giantswarm/aws-operator

Update logic for cloudconfig in S3 bucket

rossf7 opened this issue · 3 comments

rossf7 commented

Towards giantswarm/giantswarm#2023

Currently we update the cloud config in the S3 bucket when clusters are reprocessed. This can cause problems if breaking changes are deployed.

  • In k8scloudconfig we want to use semver to identify breaking / non breaking changes.
  • We also need to version the cloudconfig stored in the S3 bucket.
  • For non breaking changes we should update the existing cloud config and roll the nodes to apply the changes.
  • For breaking changes a new version should be created in the S3 bucket and the cluster should not be updated.

Cloud Config operator

  • Until the cloud config operator is in place aws-operator will need to determine which k8scloudconfig version to use.
  • Once the cloud config operator is in place it will generate the cloud config but aws-operator will still need to upload it to S3.
xh3b4sd commented

@rossf7 please have a look again and sort the issue from an organizational point of view. <3

rossf7 commented

Moving this to Team Batman as its part of the AWS updates story.

rossf7 commented

Closing. This is ancient and superseded by work done on the updates story.