nodowntime

Copyright (c) 2017 Michael Welter [e-mail]

About

Simple module to gracefully restart worker processes one at a time to ensure no downtime.

To work correctly workers will need to handle disconnect message.

Install

$ npm install --save nodowntime

Restart

kill -s HUP $MAINPID

Usage

const cluster = require("cluster");
 
if (cluster.isMaster) {
    // no downtime module
    return require("./lib/nodowntime");
}
 
// graceful exit if our master requests it
cluster.worker.on("disconnect", () => {
    async.series([
        (callback) => server.close(() => callback()),
        (callback) => mongoose.disconnect(() => callback())
    ], () => process.exit());
});

Debug

NODE_DEBUG=nodowntime