DaemonSets

Welcome to this guide on DaemonSets in Kubernetes. In this tutorial, you'll learn how DaemonSets work, their common use cases, and how to create one.

DaemonSets enable you to run exactly one instance of a Pod on every node within your cluster. As your cluster scales—by adding or removing nodes—the DaemonSet automatically ensures that each node has the designated Pod running. This approach is particularly useful for deploying essential services like monitoring agents, log collectors, and networking components (for example, kube-proxy) consistently across all nodes.

Use Cases for DaemonSets

DaemonSets are primarily used in the following scenarios:

• Monitoring and Logging: Deploy agents responsible for system monitoring and log collection across all nodes.
• Networking: Ensure a networking solution agent (e.g., VNet components or weave-net) is deployed on every node.
• Critical Infrastructure Components: Deploy essential components like kube-proxy that need to reside on every node.

Below is an image that illustrates the use case for DaemonSets, highlighting the connection between a monitoring solution, a logs viewer, and multiple nodes:

Another common scenario involves networking. As mentioned, some networking solutions require an agent on every node. Understanding this use case is vital before diving deeper into networking concepts later in the course:

Creating a DaemonSet

Creating a DaemonSet is quite similar to creating a ReplicaSet. The YAML definition file begins with apiVersion, kind, metadata, and spec sections. The primary difference is that the kind is set to DaemonSet, and it manages a Pod on every node rather than a specified number of replicas.

Below is an example DaemonSet definition file:

# daemon-set-definition.yaml
apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: monitoring-daemon
spec:
  selector:
    matchLabels:
      app: monitoring-agent
  template:
    metadata:
      labels:
        app: monitoring-agent
    spec:
      containers:
      - name: monitoring-agent
        image: monitoring-agent

For reference, here’s a similar ReplicaSet definition that deploys the same monitoring agent. Notice that the structure is nearly identical except for the value of kind:

# replicaset-definition.yaml
apiVersion: apps/v1
kind: ReplicaSet
metadata:
  name: monitoring-daemon
spec:
  selector:
    matchLabels:
      app: monitoring-agent
  template:
    metadata:
      labels:
        app: monitoring-agent
    spec:
      containers:
      - name: monitoring-agent
        image: monitoring-agent

Ensure that the labels in the selector match those in the Pod template to guarantee proper functioning. Once your DaemonSet definition is ready, create it using the following kubectl command:

kubectl apply -f daemon-set-definition.yaml

After applying the YAML file, verify your DaemonSet with:

kubectl get daemonsets

The console output might look similar to this:

NAME                DESIRED   CURRENT   READY   UP-TO-DATE   AVAILABLE   AGE
monitoring-daemon   1         1         1       1            1           41

How DaemonSets Work

DaemonSets automatically schedule Pods on every node in your cluster. In earlier Kubernetes versions, the nodeName property was used to assign Pods directly to nodes, bypassing the scheduler. However, since version 1.12, DaemonSets leverage the default scheduler and node affinity rules to manage Pod placement.

To summarize, a DaemonSet ensures that a specific Pod is running on all nodes in your cluster—making it ideal for running services such as:

• Monitoring
• Logging
• Networking

Happy clustering!