Currently the cluster autoscaler carries many patches that are not in upstream. Some of these are intentional and OpenShift specific, but others could be contributed to upstream to clean up our repository.
The cluster autoscaler carry patches should be reviewed, and squashed where possible, or contributed upstream. The scale from zero patches are known to be carried by us, but the others should be evaluated.

Based on a conversation with agarcial@redhat.com, these are some high priority commits to investigate:
https://github.com/openshift/kubernetes-autoscaler/commit/a43c8c2f88313eb3a719983e2fa2c2e4f0f4d47c
https://github.com/openshift/kubernetes-autoscaler/commit/38c7e5102be3b731b5f93bce56e7755684c7c179
https://github.com/openshift/kubernetes-autoscaler/commit/32f60826c10977447a0ff0df441038f71079600d
https://github.com/openshift/kubernetes-autoscaler/commit/457274d866b47591fe8f82d9aeae435247de47d6

these commits could be squashed together
https://github.com/openshift/kubernetes-autoscaler/commit/e3cc2534024707703a37d5311bb33b9988d989af
https://github.com/openshift/kubernetes-autoscaler/commit/3b9c2c9bcb50d372320b6e1ac21b5c303dad4fd7

DoD

• contribute what we can to upstream
• squash similar commits that carry openshift specific data (eg dockerfiles, owners)