Continuous Deployment (Integrations for GitHub, Heroku, ...)

[marktani]

Issue by marktani Thursday Sep 28, 2017 at 09:55 GMT Originally opened as https://github.com/graphcool/prisma/issues/609

---

What feature are you missing?

Version 0.4 opens up new workflows for continuous integration and deployment. However, this needs to be setup manually and requires some configuration and wiring overhead.

I'd like to kickstart a discussion about a seamless integration that allows you to connect a Graphcool project with a Github project, so that new pushes to the repo will initialize a new deployment.

How could this feature look like in detail? Tradeoffs?

Components

• Ability to connect Github repo + Graphcool project
• Ability to deploy any commit from the repo to the Graphcool project
• Integration with CI, so a pre-deploy script can be executed before deploying, for testing etc.
• A migrations concept (also see https://github.com/graphcool/graphcool/issues/627)
• Multi stage deployments (dev, staging, prod).

Prior Art

Github integration, Continuous Deployment, Multi stage deployments

• Heroku
• Netlify
• Serverless

[marktani]

Comment by kbrandwijk Thursday Sep 28, 2017 at 10:18 GMT

---

My first 5 cents on this:

The CI part for me would be:

• Doing a dry-run deploy (see the other FR for that)
• Executing tests

The deploy part for me would be:

• Having a Graphcool integration app on Github. This could be automatically setup using the cli like other Github integrations do (so you don't have to manually setup the webhooks on GH etc.)
• I really like the production/preview/branch concept from netlify
• I like the 'wait for CI to complete' setting on Heroku

Challenges:

• Authentication
• Configuring projectIds for the different environments without committing .graphcoolrc file to GH
• Using environment variables/secrets without committing them to GH
• Seeding (test data setup/production data setup)