Private Cloud Platform Team - 14-Day Image Pruning Issue

[LolandaE]

Describe the task To determine path forward to image management.

Purpose So that in future, there is an automated process to save/move/back up images located in applications supported by the Sustainment Team before they are pruned.

Acceptance Criteria

• [ ] Meet and develop strategy to address deleted images based on info gathered
• [ ] Devs to watch DevOps Alerts RocketChat channel daily
• [ ] Devs to subscribe to platform services updates newsletter

Additional context

• More than 500 images were removed as a result of the image pruning
• Per Feb 7th email from Steve Barre: The 14 day pruning was a one time thing as part of this round of upgrades. So it shouldn’t cause ongoing issues once they deal with the initial fallout. It was to clear out unused images and reduce the size of the ETCD database. If we were to run it again, there isn’t a way to exclude someone.

 is it deleting image tags that could be restored or is it deleting images completely

We didn’t take any backups as the image registry is over a terabyte in size.

Only images that are not referenced by an object in openshift were pruned. So this should have no impact on images deployed to production.

Where this seems to have been biting people is builds where the image is referenced from the Dockerfile. So OCP has no idea of the connection between the build and the image.

[jakemorr]

rebuilding imgaes as needed and storing images that we rebuild in artifactory