NOTE: up has been renamed in npm as "up-time" https://npmjs.org/package/up-time
Zero-downtime reloads built on top of the distribute load balancer.
Simply running
$ up --port 80 --watch my-http-server.jsWill start my-http-server.js on port 80, then reload it with no downtime
when files change in the working directory.
This project is deprecated!
- Works with Node 0.6+
- Works at the HTTP request level. It never drops requests or destroys
Keep-Alivesockets while reloading. - Compatible with any HTTP server.
- Easy-to-use CLI interface for development with automatic reloading upon file changes.
- Gracefully handles reloads with syntax errors during development.
- Built on distribute.
- Supports transpilers such as CoffeeScript.
Make sure you structure your code so that your http server lives in a
separate module that can be required.
server.js
module.exports = http.Server(function (req, res) {
res.writeHead(200);
res.end('Hello World');
});To get the up command, make sure to install with the -g flag:
$ npm install -g upUsage: up [options]
The up command accepts the following options:
-
-p/--port- The port to listen on. Not required if the module already
listens. - Defaults to
3000.
- The port to listen on. Not required if the module already
-
-w/--watch- Whether to watch for changes.
- Watches the working directory for changes.
-
-r/--require<mod>- Specifies a module to require from each worker.
- Can be used multiple times.
- Tranpilers such as CoffeeScript can be supported with
--require|-r coffee-script
-
-n/--number- Number of workers. It gets evaluated with eq.js.
- You can optionally use the
cpusvariable. eg:cpus + 2. - You can use all the
Mathmethods. eg:round(cpus / 2). - Defaults to number of CPUS, or
1ifNODE_ENVisdevelopment.
-
-t/--timeout- Number of ms after which a worker is killed once it becomes inactive.
- Strings like
'10s'are accepted. - Defaults to
'10m', or'500ms'ifNODE_ENVisdevelopment.
-
-k/--keepalive- start a new worker after one dies unexpectedly
-
-f/--pidfile- A filename to write the pid to
- If specified, restarts can be achieved with: "kill -s SIGUSR2
cat pidfile.txt"
-
-T/--title- A value to give
process.title. - Defaults to
up. - The value will be appended with
masterorworker(e.g "up master", "up worker").
- A value to give
var up = require('up')
, master = http.Server().listen(3000)
// initialize up
var srv = up(master, __dirname + '/server');
process.on('SIGUSR2', function () {
srv.reload();
});require('up') exports the UpServer constructor, which takes three
parameters:
- server (
http.Server) server to accept connections on - module (
String) absolute path to the module. - options (
Object)numWorkers: (Number|String): see--workersabove.workerTimeout: (Number|String): see--timeoutabove.title: (String): see--titleabove.assumeReady: (Boolean): see Worker readiness below.keepAlive: (Boolean): see--keepaliveabove.minExpectedLifetime: (Number|String): Number of ms a worker is expected to live. Don't auto-respawn if a worker dies earlier. Strings like'10s'are accepted. Defaults to'20s'.
An UpServer inherits from a Distributor, which means you can use()
any distribute middleware.
The main difference is that the "default handler" of up (ie: the last function in the middleware chain) is the one that executes the round-robin load balancing.
To reload the workers, call srv.reload(). In the example above and CLI,
this is called by sending the SIGUSR2 signal:
$ kill -s SIGUSR2 <process id>If you're running with up CLI, this command is output to stderr for your
convenience.
The CLI tool also auto-reloads if you pass the --watch option and a file
changes in the working directory.
- An up server starts with an arbitrary number of workers, which defaults to the number of CPUs.
- When a reload instruction is received, it spawns an identical number of workers.
- Upon the first of those workers binding to a port, any subsequent requests are sent to that worker, and all the workers containing old code are discarded.
- The discarded workers could have been processing requests, so they only
truly die after the configured
workerTimeout, which defaults to 10 minutes in production. This means that if a user was uploading a file, his request will be processed without interruptions. - As other workers bind and become available, they join the round-robin round.
By default up assume that new workers are ready for new connections,
immediately after they have been required. This can be changed by setting
assumeReady to false, on the options object when initializing
the up server through the JavaScript API.
The worker then needs to tell up, when it's ready, like this:
var up = require('up');
// Dummy async event
setTimeout(function(){
up.ready();
}, 1000);(The MIT License)
Copyright (c) 2011 Guillermo Rauch <[email protected]>
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the 'Software'), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED 'AS IS', WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
![Kimberly Thach [Ghostglass] avatar](https://avatars.githubusercontent.com/u/64843479?s=40&u=f39105d57ba7536e4c522fcb6740cafcce2a7144&v=4 "Kimberly Thach [Ghostglass]")
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent