I've spent countless hours trying to understand how to do this correctly.
I have a nodejs application with angular front-end which are both contained in the same project.
I would simply like to have Gitlab CE CI build the project and then copy the resulting dist folder and package.json file to the production server and restart
I have a shared gitlab runner setup and was able to successfully configure the ssh runner.
Using the Gitlab runner ssh I was able to copy the entire project to the production server but cannot get it to build (plus I really don't want to have all files on the server, just the production required files.
what am I missing. do you use a docker runner with a node image to build the project and they scp files to the production server?
Any guidance would greatly be appreciated.
Yes, I think you already state the solution.
I would recommend looking into Job Artifacts to build your application in a Docker-based build job and upload the production files as an artifact.
The next job could then deploy (by scp or ssh+wget/curl) the artifact files to your production server and restart the webserver.
Related
i'm already thankfull for anyone reading this.
I've stumbled across some problems trying to deploy a Nuxt application in a correct way. For testing purposes i've created a clean installed Nuxt application so i'm sure nothing is wrong with my codebase. What my ultimate goal is is to push all my nuxt code to github which than gets picked up from a azure deployment pipeline and generates the needed build files and drops them to the webapp service and runs the needed start command. There's no documentation to be find about Azure which is really annoying for me. I'm not used to deploying stuff through a pipeline but need this to be working for this project.
Has anyone experience with Nuxt and deployments through Github -> Azure pipeline -> Build -> Web app running immediatly
What's i've tried already is pushing all the source code to a repository which get's picked up by azure pipeline. The only thing the pipeline does is paste the code in the wwwroot folder. Which is obviously not enough to make the application run automated.
What i expect is some insight from someone who is experienced with nuxt and azure deployments through github/bitbucket (doesn't really make a difference)
You have to configure your pipeline to build the nuxt files and deploy them. You can usually achieve the first step through npm commands, npm install then npm run build.
There's a few different ways to deploy them, which you'll have to decide between depending on your use case and environment. You can create an endpoint pointing to a CDN serving a blob, zip the built files and publish the artifact which is then used by a release pipeline, etc. Once the next files are created, you have all you need to deploy.
I'm looking into using Nuxt.js for a new Web application with an Express server / API attached in the project itself. I've created the Nuxt.js application with the npx create-nuxt-app command and selected Express as "custom server framework".
When using the development environment (npm run dev), it works like a charm. Now, I want to deploy the application to a server which is running pm2 using Gitlab CI. I struggle with the right way to deploy the application, as I don't think I should "upload" the sourcecode to the server and then run the neccessary commands like npm run build or npm start. So, what is actually "the right way" to deploy this application with the server?
I have SSH access to a server where I want to deploy this application & API. The server is running pm2 for managing the Node processes.
Furthermore, the sourcecode is in a repository on a Gitlab server with a Gitlab CI Runner available. The current gitlab-ci.yml does the following:
npm install
npm run build
lftp the dist folder to the server
With a SPA, only the client is built and the server / API is not. We now have a, in my opinion, dirty workaround which doesn't work very well that does the following:
npm install
npm run build
mirror the whole project to the server
use a watcher in pm2 to detect changes in files
build project again???
It looks like it doesn't work as intended and I have the feeling we are creating problems instead of solving the deployment issue the right way. We don't use Heroku, Google or any other services for hosting the application, it's just a webhosting with SSH access basically.
I am having self-hosted GitLab-EE. I want to enable CI / CD. I already have gitlab-runner on windows and it is activated for my project.
I also have custom server for hosting my ASP.NET Core 2.0 with Angular 5.0 application.
My final idea is when commit is made to GitLab, build and deploy (to custom server) is executed. Deploy path would be different if commit to master was made or any other branch (from merge request) conditional by Git Tag.
Almost all tutorials use docker, but I couldn't found why? What are the cons using docker?
I thought I only need msbuild (I can also install Visual Studio) on machine where GitLab-runner is running. It would build and deploy application. I found this configuration file which doesn't use docker, but question remains. Do I need docker and why?
A docker image would contain all the msbuild dependencies and build tools needed to build your application with out you going through the trouble of manually installing them on your server.
So basically docker helps you to manage dependencies of you application more efficiently.
We are developing a mobile app using IBM MobileFirst platform.
We compile the code using the command line tool with the command 'mfp build' and deploy it using 'mfp deploy', and we are able to preview the application from the URL mentioned below: http://localhost:xxxxx/worklightconsole/index.html
After doing 'mfp build' I get these 6 files:
mobapp.war
mobAppAdapter.adapter
Dashboard-all.wlapp
Dashboard-common.wlapp
Dashboard-desktopbrowser-1.0.wlapp
Dashboard-ipad-1.2.wlapp
Questions:
Where can I find the 6 files in 'worklight console'? (or)
Which other URL I have to refer to verify whether the files have been deployed correctly or not?
Because when I type in 'mfp build' it deploys the files but we are not sure where its getting deployed. We don't have WAS installed instead 'liberty' is used to our knowledge
The MobileFirst Platform CLI tool contains an embedded WebSphere Liberty profile server inside of it. It is this server and its internal database (during development time(!)), that the artifacts you have mentioned get deployed to.
You create a project ("mfp create").
You then start the server ("mfp start").
You then create your applications, develop them, etc...
Eventually you build and deploy it to the server ("mfp build", "mfp deploy").
The resulting files of the build command are the .adapter and .wlapp files.
These get deployed using the deploy command to the server's database.
You then see these files in the console using "mfp console".
The .war file is your project's runtime, containing metadata required for various server operations. This file is part of the server and handled automatically by the underlying tooling. This is invisible to you during development.
Is there a way in glassfish v2 asadmin utility to copy jar files into the server lib or the domain's lib? I want to do a remote deployment using hudson on a different box so i have a script that executes a bunch of asadmin commands.
asadmin deploy moves your deployable code to the autodeploy directory making glassfish aware of its existence to deploy. I believe copying a simple library out there won't net any effect without proper deployment descriptors. In any event I don't think the asadmin deploy would have access to your domain/lib or server/lib directory but rather only your application directory structure itself.
To do what you are trying to achieve, it sounds more like writing a simple ant script.