Kubernates notes Manual scheduling

Use of binding schedular

In Kubernetes (K8s), ==schedulers are responsible for assigning pods to the most appropriate nodes in the cluster==. The default scheduler is built-in, but you can also implement custom schedulers, including a manual scheduler. Here’s an overview of both:

1. Manual Scheduler:

• Definition: As the name suggests, the user manually selects a node for the pod to run on.

• How: Specify the node name in the pod spec (nodeName field).

• Use Case: Useful in testing or small setups where you want full control over node selection.

• YAML Example:

  apiVersion: v1
  kind: Pod
  metadata:
    name: manual-pod
  spec:
    containers:
    - name: nginx
      image: nginx
    nodeName: worker-node1

• Drawback: Doesn’t scale well in large clusters, prone to human error.

2. Default Scheduler:

• Definition: The built-in Kubernetes scheduler that automatically assigns pods to nodes based on resource requests and other constraints (e.g., CPU, memory).
• How: Automatically selects nodes based on available resources and other policies.
• Use Case: Most common, handles most scheduling requirements in production.
• Factors Considered:
  • Resource requests and limits (CPU, memory).
  • Node availability.
  • Taints and tolerations.
  • Affinity and anti-affinity rules.
  • Node selectors.

3. Custom Scheduler:

• Definition: Custom schedulers allow you to create specialized scheduling logic using your algorithms.
• How: You define custom logic to assign pods to nodes (often using a custom controller).
• Use Case: Needed in very specific use cases where default logic doesn’t suffice (e.g., custom resource optimization).
• Example: Scheduling for GPUs or specific latency-sensitive workloads.

4. Multiple Schedulers:

• Definition: You can deploy and configure multiple schedulers in the same cluster.

• How: Pods can use the schedulerName field to specify which scheduler to use.

• Use Case: Useful if different teams or workloads require different scheduling strategies.

• YAML Example:

  apiVersion: v1
  kind: Pod
  metadata:
    name: custom-scheduler-pod
  spec:
    schedulerName: my-custom-scheduler
    containers:
    - name: nginx
      image: nginx

Key Points for CKA:

• Understand how the default scheduler works and the factors affecting it.
• Practice using manual scheduling and node selectors.
• Be familiar with how to implement custom schedulers using schedulerName.

Real-world Use Cases:

• Default scheduler handles most cases, scaling across multiple nodes.
• Custom schedulers are used for specific needs (e.g., workload balancing, latency-sensitive apps).

Tips:

• Use node taints and tolerations to control pod placement effectively.
• Use node affinity/anti-affinity to manage pod distribution across nodes.
• Avoid manual scheduling in large environments to minimize human errors.

Let me know if you’d like more detail on any of these!