Node v0.11 compatible
Log errors and stack traces in Sentry from within your Node.js applications. Includes middleware support for Connect/Express.
All processing and sending happens asynchronously to not slow things down if/when Sentry is down or slow.
- 0.6.x
- 0.8.x
- 0.10.x
- 0.11.x
Raven 0.6+ requires Sentry 6.0+
$ npm install raven
var raven = require('raven');
var client = new raven.Client('{{ SENTRY_DSN }}');
client.captureMessage('Hello, world!');
You can specify a level in the second optional parameter. Default level is error
There are 5 logging levels (in order):
- debug (the least serious)
- info
- warning
- error
- fatal (the most serious)
var raven = require('raven');
var client = new raven.Client('{{ SENTRY_DSN }}');
client.captureMessage("Another message", {level: 'info'})
Adding extra info to messages
var raven = require('raven');
var client = new raven.Client('{{ SENTRY_DSN }}');
client.captureMessage("Another message", {level: 'info', extra: {'key': 'value'}})
client.captureError(new Error('Broke!'));
client.captureQuery('SELECT * FROM `awesome`', 'mysql');
client.captureMessage('Hello, world!', function(result) {
console.log(client.getIdent(result));
});
client.captureError(new Error('Broke!'), function(result) {
console.log(client.getIdent(result));
});
Note: client.captureMessage
will also return the result directly without the need for a callback, such as: var result = client.captureMessage('Hello, world!');
If you really care if the event was logged or errored out, Client emits two events, logged
and error
:
client.on('logged', function(){
console.log('Yay, it worked!');
});
client.on('error', function(e){
console.log('oh well, Sentry is broke.');
})
client.captureMessage('Boom');
The event error is augmented with the original Sentry response object as well as the response body and statusCode for easier debugging.
client.on('error', function(e){
console.log(e.reason); // raw response body, usually contains a message explaining the failure
console.log(e.statusCode); // status code of the http request
console.log(e.response); // entire raw http response object
});
Optionally declare the DSN to use for the client through the environment. Initializing the client in your app won't require setting the DSN.
Optionally set the name for the client to use. What is name?
Optionally set the site for the client to use. What is site?
For those times when you don't catch all errors in your application. ;)
client.patchGlobal();
// or
raven.patchGlobal(client);
// or
raven.patchGlobal('{{ SENTRY_DSN }}');
It is recommended that you don't leave the process running after receiving an uncaughtException
(http://nodejs.org/api/process.html#process_event_uncaughtexception), so an optional callback is provided to allow you to hook in something like:
client.patchGlobal(function() {
console.log('Bye, bye, world.')
process.exit(1);
});
The callback is called after the event has been sent to the Sentry server.
new raven.Client(dsn[, options])
client.captureMessage(string[,callback])
client.captureError(Error[,callback])
client.captureQuery(string, string[,callback])
The Raven middleware can be used as-is with either Connect or Express in the same way. Take note that in your middlewares, Raven must appear after your main handler to pick up any errors that may result from handling a request.
var connect = require('connect');
function mainHandler(req, res) {
throw new Error('Broke!');
}
function onError(err, req, res, next) {
// The error id is attached to `res.sentry` to be returned
// and optionally displayed to the user for support.
res.statusCode = 500;
res.end(res.sentry+'\n');
}
connect(
connect.bodyParser(),
connect.cookieParser(),
mainHandler,
raven.middleware.connect('{{ SENTRY_DSN }}'),
onError, // optional error handler if you want to display the error id to a user
).listen(3000);
var app = require('express')();
app.use(app.router);
app.use(raven.middleware.express('{{ SENTRY_DSN }}'));
app.use(onError); // optional error handler if you want to display the error id to a user
app.get('/', function mainHandler(req, res) {
throw new Error('Broke!');
});
app.listen(3000);
Note: raven.middleware.express
or raven.middleware.connect
must be added to the middleware stack before any other error handling middlewares or there's a chance that the error will never get to Sentry.
In order to use raven-node with coffee-script or another library which overwrites Error.prepareStackTrace you might run into the exception "Traceback does not support Error.prepareStackTrace being defined already."
In order to not have raven-node (and the underlying raw-stacktrace library) require Traceback you can pass your own stackFunction in the options. For example:
var client = new raven.Client('{{ SENTRY_DSN }}', { stackFunction: {{ Your stack function }}});
So for example:
var client = new raven.Client('{{ SENTRY_DSN }}', {
stackFunction: Error.prepareStackTrace
});
Pass false
as the DSN (or any falsey value).
client = new raven.Client(process.env.NODE_ENV === 'production' && '{{ SENTRY_DSN }}')
Note: We don't infer this from NODE_ENV
automatically anymore. It's up to you to implement whatever logic you'd like.
You can find me on IRC. I troll in #sentry
on freenode
.