A command-line tool for running node.js servers.
cluster to automatically fork server processes into the backgrounddomain module to catch and report errors per-request$ ls .
package.json node_modules/ app.js
$ cat app.js
module.exports = http.createServer(function(req, resp) {
resp.end('woo')
})
$ utensil -p 8000
spawning 8 processes listening on 0.0.0.0:8000
Zing! utensil will also handle unhandled exceptions for you, serving up a
500 page where appropriate (without taking down your node process!)
The port to bind to.
The address to bind to.
The number of processes to fork. Defaults to the number of cpus
reported by require('os').cpus().length.
The millisecond interval that child process should report their memory usage.
Path to a JavaScript module that exports a single function taking pid and msg.
Messages are of the form:
{ "type": "metrics" | "error" | "console"
, "subtype": <subtype>
, "data": <message> }
data will include the output of process.memoryUsage().
{ "type": "metrics"
, "subtype": "request"
, "data": { "start": <unix ms epoch timestamp>
, "elapsed": <ms from request received till response>
, "status": <HTTP status code of response>
, "url": <contents of `req.url`> } }
For global, pre-runtime errors:
{ "type": "error"
, "subtype": "application"
, "data": { "err": <err+''>
, "stack": <err.stack if available> } }
For errors during the duration of a request:
{ "type": "error"
, "subtype": "request"
, "data": { "err": <err+''>
, "start": <unix ms epoch timestamp>
, "elapsed": <ms from request received till response>
, "stack": <err.stack if available> } }
All console output will be redirected to the adaptor, as well.
{ "type": "console"
, "subtype": "error" | "warn" | "log"
, "data": [arguments to console] }
// console.js
// simply output everything.
module.exports = function(pid, msg) {
console.log('#%s %s/%s', pid, msg.type, msg.subtype, msg.data)
}
// run it with
// utensil --adaptor path/to/console.js
MIT