After upgrading from 1.14.something then to 1.15.something then to 1.16.13-gke.1.
It started behaving weird. When I delete Deployment replicasets and pods don't get garbage collected.
Also cronjob spams with pods to the point I exceeded a quota and cannot deploy anything anymore even tho I deleted all those pods.
It had 2-4 copies of every system replicasets.
Check events for possible errors. Raise this issue in gcp or your provider. Deployments track replica set and pods by labels and selectors. May be those got changed somehow.