RFC - Automated recovery when E2E fail in production (CI/CD) #48
Replies: 5 comments
-
We could use a blue-green strategy, I personally used it in project a while ago (not a next.js project), it's a robust solution, the problem I see with just rolling back when e2e tests fails is that production will have a bad deployment live for quite some time (the time to finish up e2e).
|
Beta Was this translation helpful? Give feedback.
-
More relevant links: https://martinfowler.com/bliki/BlueGreenDeployment.html |
Beta Was this translation helpful? Give feedback.
-
The design is very interesting, but the implementation is non-trivial, really. Since Vercel only allows one production deployment per project, this would mean it'd switch between 2 different projects (blue/green), there is a lot of stuff behind such automation. And it'd be a very different implementation depending on the hosting provider. |
Beta Was this translation helpful? Give feedback.
-
I think just by using alias would make things easier see: https://spectrum.chat/zeit/now/blue-green-strategy~d1e657ae-761d-4668-9ed6-8f1648cf358e |
Beta Was this translation helpful? Give feedback.
-
Indeed, that may simplify things quite a lot. |
Beta Was this translation helpful? Give feedback.
-
Goal: Discuss desired behaviour about automated rollback to the "previously successful" deployment when a new deployment fails E2E tests in production stage.
Beta Was this translation helpful? Give feedback.
All reactions