I've an app which initially creates static config files (once) and after files were written I need to reinitialize/restart the application.
Is there something to restart a node.js app from itself?
This is required cause I've an application running in two runlevels in node.js.
The initial one starts completly synchronus and after this level has been completed app is in async runlevel in a previously started environment.
I know there are tools like nodemon but that's not what I need in my case.
I tried to kill the app via process.kill() which is working but I can't listen to the kill event:
// Add the listener
process.on('exit', function(code) {
console.log('About to exit with code:', code);
// Start app again but how?
});
// Kill application
process.kill();
Or is there a better, cleaner way to handle this?
Found a working case to get node.js restarted from app itself:
Example:
// Optional part (if there's an running webserver which blocks a port required for next startup
try {
APP.webserver.close(); // Express.js instance
APP.logger("Webserver was halted", 'success');
} catch (e) {
APP.logger("Cant't stop webserver:", 'error'); // No server started
APP.logger(e, 'error');
}
// First I create an exec command which is executed before current process is killed
var cmd = "node " + APP.config.settings.ROOT_DIR + 'app.js';
// Then I look if there's already something ele killing the process
if (APP.killed === undefined) {
APP.killed = true;
// Then I excute the command and kill the app if starting was successful
var exec = require('child_process').exec;
exec(cmd, function () {
APP.logger('APPLICATION RESTARTED', 'success');
process.kill();
});
}
The only con I can see here is to loose outputs on console but if anything is logged into logfiles it's not a problem.
Related
If I have a node.js web server running with express, is there a way to get a command line? Kind of like the way the browser has a javascript console. I want to type in variables and inspect them in node.js runtime.
If I type in 'node' then it gives me a command line. but if I type in 'node server.js' that command line is not available. How can I make the command line always available (in bottom of screen) regardless of the server's state?
To do this type node --inspect server.js
Then to connect to it type in chrome://inspect/ in url bar of Chrome. Then click "inspect" on that instance under Remote Target. Then I found I can see variables/objects that are under the global namespace.
Also, a very basic way to add keyboard interaction on the server window itself without having to connect to it:
var readline = require('readline');
readline.emitKeypressEvents(process.stdin);
process.stdin.setRawMode(true);
setInterval(function() {
console.log('server stuff');
}, 1000);
process.stdin.on('keypress', (str, key) => {
if(key.sequence === '\u0003') {
console.log('ctrl-c pressed');
process.exit();
}
console.log('you pressed: ' + key.name);
});
I have a bash script that I am running server side from meteor. I have tested that I am successfully able to run shell commands by running 'ls' and getting the expected response back. However, when I run my shell script no output is ever logged to the server console and none of the intended effects of the script are successful. I am printing stderr,stdout, and error yet they print nothing when I run my script.
Meteor.methods({
grade: function(unittest, code) {
this.unblock();
var sys = Npm.require('sys');
var exec = Npm.require('child_process').exec;
// exec('/bin/ls /srv/srcalyzer/scripts', function(error,stdout,stderr){
// console.log('error: ',error);
// console.log('stdout: ',stdout);
// console.log('stderr: ',stderr);
// });
console.log('running grade')
exec('/bin/bash /srv/srcalyzer/scripts/grade.sh', function(error,stdout,stderr){
console.log('error: ',error);
console.log('stdout: ',stdout);
console.log('stderr: ',stderr);
});
console.log('just finished.');
},
Although it is currently commented out the /bin/ls /some/path logs the expected output to console. However when I run the /bin/bash /path/to/.sh that I know is in place. The console output looks like
I20161207-15:22:07.031(-5)? running grade
I20161207-15:22:07.045(-5)? just finished.
The script does take a short time to run (~15-20 seconds). I am uncertain if this is relevant or not.
Can someone please help me understand what is happening?
there's a hint here:
I20161207-15:22:07.031(-5)? running grade
I20161207-15:22:07.045(-5)? just finished.
that's only taking a few ms to run. meaning, your Meteor method is likely exiting before the exec() finishes. i've never run an exec() in Meteor, so i'm unsure if that shell script keeps running after your method exits.
what you need is to run a future in your Meteor method, so it doesn't exit until your shell script comes back.
something like:
let future = new Future();
exec('/bin/bash /srv/srcalyzer/scripts/grade.sh', function(error,stdout,stderr){
console.log('error: ',error);
console.log('stdout: ',stdout);
console.log('stderr: ',stderr);
future.return(stdout.toString());
});
return future.wait();
now your Meteor method will wait until your script finishes.
(caveat: i didn't try this, just typing this solution from memory on my non-work machine)
INFO: I'm referring to this question I asked on Super User, but couldn't get an answer and I think this is a good place to ask, since the problem is probably code related.
I'm currently running a simple Node.JS server with express.js on my RaspberryPi with Debian installed on it. Everything works fine, but every morning I wake up to see my server isn't running anymore (the server process I started with the command node main.js).
My first guess was, that the Pi has some kind of sleep mode, which it enters after a couple of hours without traffic/etc, and which shuts down the server, but I also run a dydns-client, which is still up every morning (I also was informed, that the RaspberryPi doesn't come with a sleep mode).
I wrote a simple script to check whether the process is running and writes it into a log file, but today morning I had to notice, that this script was wasn't running as well (only for around two hours, it logs the server state every 15 seconds and the last state was running).
Here is the script:
#!/bin/sh
MATCH="SOME_PROCESS_NAME"
while [ true ]
do
if [ "$(ps -ef | grep ${MATCH} | grep -v grep)" ]; then
echo "$(date): Process is running ..."
else
echo "$(date): Process has ended ..."
fi
sleep 15
done
Is there a way to track a process after I started it to check tomorrow morning, what killed my process or why it ended (the script obviously didn't work)?
The server itself looks pretty simple and I don't think there is some kind of auto-shutdown I missed. Here is the code I used.
var express = require('express');
var path = require('path');
var server = express();
server.use(express.static(path.join(__dirname, 'public')));
server.listen(1337);
console.log("Server listening (PORT: " + 1337 + ") ...");
Any idea what to do, to keep the server running/find out what is the stopping reason?
UPDATE: I received a working answer over at RaspberryPi-stackexchange.
My guess is the Raspberry Pi restarts at midnight or something similar. to fix this maybe add an entry for your server process rc.local file. you can add commands to the rc.local file by editing /etc/rc.local
Would this helps https://raspberrypi.stackexchange.com/questions/8741/when-does-the-os-kill-an-application ?
I would like to suggest a different approach to monitor your process until you can get more information, to edit, then check, then start (wrote on the fly)
var fs = require('fs')
var spawn = require('child_process');
var child = spawn(process.argv[0], 'your/bin.js', {stdio:['ignore', 'pipe', 'pipe']})
child.stdout.pipe(fs.createReadStream('stdout.log'))
child.stderr.pipe(fs.createReadStream('stderr.log'))
child.on('error', function (err) {
fs.writeFile('error.log', JSON.stringify(err), function () { /* void */ })
})
child.on('close', function (code, signal) {
fs.writeFile('exit.log', "code="+code+" signal="+signal, function () { /* void */ })
})
Here is my setup.
I have a grunt task that does 2 things: 1) start a http server listening on some port 2) triggers another grunt task
The grunt task triggered above is a testem task that runs tests on test-index.html page in PhantomJS.
The test-index.html page sends a POST call on the port on which I start a server in the first grunt task.
Issue: The POST call doesn't hit my server.
Note: If I run the same server manually (not from grunt) and then run the test grunt task, the POST call hits the server.
Heres the code:
Grunt task
grunt.registerMultiTask('coverage', 'Generates coverage reports for JS using Istanbul', function () {
var server = http.createServer(function(req, resp) {
resp.setHeader("Access-Control-Allow-Origin", "*");
console.log('Got something');
req.pipe(fs.createWriteStream('coverage.json'))
resp.end();
});
var port = 7358;
server.listen(port);
// This task simply executes a command: `testem.js ci`
grunt.task.run('testem').then(function() {
server.close();
});
});
test-index.html (somewhere in the )
function onTestemLoad() {
Testem.on('all-test-results', function(results){
var xhr = new XMLHttpRequest();
xhr.open('POST', 'http://localhost:7358/');
xhr.send(JSON.stringify(window.__coverage__))
});
}
Can anyone point what might be going wrong here?
you've been misled by grunt.task.run!
This line does NOT run the task, it just puts it on the top of the list of tasks to run once the current task is done (http://gruntjs.com/api/grunt.task#grunt.task.run).
A second issue is that when a task completes, it takes down all its running processes with it.
So what happens is:
coverage starts, spins on your server, then register testem as next task
coverage ends, taking down your server
next task (= testem) runs, but no server is here to answer...
You can fix it by doing the following 2 things:
inline your test: replace grunt.task.run('testem') with a direct call to testem
ensure your tests run synchronously - or keep your task alive until your tests are done (using this.async - see http://gruntjs.com/api/inside-tasks#this.async).
I want to wait for a server to be ready, then run a task on top of it
EDIT:
I've got some of this figured out. I am properly filtering for my string. However, how can I keep this running in the background, and have it terminate when a grunt sequence is complete?
Consider
grunt common server spec-e2e
This will run the common tasks, run the grunt server, then (without stopping the server) runs the next task on top. After all of this is complete, it turns off the server automatically.
That's the kind of functionality that I'm trying to create.
A couple important points that I learned:
Processes started from node - e.g. require('child_process').spawn(...) - will continue to run for all subsequent grunt tasks (they won't be closed once the single task completes).
Once grunt is finished running the sequence of tasks, it will terminate all spawned processes.
Therefore, the command grunt selendroid:selendroid.jar spec-e2e - where the selendroid task starts a server - will keep the server running until after the spec-e2e task completes.
The code for my selendroid task is:
grunt.registerTask('selendroid', 'Start selendroid server', function (appLocation) {
var done = this.async();
var timeout = setTimeout(function() {
console.log('Selendroid took too long to start');
done(false);
}, 30000);
function checkLoaded(data) {
data = data.toString();
if (data.indexOf('Selendroid standalone server has been started on port') > -1) {
clearTimeout(timeout);
done();
}
}
if (typeof appLocation === 'undefined') {
console.log('App to test not set, aborting...'.red);
done(false);
}
var cmd = require('child_process').spawn('java', ['-jar', 'selendroid.jar', '-app', appLocation]);
cmd.stdout.on('data', function(data) {
checkLoaded(data);
});
cmd.stderr.on('data', function(data) {
checkLoaded(data);
});
});