3. Scaling and Scheduling Pods in Kubernetes: All About DaemonSets

What is a DaemonSet in Kubernetes?

Think of a DaemonSet as a dedicated delivery system that ensures every stop on a route gets exactly what it needs. In Kubernetes, it guarantees that a specific pod runs on every node in your cluster (or a subset of nodes, if you prefer).

---

Let’s Break it Down with an Example:

Imagine you’re managing a pizza chain. Each branch (node) needs:

• A pizza oven (to make pizza).

• A cash register system (to handle payments).

Now, to ensure smooth operations:

1. You deploy one pizza oven and one cash register at every branch.

2. Anytime a new branch opens, it automatically gets these essentials.

3. If a branch closes, the pizza oven and cash register are removed.

In Kubernetes, the pizza oven and cash register are like the pods deployed by a DaemonSet. It ensures that all nodes (branches) have these tools running consistently.

---

Real-Life Tech Analogy:

Imagine you’re running a monitoring or logging service (like Prometheus Node Exporter or Fluentd). You want every machine in your data center (node in your cluster) to:

1. Track system metrics like CPU usage or memory.

2. Send logs for analysis.

A DaemonSet ensures that these monitoring tools run on every machine, no matter how many machines are added or removed. It's your "set-it-and-forget-it" solution.

---

Why is it Cool?

• Automation: New nodes automatically get the pods they need.

• Efficiency: You don’t have to manually deploy monitoring/logging services on each node.

• Consistency: All nodes are equipped with the same tools.

---

Why DaemonSet is Beneficial:

1. No Manual Placement: You don’t need to say, "Place this pod on every node." DaemonSet handles it.

2. Dynamic Scaling: If a new node joins the cluster, DaemonSet automatically deploys the pod there. No extra config is required.

3. Uniformity: Ensures every node has the same service (like monitoring or logging), making it reliable for tasks that must run on all nodes.

4. No Overlapping Pods: Unlike ReplicaSets, which might run multiple pods on one node, DaemonSet ensures one pod per node for its task.

---

Steps to use DaemonSets:

Please refer to this blog for creating our clusters, and then follow the steps below:

1. Create daemonset-deploy.yaml file.

apiVersion: apps/v1
kind:  DaemonSet
metadata:
  name: nginx-daemonset
  labels:
    app : nginx
spec:
  template:
    metadata:
      labels:
        app: nginx
      name: nginx
    spec:
      containers:
      - image: nginx
        name: nginx
        ports:
        - containerPort: 80
  selector:
    matchLabels:
      app: nginx

This daemonset-deploy.yaml file is similar to our deployment.yml file, doesn't it? The main difference is the kind type. By using DaemonSet as the kind type, it assigns each pod to an individual node.

2. Now apply the changes:

kubectl apply -f daemonset-deploy.yaml

Although we have 2 worker nodes and the file specifies creating only one pod, running the command kubectl get ds will show 2 pods. This is because DaemonSet automatically assigns one pod to each node.

3. Now you can delete daemonset-deploy.yml and load-generator pod:

 kubectl delete  pod load-generator
kubectl delete -f daemonset-deploy.yaml

This is how a DaemonSet operates.

---

Happy Learning :)

Chetan Mohod ✨

For more DevOps updates, you can follow me on LinkedIn.