Can anyone share how they setup their CI/CD pipeline with payload

techinverted2 years ago

2

• 

  chris_toxicness2 years ago

  Did you made project dockerized?

• 

  hades2000822 years ago

  In my most recent deployment, I used DevOps Pipelines deploying to Azure App Services with a MongoDB Atlas instance from the Azure Marketplace.

  
  
  

  I created a new Dockerfile for each environment (dev, qa, uat, production).

  
  
  

  In each Dockerfile, I added a line before the npm install

  ENV PAYLOAD_PUBLIC_SERVER_URL=https://myserver.com

  (different URL in each reflecting the environment's URL.

  
  
  

  In my Payload Config, I use the environment variable to set the

  serverUrl

  property.

  
  
  

  Note: This could also have been achieved using .env.<environment-name> files and a single Dockerfile.
  

  
  

  In my DevOps

  azure-pipelines.yml

  file I use variables to determine which Dockerfile to use depending on which git branch triggered the pipeline.

  
  
  

  I build and publish the relevant Dockerfile to an Azure Container Registry.

  
  
  

  The Azure App Services were created as Docker app services and are deployed from the Azure Container Registry, re-deploying via webhook whenever the associated docker image is updated.

  
  
  

  I found this was the only way to get it working in Azure App Services due to the fact that the

  PAYLOAD_PUBLIC_*

  variables need to be present at

  build time

  so they are passed to the admin app, and a bug in Kudu ZipDeploy that breaks

  node_modules/.bin

  symlinks preventing any kind of

  npm build

  on the app service itself. (

  npm rebuild

  also doesn't work on app services due to file permissions)

  
  
  

  I've opened a discussion at

  https://github.com/payloadcms/payload/discussions/2288

  to talk about improving the Payload documentation in regards to production deployments and CI/CD.