I have deployed codepipeline and codedeploy to take my code from a zip file in S3 and to deploy it into an EC2 autoscale farm. An external process updates EC2, CWE then triggers the pipeline.
When it works it works fine. But if there is ever a failure to deploy onto EC2 eg, an error with the application stop script, the latest revision appears to be lost. If an autoscale event is triggered the previous revision is deployed. It would appear that a revision is only marked as good once a successful deployment has occured. This makes it very difficult to fix any issues as I have to rely on the old revision deploying successfully before I can release the new. I would prefer that I could fix the code, deploy/trigger a new revision, then terminate all machines and let them build now.
Is there a codepipeline step I can use that will create a new application revision without having to deploy it to EC2 for it to be made active?
AWS OFICIALAtualizada há 2 anos