Brought up in @kubernetes/sig-apps today, there seems to be a great deal of common patterns between workload controllers (Deployments, DaemonSets, StatefulSets). Namely all three of them in one way or another need:

• upgrades (http://kubernetes.io/docs/user-guide/deployments/#updating-a-deployment, [WiP] DaemonSet updates #31693, Pet set upgrades #28706)
• failure detection (Add docs for perma-failed Deployments website#1699, k8s daemonset, having unexpected behaviour #30206, DaemonSet PODs randomly fail to start on management nodes after reboot #36482)
• the ability to rollback (either manually or automatically after a permanent failure has been detected - Automatic rollback for failed deployments #23211)
• historical info per se (changeset of my app)
• a policy for cleaning up history
• user-extensible strategies (RFC: Support user defined and extensible deployment strategies #14510)

Even more features that could possibly make sense for all of the workload controllers can be found in https://github.com/kubernetes/community/wiki/Roadmap:-Deployment

This issue serves for discussing similarities, differences, what makes sense for all three controllers, and what can we do in order to be consistent across all wherever it is possible.