Kube Scheduler

kube-scheduler is a critical component of Kubernetes that ensures the optimal placement of pods on nodes within a cluster, considering various constraints and factors to maintain application performance, resource utilization, and high availability. It plays a pivotal role in automating workload distribution in a Kubernetes environment.

Responsibilities

  1. Pod Scheduling Request: When a user or system administrator creates a pod deployment, replication controller, or other workload with a desired number of replicas, Kubernetes sends a pod scheduling request to the kube-scheduler.

  2. Node Selection: The kube-scheduler evaluates various factors and constraints to select the most suitable node for the pod. These factors and constraints include:

    • Resource Requirements: It considers the CPU and memory resource requests and limits defined in the pod’s configuration to ensure that the node has enough resources available.

    • Node Affinity/Anti-Affinity: Node affinity rules allow you to specify conditions that must be met for a pod to be scheduled on a particular node. Node anti-affinity rules prevent pods from being scheduled on nodes that already have pods that match certain criteria.

    • Pod Affinity/Anti-Affinity: Similar to node affinity, these rules define how pods should be placed in relation to other pods.

    • Taints and Tolerations: Nodes can be “tainted” to repel certain pods unless those pods have specific “tolerations” that allow them to run on tainted nodes.

    • Resource Constraints: The scheduler considers node-specific resource constraints, such as the amount of available CPU and memory on each node.

    • Pod Priority: Pods can be assigned priorities, which influence their scheduling order. Higher-priority pods are scheduled before lower-priority pods.

    • Inter-Pod Spreading: To enhance application availability, the scheduler can be configured to spread pods across different nodes to avoid single points of failure.

  3. Node Scoring: The scheduler assigns a score to each node based on the criteria mentioned above. Nodes with higher scores are better candidates for the pod.

  4. Node Selection: The scheduler selects the node with the highest score as the destination for the pod.

  5. Binding: Once the scheduler has selected a node, it updates the Kubernetes API server with the binding information, indicating which node the pod should be scheduled on.

  6. Pod Placement: The Kubernetes control plane, including the kubelet on the selected node, then takes care of placing the pod’s containers onto the chosen node.

  7. Monitoring: The kube-scheduler continually monitors the cluster’s state and reschedules pods if necessary. For example, if a node goes offline or becomes unsuitable due to resource constraints, the scheduler can relocate pods to other available nodes.

Commands 

# VView kube-scheduler options (kubeadm installation)
cat /etc/kubernetes/manifests/kube-scheduler.yaml

# View api-server options by listing the process on the master node and searching for kube-apiserver
ps -aux | grep kube-scheduler
Edit this page on GitHub
← Binding Pods to Nodes
Kubelet →