents field can be used when adding a stream to control whether "error" events are re-emitted on the Logger. For example:

var EventEmitter = require('events').EventEmitter;
var util = require('util');

function MyFlakyStream() {}
util.inherits(MyFlakyStream, EventEmitter);

MyFlakyStream.prototype.write = function (rec) {
    this.emit('error', new Error('boom'));
}

var log = bunyan.createLogger({
    name: 'this-is-flaky',
    streams: [
        {
            type: 'raw',
            stream: new MyFlakyStream(),
            reemitErrorEvents: true
        }
    ]
});
log.info('hi there');

The behaviour is as follows:

• reemitErrorEvents not specified: file streams will re-emit error events on the Logger instance.
• reemitErrorEvents: true: error events will be re-emitted on the Logger for any stream with a .on() function -- which includes file streams, process.stdout/stderr, and any object that inherits from EventEmitter.
• reemitErrorEvents: false: error events will not be re-emitted for any streams.

Note: "error" events are not related to log records at the "error" level as produced by log.error(...). See the node.js docs on error events for details.

stream type: stream

A type === 'stream' is a plain ol' node.js Writable Stream. A "stream" (the writable stream) field is required. E.g.: process.stdout, process.stderr.

1var log = bunyan.createLogger({
2    name: 'foo',
3    streams: [{
4        stream: process.stderr
5        // `type: 'stream'` is implied
6    }]
7});

Field	Required?	Default	Description
stream	Yes	-	A "Writable Stream", e.g. a std handle or an open file write stream.
type	No	n/a	`type == 'stream'` is implied if the `stream` field is given.
level	No	info	The level to which logging to this stream is enabled. If not specified it defaults to "info". If specified this can be one of the level strings ("trace", "debug", ...) or constants (`bunyan.TRACE`, `bunyan.DEBUG`, ...). This serves as a severity threshold for that stream so logs of greater severity will also pass through (i.e. If level="warn", error and fatal will also pass through this stream).
name	No	-	A name for this stream. This may be useful for usage of `log.level(NAME, LEVEL)`. See the [Levels section](#levels) for details. A stream "name" isn't used for anything else.

stream type: file

A type === 'file' stream requires a "path" field. Bunyan will open this file for appending. E.g.:

1var log = bunyan.createLogger({
2    name: 'foo',
3    streams: [{
4        path: '/var/log/foo.log',
5        // `type: 'file'` is implied
6    }]
7});

Field	Required?	Default	Description
path	Yes	-	A file path to which to log.
type	No	n/a	`type == 'file'` is implied if the `path` field is given.
level	No	info	The level to which logging to this stream is enabled. If not specified it defaults to "info". If specified this can be one of the level strings ("trace", "debug", ...) or constants (`bunyan.TRACE`, `bunyan.DEBUG`, ...). This serves as a severity threshold for that stream so logs of greater severity will also pass through (i.e. If level="warn", error and fatal will also pass through this stream).
name	No	-	A name for this stream. This may be useful for usage of `log.level(NAME, LEVEL)`. See the [Levels section](#levels) for details. A stream "name" isn't used for anything else.

stream type: rotating-file

WARNING on node 0.8 usage: Users of Bunyan's rotating-file should (a) be using at least bunyan 0.23.1 (with the fix for this issue), and (b) should use at least node 0.10 (node 0.8 does not support the unref() method on setTimeout(...) needed for the mentioned fix). The symptom is that process termination will hang for up to a full rotation period.

WARNING on cluster usage: Using Bunyan's rotating-file stream with node.js's "cluster" module can result in unexpected file rotation. You must not have multiple processes in the cluster logging to the same file path. In other words, you must have a separate log file path for the master and each worker in the cluster. Alternatively, consider using a system file rotation facility such as logrotate on Linux or logadm on SmartOS/Illumos. See this comment on issue #117 for details.

A type === 'rotating-file' is a file stream that handles file automatic rotation.

1var log = bunyan.createLogger({
2    name: 'foo',
3    streams: [{
4        type: 'rotating-file',
5        path: '/var/log/foo.log',
6        period: '1d',   // daily rotation
7        count: 3        // keep 3 back copies
8    }]
9});

This will rotate '/var/log/foo.log' every day (at midnight) to:

1/var/log/foo.log.0     # yesterday
2/var/log/foo.log.1     # 1 day ago
3/var/log/foo.log.2     # 2 days ago

Currently, there is no support for providing a template for the rotated files, or for rotating when the log reaches a threshold size.

Field	Required?	Default	Description
type	Yes	-	"rotating-file"
path	Yes	-	A file path to which to log. Rotated files will be "$path.0", "$path.1", ...
period	No	1d	The period at which to rotate. This is a string of the format "$number$scope" where "$scope" is one of "ms" (milliseconds -- only useful for testing), "h" (hours), "d" (days), "w" (weeks), "m" (months), "y" (years). Or one of the following names can be used "hourly" (means 1h), "daily" (1d), "weekly" (1w), "monthly" (1m), "yearly" (1y). Rotation is done at the start of the scope: top of the hour (h), midnight (d), start of Sunday (w), start of the 1st of the month (m), start of Jan 1st (y).
count	No	10	The number of rotated files to keep.
level	No	info	The level at which logging to this stream is enabled. If not specified it defaults to "info". If specified this can be one of the level strings ("trace", "debug", ...) or constants (`bunyan.TRACE`, `bunyan.DEBUG`, ...).
name	No	-	A name for this stream. This may be useful for usage of `log.level(NAME, LEVEL)`. See the [Levels section](#levels) for details. A stream "name" isn't used for anything else.

Note on log rotation: Often you may be using external log rotation utilities like logrotate on Linux or logadm on SmartOS/Illumos. In those cases, unless you are ensuring "copy and truncate" semantics (via copytruncate with logrotate or -c with logadm) then the fd for your 'file' stream will change. You can tell bunyan to reopen the file stream with code like this in your app:

1var log = bunyan.createLogger(...);
2...
3process.on('SIGUSR2', function () {
4    log.reopenFileStreams();
5});

where you'd configure your log rotation to send SIGUSR2 (or some other signal) to your process. Any other mechanism to signal your app to run log.reopenFileStreams() would work as well.

stream type: raw

• raw: Similar to a "stream" writable stream, except that the write method is given raw log record Objects instead of a JSON-stringified string. This can be useful for hooking on further processing to all Bunyan logging: pushing to an external service, a RingBuffer (see below), etc.

raw + RingBuffer Stream

Bunyan comes with a special stream called a RingBuffer which keeps the last N records in memory and does not write the data anywhere else. One common strategy is to log 'info' and higher to a normal log file but log all records (including 'trace') to a ringbuffer that you can access via a debugger, or your own HTTP interface, or a post-mortem facility like MDB or node-panic.

To use a RingBuffer:

1/* Create a ring buffer that stores the last 100 records. */
2var bunyan = require('bunyan');
3var ringbuffer = new bunyan.RingBuffer({ limit: 100 });
4var log = bunyan.createLogger({
5    name: 'foo',
6    streams: [
7        {
8            level: 'info',
9            stream: process.stdout
10        },
11        {
12            level: 'trace',
13            type: 'raw',    // use 'raw' to get raw log record objects
14            stream: ringbuffer
15        }
16    ]
17});
18
19log.info('hello world');
20console.log(ringbuffer.records);

This example emits:

1[ { name: 'foo',
2    hostname: '912d2b29',
3    pid: 50346,
4    level: 30,
5    msg: 'hello world',
6    time: '2012-06-19T21:34:19.906Z',
7    v: 0 } ]

third-party streams

See the user-maintained list in the Bunyan wiki.

Runtime log snooping via DTrace

On systems that support DTrace (e.g., illumos derivatives like SmartOS and OmniOS, FreeBSD, Mac), Bunyan will create a DTrace provider (bunyan) that makes available the following probes:

1log-trace
2log-debug
3log-info
4log-warn
5log-error
6log-fatal

Each of these probes has a single argument: the string that would be written to the log. Note that when a probe is enabled, it will fire whenever the corresponding function is called, even if the level of the log message is less than that of any stream.

DTrace examples

Trace all log messages coming from any Bunyan module on the system. (The -x strsize=4k is to raise dtrace's default 256 byte buffer size because log messages are longer than typical dtrace probes.)

1dtrace -x strsize=4k -qn 'bunyan*:::log-*{printf("%d: %s: %s", pid, probefunc, copyinstr(arg0))}'

Trace all log messages coming from the "wuzzle" component:

1dtrace -x strsize=4k -qn 'bunyan*:::log-*/strstr(this->str = copyinstr(arg0), "\"component\":\"wuzzle\"") != NULL/{printf("%s", this->str)}'

Aggregate debug messages from process 1234, by message:

1dtrace -x strsize=4k -n 'bunyan1234:::log-debug{@[copyinstr(arg0)] = count()}'

Have the bunyan CLI pretty-print the traced logs:

1dtrace -x strsize=4k -qn 'bunyan1234:::log-*{printf("%s", copyinstr(arg0))}' | bunyan

A convenience handle has been made for this:

1bunyan -p 1234

On systems that support the jstack action via a node.js helper, get a stack backtrace for any debug message that includes the string "danger!":

1dtrace -x strsize=4k -qn 'log-debug/strstr(copyinstr(arg0), "danger!") != NULL/{printf("\n%s", copyinstr(arg0)); jstack()}'

Output of the above might be:

{"name":"foo","hostname":"763bf293-d65c-42d5-872b-4abe25d5c4c7.local","pid":12747,"level":20,"msg":"danger!","time":"2012-10-30T18:28:57.115Z","v":0}

          node`0x87e2010
          DTraceProviderBindings.node`usdt_fire_probe+0x32
          DTraceProviderBindings.node`_ZN4node11DTraceProbe5_fireEN2v85LocalINS1_5ValueEEE+0x32d
          DTraceProviderBindings.node`_ZN4node11DTraceProbe4FireERKN2v89ArgumentsE+0x77
          << internal code >>
          (anon) as (anon) at /root/node-bunyan/lib/bunyan.js position 40484
          << adaptor >>
          (anon) as doit at /root/my-prog.js position 360
          (anon) as list.ontimeout at timers.js position 4960
          << adaptor >>
          << internal >>
          << entry >>
          node`_ZN2v88internalL6InvokeEbNS0_6HandleINS0_10JSFunctionEEENS1_INS0_6ObjectEEEiPS5_Pb+0x101
          node`_ZN2v88internal9Execution4CallENS0_6HandleINS0_6ObjectEEES4_iPS4_Pbb+0xcb
          node`_ZN2v88Function4CallENS_6HandleINS_6ObjectEEEiPNS1_INS_5ValueEEE+0xf0
          node`_ZN4node12MakeCallbackEN2v86HandleINS0_6ObjectEEENS1_INS0_8FunctionEEEiPNS1_INS0_5ValueEEE+0x11f
          node`_ZN4node12MakeCallbackEN2v86HandleINS0_6ObjectEEENS1_INS0_6StringEEEiPNS1_INS0_5ValueEEE+0x66
          node`_ZN4node9TimerWrap9OnTimeoutEP10uv_timer_si+0x63
          node`uv__run_timers+0x66
          node`uv__run+0x1b
          node`uv_run+0x17
          node`_ZN4node5StartEiPPc+0x1d0
          node`main+0x1b
          node`_start+0x83

          node`0x87e2010
          DTraceProviderBindings.node`usdt_fire_probe+0x32
          DTraceProviderBindings.node`_ZN4node11DTraceProbe5_fireEN2v85LocalINS1_5ValueEEE+0x32d
          DTraceProviderBindings.node`_ZN4node11DTraceProbe4FireERKN2v89ArgumentsE+0x77
          << internal code >>
          (anon) as (anon) at /root/node-bunyan/lib/bunyan.js position 40484
          << adaptor >>
          (anon) as doit at /root/my-prog.js position 360
          (anon) as list.ontimeout at timers.js position 4960
          << adaptor >>
          << internal >>
          << entry >>
          node`_ZN2v88internalL6InvokeEbNS0_6HandleINS0_10JSFunctionEEENS1_INS0_6ObjectEEEiPS5_Pb+0x101
          node`_ZN2v88internal9Execution4CallENS0_6HandleINS0_6ObjectEEES4_iPS4_Pbb+0xcb
          node`_ZN2v88Function4CallENS_6HandleINS_6ObjectEEEiPNS1_INS_5ValueEEE+0xf0
          node`_ZN4node12MakeCallbackEN2v86HandleINS0_6ObjectEEENS1_INS0_8FunctionEEEiPNS1_INS0_5ValueEEE+0x11f
          node`_ZN4node12MakeCallbackEN2v86HandleINS0_6ObjectEEENS1_INS0_6StringEEEiPNS1_INS0_5ValueEEE+0x66
          node`_ZN4node9TimerWrap9OnTimeoutEP10uv_timer_si+0x63
          node`uv__run_timers+0x66
          node`uv__run+0x1b
          node`uv_run+0x17
          node`_ZN4node5StartEiPPc+0x1d0
          node`main+0x1b
          node`_start+0x83

Runtime environments

Node-bunyan supports running in a few runtime environments:

• Node.js
• Browserify: See the Browserify section below.
• Webpack: See the Webpack section below.
• NW.js

Support for other runtime environments is welcome. If you have suggestions, fixes, or mentions that node-bunyan already works in some other JavaScript runtime, please open an issue or a pull request.

The primary target is Node.js. It is the only environment in which I regularly test. If you have suggestions for how to automate testing for other environments, I'd appreciate feedback on this automated testing issue.

Browserify

As the Browserify site says it "lets you require('modules') in the browser by bundling up all of your dependencies." It is a build tool to run on your node.js script to bundle up your script and all its node.js dependencies into a single file that is runnable in the browser via:

1<script src="play.browser.js"></script>

As of version 1.1.0, node-bunyan supports being run via Browserify. The default stream when running in the browser is one that emits raw log records to console.log/info/warn/error.

Here is a quick example showing you how you can get this working for your script.

1. Get browserify and bunyan installed in your module:

   1$ npm install browserify bunyan

2. An example script using Bunyan, "play.js":

   1var bunyan = require('bunyan');
   2var log = bunyan.createLogger({name: 'play', level: 'debug'});
   3log.trace('this one does not emit');
   4log.debug('hi on debug');   // console.log
   5log.info('hi on info');     // console.info
   6log.warn('hi on warn');     // console.warn
   7log.error('hi on error');   // console.error

3. Build this into a bundle to run in the browser, "play.browser.js":

   1$ ./node_modules/.bin/browserify play.js -o play.browser.js

4. Put that into an HTML file, "play.html":

   1<!DOCTYPE html>
   2<html>
   3<head>
   4  <meta charset="utf-8">
   5  <script src="play.browser.js"></script>
   6</head>
   7<body>
   8  <div>hi</div>
   9</body>
   10</html>

5. Open that in your browser and open your browser console:

   1$ open play.html

Here is what it looks like in Firefox's console:

For some, the raw log records might not be desired. To have a rendered log line you'll want to add your own stream, starting with something like this:

1var bunyan = require('./lib/bunyan');
2
3function MyRawStream() {}
4MyRawStream.prototype.write = function (rec) {
5    console.log('[%s] %s: %s',
6        rec.time.toISOString(),
7        bunyan.nameFromLevel[rec.level],
8        rec.msg);
9}
10
11var log = bunyan.createLogger({
12    name: 'play',
13    streams: [
14        {
15            level: 'info',
16            stream: new MyRawStream(),
17            type: 'raw'
18        }
19    ]
20});
21
22log.info('hi on info');

webpack

To include bunyan in your webpack bundle you need to tell webpack to ignore the optional dependencies that are unavailable in browser environments.

Mark the following dependencies as externals in your webpack configuration file to exclude them from the bundle:

module: {
    externals: ['dtrace-provider', 'fs', 'mv', 'os', 'source-map-support']
}

Versioning

All versions are <major>.<minor>.<patch> which will be incremented for breaking backward compat and major reworks, new features without breaking change, and bug fixes, respectively. tl;dr: Semantic versioning.

License

MIT.

See Also

See the user-maintained list of Bunyan-related software in the Bunyan wiki.