I am newbie to MEAN stack building MEAN stack development environment for start-up project(web service). Below is a topology where I build up MEAN stack environment.
topology image link: http://147.46.215.116/Picture1.png
Remote PC which developers will use Sublime Text(Code Editor) to code. Git client, NodeJS, its plugins and other stuffs(bower, grunt-client and so on...) are also set on their PC.
I made a Git-hub repository (forked from https://github.com/linnovate/mean.git) for SCM. I also built Jenkins server for CI. Web server is on cloud(Digital ocean).
When I started sample MEAN web service on each machine, It worked like a charm.
After typing following commands, I could see the empty sample page at localhost:3000.
git clone https://github.com/myname/mean.git (which is forked from ~/linnovate/mean.git)
cd mean && npm install
grunt
Here is where my trouble started.
At remote PC, I modified a js source a little bit, and I pushed this to the git-hub repository. It seemed that, by setting .gitignore, MEAN stack didn't let whole project files pushed to git-hub. Instead, it uploaded small part of the project, requiring developers 'npm install' when they git-cloned the project.
One newbie developer git-cloned this project to his remote PC. After 'npm install'ed, however, it didn't work. This is what came up instead.
[4mRunning "clean:0" (clean) task[24m
[4mRunning "jshint:all" (jshint) task[24m
[1m Gruntfile.js[22m
1 |[90m'use strict';[39m
^ Use the function form of "use strict".
9 |[90mmodule.exports = function(grunt) {[39m
^ 'module' is not defined.
11 |[90m if (process.env.NODE_ENV !== 'production') {[39m
^ 'process' is not defined.
12 |[90m require('time-grunt')(grunt);[39m
^ 'require' is not defined.
......
Here's my question.
Is it right to 'npm install' whenever I git-cloned/pulled the project?
How do you guys usually develop web service project w/ MEAN stack and Git?
I wonder whether I misunderstood mean stack development(w/ git) process.
If it is, why doesn't it work?
I also feel confused while using Jenkins. If 'npm install' should happen whenever builds,
Should 'npm install' process occur in Jenkins server (and output files goes to web server)?
or in web server (after just source files are transferred to web server)?
a grass-hopper waits for you guys advice. please help-.
RTFM
How to install MEAN.js http://meanjs.org/docs.html#getting-started
How to deploy via Git http://krisjordan.com/essays/setting-up-push-to-deploy-with-git
If you do now understand how to use Git and MEAN.js what the reason to make things worth and to use Jenkins?
Related
Bit of a n00b question:
I created project with vue-cli using webpack.
On my windows machine I run "npm un dev" and I get a frontend server with HMR and so on.
Now I want to deploy my app to a production machine - ubuntu on DigitalOcean.
What are the steps I must take? I'm not very familiar with the logic of how it's supposed to work. If my ubuntu machine has NODE_ENV set to production, it won't install any of the devDependancies and i'm not able to build anything. So I guess I'll have to change that? If yes then it doesn't make any sense since it's a production machine.
And do I have to create another node/express server to serve index.html? Won't it supposed to work out-of-the-box somehow?
Thanks :)
TL;DR Build on your local machine and everything you need will be outputted in the ./dist/ directory, just copy the contents over to the webroot on your production server and you're good to go.
The webpack template handles most of the stuff for you.
Step you need to take to release:
Run npm run build on your local machine
Copy the contents of the generated ./dist/ directory to your server webroot
That's it!
When you run npm run build, the pre-configured build script sets the node environment to production, and builds with only the stuff that should be in production, it also optimizes the code and removes debug capabilities. When it comes to dependencies webpack takes care of that and includes them in the generated javascript files located in the ./dist/js/, so you need not concern yourself with copying over the node_modules/ directory either.
It also copies over everything in your static directory and src/assets directory to the ./dist/ directory to be prepare for a release. And resolves all the references to the new path generated by webpack.
The production server should not be concerned with building the vue app, run the build command on your local machine to keep dev dependencies away from your production server. I recommend against installing webpack and other dev tools on your production server. It just pollutes the server with things not needed there.
Some development tools could potentially produce alot of issues on production servers. So best practice is to never install them.
You could optionally create your own release script that uses ftp or rsync, whatver you prefer to copy everything in the ./dist/ directory to the production server webroot. This could be a script in bash, if on windows, run it in git bash or something similar for example.
Hope that cleared things up, congrats on your first vue release!
I am attempting to set up a xcode project and bot that will build a nodejs application on commit from a github repository and restart the server after the build completes. The bot is currently picking up on the repository changes but fails to build correctly.
I am using a xcode external build tool project that uses /bin/bash as the tool path and the working directory is set to the local repository path.
The bot's after integration script is something like,
npm install --production
npm run build
npm run server:restart
I am getting errors like [npm|node] is not recognized.
Just looking for some clarity to what I might be missing or what could be going wrong.
Add this to the beginning of your script and review the output:
which node
set | grep PATH
This will happen if node is not in your path, which may happen because build scripts have a pretty basic environment - they're not running as a normal user. You may need to add it to your PATH at the start of your build script.
I've developed an Angular website that lives in a Node-based project. Gulp is used to build sources into development and production versions of the site. The project includes an ExpressJS server that can serve either the dev or prod versions of the files. Now I want to build and deploy the site in CI fashion.
I have a private NPM registry that I'm able to publish the entire project module to. The easy route would be:
In the build environment, check out the project repository
npm install
pass tests
npm publish
In the production environment..
npm install project
build for production
Run the server over the newly built prod files
But this doesn't seem right. Shouldn't I output the production files as part of the build process, and publish these as a versioned artifact with NPM? Are there acceptable, different ways to publish an NPM module for development and deployment?
Or am I stuck building my sources in the production environment? Doesn't this defeat the purpose of NODE_ENV=production?
It depends where you are hosting your server.
If you want to deploy to heroku, you can follow https://devcenter.heroku.com/articles/getting-started-with-nodejs#introduction and install the gulp build task https://github.com/appstack/heroku-buildpack-nodejs-gulp. Basically, you will push the version you want to deploy to heroku's git server and it will automatically trigger the build process for dev or prod and start the express server. Assuming you are already using git for your project, it is only a matter of adding a new remote.
If you want to deploy to Amazon's elastic beanstalk, it's a little bit more complicated but it all boils down to pushing your application code to the service and this will trigger the build process in the host instance.
If you want to deploy it on your environment you could still use git push and git hook to trigger the deployment. Check out How can I automatically deploy my app after a git push ( GitHub and node.js)? for various implementations.
As for your question "Shouldn't I output the production files as part of the build process, and publish these as a versioned artifact with NPM": you could, but this is not what most people do. As long as your build process is repeatable, there is no reason to package and publish the built version of your app. I am assuming here that your are building an application and not a reusable library in which case it would be a different story.
I'm trying to use cute-files on my openshift account
https://www.npmjs.com/package/cute-files
I know node.js works as I did in localhost:3000 but not as expert,
I'm ended up with http://cutefiles-powerupware.rhcloud.com/ that's not I'm looking it. It should be Cute-files app as start.
I did PuTTY to access terminal to do npm-install but no luck due to permission.
The problem is... I cannot understand how openshift works, I need to know how to access console to do npm-install then, execute "$ cute-files", alternativley I've used Git to commit and push, the files are updated...I got no results... I've spent like 6 or 7 hours, Openshift has been very complicated me to use, can anyone explain how to install and execute simple node.js app in openshift?
It's not like you can access Openshift and run any command. Openshift isn't a simple hosting or vps service, it's a PaaS. That means they offer you an environment to run certain services in particular as you demand.
When you create an app it's bound to a git repository which when pushed will be deployed to openshift.
About your application being deployed and runned, it's NOT about node.js running in port 3000. Openshift has some default application file names and port to expect.
So, working in a local clone of such repository you might perform the npm install command, that will download the node_modules folder to be there.
How to do all this?
First you should install rhc in your system.
Manage to clone any of you applications via rhc.
Make changes, and do git.
Those changes may include doing npm install getting your dependencies.
When doing git push, expect the changes to be deployed.
You really should read this guide.
My application is based on node.js, and uses bower.js and others task runners to compile assets and build the actual assets (minify, concat, inject...).
Since this is my first application that will be running in a scalable enviroment on Heroku, I was wondering how is the process of deploying.
I mean, my current workflow is:
cd myRepo
git commit [blabla...]
git push heroku
And when running it, it runs npm run wich calls geddy and runs the server.
If I build before pushing, there will be files that are kind of redundant, but if I push the unbuilt project, it should build it on the cloud. Is that the main idea?
Thanks
it should build it on the cloud. Is that the main idea?
Correct; checkout the Dev Center for more Nodejs information:
Heroku Dev Center: Getting Started with Nodejs
Specifically here, it lists information on bower:
Heroku Node.js Support: Customizing the Build Process