When I run pm2, it creates two cron-out and cron-error files in the logs folder. How to not create these files when starting pm2?
You can Generate Configuration
pm2 ecosystem
This will generate following file (ecosystem.config.js) ,pass error_file: "/dev/null" to not to generate pm2 logs
module.exports = {
apps : [{
name: "app",
script: "./app.js",
error_file: "/dev/null"
env: {
NODE_ENV: "development",
},
env_production: {
NODE_ENV: "production",
}
}]
}
pm2 [start|restart|stop|delete] ecosystem.config.js
This is the link I followed -http://pm2.keymetrics.io/docs/usage/application-declaration/
Or Shortest Way
pm2 start app /dev/null
Related
I have two nodejs app deployed on Digital Ocean, with PM2, which I am learning.
I'd like to know how I could update my apps (services) separately.
At the moment my pm2 config looks like this:
// ecosystem.config.js
module.exports = {
apps: [
{
name: `app1`,
script: './app1.js',
// ...
},
{
name: `app2`,
script: './app2.js',
// ...
}
],
deploy: {
production: {
key: '',
user: 'admin',
host: '',
ref: '',
repo: '',
path: '/home/admin/app/prod',
'post-deploy': './deploy.sh production'
},
development: {
key: '',
user: 'admin',
host: '',
ref: '',
repo: '',
path: '/home/admin/app/development',
'post-deploy': './deploy.sh development'
},
}
}
I can deploy per environment, like pm2 deploy production.
But, can I do something like pm2 deploy development app2 to only update app2?
You have many ways to do that, for me I always create ecosystem file in the client folder and server folder, then create publish ecosystem file in the root folder, like:
client
ecosystem.config.js
server
ecosystem.config.js
ecosystem.config.js
The root ecosystem file like:
deploy: {
frontend: {
...other,
path: '/app/easyv_spaceship',
'post-deploy': 'cd ./client && npm install && npm run build'
},
backend: {
...other,
path: '/app/easyv_spaceship',
'post-deploy': 'cd ./server && npm install && npm run tsc && pm2 startOrRestart ecosystem.config.js --env production'
}
}
hope this can inspire you.
I have the following ecosystem.config.js file for my application:
module.exports = {
apps: [
{
name: 'Worker A',
script: 'scripts/workerA.js',
instances: 4,
autorestart: true,
watch: false,
instance_var: 'INSTANCE_ID',
max_memory_restart: '1G'
},
{
name: 'Worker B',
script: 'scripts/workerB.js',
instances: 4,
autorestart: true,
watch: false,
instance_var: 'INSTANCE_ID',
max_memory_restart: '1G'
}
]
}
When the processes are started via pm2 start ecosystem.config.js everything is working fine. However, if I now change the instances of the applications to a different number (i.e. 8) and try to restart the processes via pm2 restart ecosystem.config.js, the same number of instances will be started as before. pm2 reload ecosystem.config.js and pm2 startOrReload ecosystem.config.js show the same behaviour.
The only way I have solved the problem is to completely delete all the processes with pm2 delete ecosystem.config.js and then start them all again. While this works, it robs me of the possiblity to use the "0-second-downtime" restart that pm2 reload ... offers.
Is there any solution to this? Thanks!
$ pm2 -v
4.2.3
scaling nodes only work in cluster mode in pm2 (https://pm2.keymetrics.io/docs/usage/cluster-mode/)
so add
exec_mode : "cluster" to your configuration
module.exports = {
apps: [
{
name: 'Worker A',
script: 'scripts/workerA.js',
instances: 4,
autorestart: true,
watch: false,
instance_var: 'INSTANCE_ID',
max_memory_restart: '1G',
exec_mode : "cluster"
},
{
name: 'Worker B',
script: 'scripts/workerB.js',
instances: 4,
autorestart: true,
watch: false,
instance_var: 'INSTANCE_ID',
max_memory_restart: '1G',
exec_mode : "cluster"
}
]
}
I searched all over, but could not find the answer
for how to config pm2 with Express.js
here is what i have so far based on other's answers and pm2 documentation.
this is on the server main file(index.js):
const port = process.env.NODE_PORT;
I get undefined unless i use || 8080 .
const port = process.env.NODE_PORT || 8080;
I need it working only on dev env for now..
but it seems it does not get what i config on ecosystem.config.js file.
and on my ecosystem.config.js:
module.exports = {
apps: [{
name: 'API',
script: 'index.js',
// Options reference: https://pm2.io/doc/en/runtime/reference/ecosystem-file/
args: 'one two',
instances: 1,
exec_mode: "fork_mode",
autorestart: true,
watch: false,
max_memory_restart: '1G',
env: {
NODE_PORT = 8080 `pm2 start app.js -f`,
NODE_PORT = 8081 `pm2 start app.js -f`,
NODE_ENV: 'development'
},
env_production: {
NODE_ENV: 'production'
}
},
{
name: 'API',
script: 'index.js',
// Options reference: https://pm2.io/doc/en/runtime/reference/ecosystem-file/
args: 'one two',
instances: 1,
exec_mode: "fork_mode",
autorestart: true,
watch: false,
max_memory_restart: '1G',
env: {
PORT: 8081,
NODE_ENV: 'development'
},
env_production: {
NODE_ENV: 'production'
}
}
],
deploy: {
production: {
user: 'node',
host: '212.83.163.1',
ref: 'origin/master',
repo: 'git#github.com:repo.git',
path: '/var/www/production',
'post-deploy': 'npm install && pm2 reload ecosystem.config.js --env
production'
}
}
};
I'm using environment variable process.env.NODE_APP_INSTANCE to do this. (https://pm2.io/doc/en/runtime/guide/load-balancing/#cluster-environment-variable)
I'm setting my PORT before starting my server, and then I'm setting server port based on PORT environment variable and NODE_APP_INSTANCE, something like this:
const nodeInstance = parseInt(process.env.NODE_APP_INSTANCE || 0, 10);
const port = process.env.PORT + nodeInstance;
I am trying to figure out the VSCode launch configuration for launching a gulp task in the inspect mode. The gulp task internally transpiles the code and runs the process via gulp-nodemon. Example is given as below:
VSCode Launch script:
{
"type": "node",
"request": "launch",
"name": "Gulp App debug",
"program": "${workspaceRoot}\\node_modules\\gulp\\bin\\gulp.js",
"args": [
"serve:dev"
],
"sourceMaps": true,
"outputCapture": "std"
}
Gulp Task:
gulp.task("serve:dev", ["release"], () => {
const through = require("through2");
const nodemon = require("gulp-nodemon");
const nodemonStream = nodemon({
script: "dist/src/app.js",
ext: "",
exec: 'node --inspect',
// nodeArgs: ['--inspect'],
watch: "none", // Disabled, will be triggered externally.
env: Object.assign(process.env, {
LOG_LEVEL: "debug",
EXEC_MODE: "dev",
NODE_PATH: "dist/src"
})
});
const watchSrc = watch("./src/**/*", vinyl => {
return compileSourcesStream(
gulp.src(vinyl.path),
gulp.dest("dist/src")
).pipe(
through.obj(function(chunk, enc, cb) {
nodemonStream.emit("restart");
cb(null, chunk);
})
);
});
In the above configuration, the VSCode launches the gulp task with the inspect mode binding to a random port. However the gulp task itself runs the nodemon with the inspect mode pointing to a different new random port. This leads to VSCode unable to track the launched process.
Any suggestion on the above setup ?
Everytime I write something in the logs I get this and the pm2 restart giving me an Service unavailable error in my webpage.
To write the logs I use
winston.info('some info');
and in the app.js I have this:
winston.add(winston.transports.File, { name: 'app-info', maxFiles: 3, filename: 'logs/app-info.log', level: 'info' });
How can avoid the pm2 to restart everytime I am writing in the log?
i solved by ignore the log and watch
watch: 'false',
ignore_watch : [ "*.log"],
so full the code :
module.exports = {
apps : [{
name: "API-XXX",
script: "/var/www/api/api-xxx-xxx/server.js",
env: {
NODE_ENV: "production",
},
env_production: {
NODE_ENV: "production",
},
watch: 'false',
ignore_watch : [ "*.log"],
}],
deploy : {
production : {
user : 'root',
host : 'localhost',
ref : 'origin/master',
repo : 'https://github.com/yogithesymbian/api-xxx-nodejs.git',
path : '/var/www/api/api-xxx-nodejs/',
'pre-deploy-local': '',
'post-deploy' : 'npm install && pm2 reload ecosystem.config.js --env production',
'pre-setup': ''
}
}
};
then run
pm2 restart ecosystem.config.js --env production
i forgot the github disscussion to refer someone said put the line .
You can resolve this problem using --no-autorestart,
example:
pm2 start app.js --no-autorestart