问题
Problem: We start pm2 in cluster mode, and pm2 starts as many processes as there are cpu cores, pm2 also tries to start as many node servers as there are cpu cores but the problem here is that it fails to start as many servers because they all try and start on the same port that is 3000, which already gets occupied by the first node server
We using nginx and proxy it to 3000 port.
we are using pm2 in cluster mode with the following configuration:
{
"apps" : [{
"script" : "npm",
"instances" : "max",
"cwd":"/home/nginx/my-pwa" ,
"args" : "run start:server:prod",
"exec_mode" : "cluster",
"wait_ready": true,
"kill_timeout" : 4000,
"watch" : true
}]
}
run start:server:prod is our script to start the server
Our express server:
var app = require('../src/app');
var port = normalizePort(process.env.PORT || '3000');
app.set('port', port);
const http = require('http');
server = http.createServer(app);
server.listen(port));
server.on('error', onError);
server.on('listening', onListening);
function onError(error) {
if (error.syscall !== 'listen') {
throw error;
}
var bind = typeof port === 'string'
? 'Pipe ' + port
: 'Port ' + port;
// handle specific listen errors with friendly messages
switch (error.code) {
case 'EACCES':
console.error(bind + ' requires elevated privileges');
process.exit(1);
break;
case 'EADDRINUSE':
console.error(bind + ' is already in use');
process.exit(1);
break;
default:
throw error;
}
}
process.on('message', function(msg) {
if (msg == 'shutdown') {
server.close();
process.exit(0);
}
});
// Listening logic
function onListening() {
var addr = server.address();
var bind = typeof addr === 'string'
? 'pipe ' + addr
: 'port ' + addr.port;
debug('Listening on ' + bind);
console.log("Server started on ", bind);
process.send('ready');
}
Please help, it's mission critical!
回答1:
The problem is that pm2 doesn't play well with npm. It is not able to start two node servers using npm script. The right way is to use node
My previous config:
{
"apps" : [{
"script" : "npm",
"instances" : "max",
"cwd":"/home/nginx/my-pwa" ,
"args" : "run start:server:prod",
"exec_mode" : "cluster",
"wait_ready": true,
"kill_timeout" : 4000,
"watch" : true
}]
}
My new config:
{
"apps" : [{
"script" : "./server/bin/www",
"instances" : "max",
"exec_mode" : "cluster",
"cwd":"/home/nginx/my-pwa" ,
"env": {"NODE_ENV" : "production"},
"name" : "my-pwa"
}]
}
As you can see that I am no longer using "script": "npm". ./server/bin/www contains my express server which pm2 will execute using node. Now pm2 is able to automatically handle clustering. So how's the output is supposed to look like now? It makes one god deamon which manages the worker server instances which are according to the number of your cpu cores. Output on our server: (It has 2 cores)
nginx 1363 1 0 05:20 ? 00:00:03 PM2 v2.10.1: God Daemon (/home/nginx/.pm2)
nginx 1373 1363 0 05:20 ? 00:00:09 node /home/nginx/my-pwa/server/bin/www
nginx 1374 1363 0 05:20 ? 00:00:09 node /home/nginx/my-pwa/server/bin/www
Also, now pm2 reload works fine. When I see logs on reload, pm2 starts new workers first and then shuts down the old workers.
回答2:
For me changing the exec_mode
to cluster_mode
worked.
My ecosystem.config.js was changed to something like this
apps: [
{
script: 'server.js',
instances: 'max',
exec_mode: 'cluster_mode' // not 'fork' or 'cluster'
}
]
回答3:
I don't think you can use pm2 cluster mode with npm scripts. First check if the process is actually in cluster mode using pm2 list
. If it isn't, try modifying your configuration to call the node script directly instead of the calling npm script.
来源:https://stackoverflow.com/questions/49691848/express-server-port-configuration-issue-with-pm2-cluster-mode