CPU Manager

Platform9 Managed Kubernetes - PMK release 5.0 includes support for Kubernetes Topology Manager and CPU Manager.

Traditionally, Kubernetes has abstracted details of pod scheduling away from the user. While this is desirable in most scenarios, there are workloads that require specific guarantees in terms of latency and performance. CPU Manager is a feature on kubelet that enables you to specify requirements of CPU allocations.

Topology Manager

Topology Manager integrates the data from CPU Manager as well as other device plugins when making decisions about pod placement. Without topology manager, CPU Manager and Device Manager make decisions independently and this can result in pod placements that are undesirable.

Topology Manager with Platform9 Managed Kubernetes (PMK)

PMK now supports configuration of Topology Manager. The following fields are added to the cluster create API of qbert: \

cpuManagerPolicy: {type: string, description: "options: none, static; default: none"},

topologyManagerPolicy: {type: string, description: "options: none, best-effort, restricted, single-numa-node; default: none"},

reservedCPUs: {type: string, description: "list of CPUs to be reserved for the system. Ex: 1-3,5"}

"cpuManagerPolicy" : "static" turns on CPU Manager. This is required for Topology Manager to work.

"topologyManagerPolicy" supports the following options: \

• none: Topology Manager is turned off
• best-effort: Kubelet uses hint providers to determine the best placement location for a pod. The pod is scheduled even if all affinity requirements are not met.
• restricted: Kubelet uses hint providers to determine the best placement location for a pod. The pod is terminated if such a node matching all affinity requirements are not met.
• single-numa-node: Kubelet uses hint providers to find a single numa node that meets all affinity requirements. The pod is scheduled if such a node exists, it is moved to a terminated state otherwise. \

You can read more about these policies here.

"reservedCPUs" is a list of CPUs reserved for general purpose system use. Kubelet will not schedule pods to run on these CPUs.