Concepts

Kubernetes v1.17 documentation is no longer actively maintained. The version you are currently viewing is a static snapshot. For up-to-date documentation, see the latest version.

Edit This Page

Scheduler Performance Tuning

FEATURE STATE: Kubernetes 1.14 beta
This feature is currently in a beta state, meaning:

  • The version names contain beta (e.g. v2beta3).
  • Code is well tested. Enabling the feature is considered safe. Enabled by default.
  • Support for the overall feature will not be dropped, though details may change.
  • The schema and/or semantics of objects may change in incompatible ways in a subsequent beta or stable release. When this happens, we will provide instructions for migrating to the next version. This may require deleting, editing, and re-creating API objects. The editing process may require some thought. This may require downtime for applications that rely on the feature.
  • Recommended for only non-business-critical uses because of potential for incompatible changes in subsequent releases. If you have multiple clusters that can be upgraded independently, you may be able to relax this restriction.
  • Please do try our beta features and give feedback on them! After they exit beta, it may not be practical for us to make more changes.

kube-scheduler is the Kubernetes default scheduler. It is responsible for placement of Pods on Nodes in a cluster.

Nodes in a cluster that meet the scheduling requirements of a Pod are called feasible Nodes for the Pod. The scheduler finds feasible Nodes for a Pod and then runs a set of functions to score the feasible Nodes, picking a Node with the highest score among the feasible ones to run the Pod. The scheduler then notifies the API server about this decision in a process called Binding.

This page explains performance tuning optimizations that are relevant for large Kubernetes clusters.

Percentage of Nodes to Score

Before Kubernetes 1.12, Kube-scheduler used to check the feasibility of all nodes in a cluster and then scored the feasible ones. Kubernetes 1.12 added a new feature that allows the scheduler to stop looking for more feasible nodes once it finds a certain number of them. This improves the scheduler’s performance in large clusters. The number is specified as a percentage of the cluster size. The percentage can be controlled by a configuration option called percentageOfNodesToScore. The range should be between 1 and 100. Larger values are considered as 100%. Zero is equivalent to not providing the config option. Kubernetes 1.14 has logic to find the percentage of nodes to score based on the size of the cluster if it is not specified in the configuration. It uses a linear formula which yields 50% for a 100-node cluster. The formula yields 10% for a 5000-node cluster. The lower bound for the automatic value is 5%. In other words, the scheduler always scores at least 5% of the cluster no matter how large the cluster is, unless the user provides the config option with a value smaller than 5.

Below is an example configuration that sets percentageOfNodesToScore to 50%.

apiVersion: kubescheduler.config.k8s.io/v1alpha1
kind: KubeSchedulerConfiguration
algorithmSource:
  provider: DefaultProvider

...

percentageOfNodesToScore: 50
Note: In clusters with less than 50 feasible nodes, the scheduler still checks all the nodes, simply because there are not enough feasible nodes to stop the scheduler’s search early.

To disable this feature, you can set percentageOfNodesToScore to 100.

Tuning percentageOfNodesToScore

percentageOfNodesToScore must be a value between 1 and 100 with the default value being calculated based on the cluster size. There is also a hardcoded minimum value of 50 nodes. This means that changing this option to lower values in clusters with several hundred nodes will not have much impact on the number of feasible nodes that the scheduler tries to find. This is intentional as this option is unlikely to improve performance noticeably in smaller clusters. In large clusters with over a 1000 nodes setting this value to lower numbers may show a noticeable performance improvement.

An important note to consider when setting this value is that when a smaller number of nodes in a cluster are checked for feasibility, some nodes are not sent to be scored for a given Pod. As a result, a Node which could possibly score a higher value for running the given Pod might not even be passed to the scoring phase. This would result in a less than ideal placement of the Pod. For this reason, the value should not be set to very low percentages. A general rule of thumb is to never set the value to anything lower than 10. Lower values should be used only when the scheduler’s throughput is critical for your application and the score of nodes is not important. In other words, you prefer to run the Pod on any Node as long as it is feasible.

If your cluster has several hundred Nodes or fewer, we do not recommend lowering the default value of this configuration option. It is unlikely to improve the scheduler’s performance significantly.

How the scheduler iterates over Nodes

This section is intended for those who want to understand the internal details of this feature.

In order to give all the Nodes in a cluster a fair chance of being considered for running Pods, the scheduler iterates over the nodes in a round robin fashion. You can imagine that Nodes are in an array. The scheduler starts from the start of the array and checks feasibility of the nodes until it finds enough Nodes as specified by percentageOfNodesToScore. For the next Pod, the scheduler continues from the point in the Node array that it stopped at when checking feasibility of Nodes for the previous Pod.

If Nodes are in multiple zones, the scheduler iterates over Nodes in various zones to ensure that Nodes from different zones are considered in the feasibility checks. As an example, consider six nodes in two zones:

Zone 1: Node 1, Node 2, Node 3, Node 4
Zone 2: Node 5, Node 6

The Scheduler evaluates feasibility of the nodes in this order:

Node 1, Node 5, Node 2, Node 6, Node 3, Node 4

After going over all the Nodes, it goes back to Node 1.

Feedback