Priorities of Pods in Google Container Cluster

Question

Is it possible to schedule upcoming Pods/Containers based on their priorities? (if container1 is critical and needs resources, the google orchestrator can kill other low priority containers) If yes, is there some specific priorities tags (like: critical, monitoring, production...)?

Answer 1

This use case is described in both the Borg paper and the Omega paper . However, it is not presently implemented within Kubernetes. Here are some related links to ongoing proposals:

• QoS Tiers
• Preemption Policy / Scheme
• Resource Quality of Service