High-speed HTTP logger for Node.js
To our knowledge, pino-http
is the fastest HTTP logger in town.
Benchmarks log each request/response pair while returning
'hello world'
, using
autocannon with 100
connections and 10 pipelined requests.
http-ndjson
(equivalent info): 7730.73 req/sechttp-ndjson
(standard minimum info): 9522.37 req/secpino-http
: 21496 req/secpino-http
(extreme): 25770.91 req/sec- no logger: 46139.64 req/sec
All benchmarks where taken on a Macbook Pro 2013 (2.6GHZ i7, 16GB of RAM).
npm i pino-http --save
'use strict'
var http = require('http')
var server = http.createServer(handle)
var logger = require('pino-http')()
function handle (req, res) {
logger(req, res)
req.log.info('something else')
res.end('hello world')
}
server.listen(3000)
$ node example.js | pino
[2016-03-31T16:53:21.079Z] INFO (46316 on MBP-di-Matteo): something else
req: {
"id": 1,
"method": "GET",
"url": "/",
"headers": {
"host": "localhost:3000",
"user-agent": "curl/7.43.0",
"accept": "*/*"
},
"remoteAddress": "::1",
"remotePort": 64386
}
[2016-03-31T16:53:21.087Z] INFO (46316 on MBP-di-Matteo): request completed
res: {
"statusCode": 200,
"header": "HTTP/1.1 200 OK\r\nX-Powered-By: restify\r\nContent-Type: text/html; charset=utf-8\r\nContent-Length: 11\r\nETag: W/\"b-XrY7u+Ae7tCTyyK7j1rNww\"\r\nDate: Thu, 31 Mar 2016 16:53:21 GMT\r\nConnection: keep-alive\r\n\r\n"
}
responseTime: 10
req: {
"id": 1,
"method": "GET",
"url": "/",
"headers": {
"host": "localhost:3000",
"user-agent": "curl/7.43.0",
"accept": "*/*"
},
"remoteAddress": "::1",
"remotePort": 64386
}
opts
: it has all the options as pino and
logger
:pino-http
can reuse a pino instance if passed with thelogger
propertygenReqId
: you can pass a function which gets used to generate a request id. The first argument is the request itself. As fallbackpino-http
is just using an integer. This default might not be the desired behavior if you're running multiple instances of the appuseLevel
: the logger levelpino-http
is using to log out the response. default:info
customLogLevel
: set to afunction (res, err) => { /* returns level name string */ }
. This function will be invoked to determine the level at which the log should be issued. This option is mutually exclusive with theuseLevel
option. The first argument is the HTTP response. The second argument is an error object if an error has occurred in the request.autoLogging
: set tofalse
, to disable the automatic "request completed" and "request errored" logging. Defaults totrue
. If set to an object, you can provide more options.autoLogging.ignorePaths
: array that holds one or many paths that should not autolog on completion. Paths will be matched exactly to the url path (using Node classURL.pathname
). This is useful for ignoring e.g. health check paths that get called every X seconds, and would fill out the logs unnecessarily. If the path matches and succeeds (http 200), it will not log any text. If it fails, it will log the error (as with any other path).stream
: same as the second parameter
stream
: the destination stream. Could be passed in as an option too.
'use strict'
var http = require('http')
var server = http.createServer(handle)
var pino = require('pino')
var logger = require('pino-http')({
// Reuse an existing logger instance
logger: pino(),
// Define a custom request id function
genReqId: function (req) { return req.id },
// Define custom serializers
serializers: {
err: pino.stdSerializers.err,
req: pino.stdSerializers.req,
res: pino.stdSerializers.res
},
// Logger level is `info` by default
useLevel: 'info',
// Define a custom logger level
customLogLevel: function (res, err) {
if (res.statusCode >= 400 && res.statusCode < 500) {
return 'warn'
} else if (res.statusCode >= 500 || err) {
return 'error'
}
return 'info'
}
})
function handle (req, res) {
logger(req, res)
req.log.info('something else')
res.end('hello world')
}
server.listen(3000)
The pinoHttp
function has a property called startTime
which contains a symbol
that is used to attach and reference a start time on the HTTP res
object. If the function
returned from pinoHttp
is not the first function to be called in an HTTP servers request
listener function then the responseTime
key in the log output will be offset by any
processing that happens before a response is logged. This can be corrected by manually attaching
the start time to the res
object with the pinoHttp.startTime
symbol, like so:
var http = require('http')
var logger = require('pino-http')()
var someImportantThingThatHasToBeFirst = require('some-important-thing')
http.createServer((req, res) => {
res[logger.startTime] = Date.now()
someImportantThingThatHasToBeFirst(req, res)
logger(req, res)
res.end('hello world')
}).listen(3000)
Generates a JSONifiable object from the HTTP request
object passed to
the createServer
callback of Node's HTTP server.
It returns an object in the form:
{
pid: 93535,
hostname: 'your host',
level: 30,
msg: 'my request',
time: '2016-03-07T12:21:48.766Z',
v: 0,
req: {
id: 42,
method: 'GET',
url: '/',
headers: {
host: 'localhost:50201',
connection: 'close'
},
remoteAddress: '::ffff:127.0.0.1',
remotePort: 50202
}
}
Generates a JSONifiable object from the HTTP response
object passed to
the createServer
callback of Node's HTTP server.
It returns an object in the form:
{
pid: 93581,
hostname: 'myhost',
level: 30,
msg: 'my response',
time: '2016-03-07T12:23:18.041Z',
v: 0,
res: {
statusCode: 200,
header: 'HTTP/1.1 200 OK\r\nDate: Mon, 07 Mar 2016 12:23:18 GMT\r\nConnection: close\r\nContent-Length: 5\r\n\r\n'
}
}
Each of the standard serializers can be extended by supplying a corresponding
custom serializer. For example, let's assume the request
object has custom
properties attached to it, and that all of the custom properties are prefixed
by foo
. In order to show these properties, along with the standard serialized
properties, in the resulting logs, we can supply a serializer like:
var http = require('http')
var logger = require('pino-http')({
serializers: {
req (req) {
Object.keys(req.raw).forEach((k) => {
if (k.startsWith('foo')) {
req[k] = req.raw[k]
}
})
return req
}
}
})
Logging of requests' bodies is disabled by default since it can cause security risks such as having private user information (password, other GDPR-protected data, etc.) logged (and persisted in most setups). However if enabled, sensitive information can be redacted as per redaction documentation.
Furthermore, logging more bytes does slow down throughput. This video by pino maintainers Matteo Collina & David Mark Clements goes into this in more detail.
After considering these factors, logging of the request body can be achieved as follows:
const http = require('http')
const logger = require('pino-http')({
serializers: {
req(req) {
req.body = req.raw.body;
return req;
},
},
});
https://www.npmjs.com/~matteo.collina
https://twitter.com/matteocollina
https://github.com/davidmarkclements
https://www.npmjs.com/~davidmarkclements
https://twitter.com/davidmarkclem
This project was kindly sponsored by nearForm.
Logo and identity designed by Beibhinn Murphy O'Brien: https://www.behance.net/BeibhinnMurphyOBrien.
MIT