Edit

Share via

Create Kubernetes DaemonSets

  • Article

Applies to: AKS on Azure Stack HCI 22H2, AKS on Windows Server

This article describes how to create and use a Kubernetes DaemonSet in AKS enabled by Azure Arc to ensure that a copy of a pod is always available on every worker node in a cluster. You can use a DaemonSet to improve cluster performance by ensuring that an app runs on all the worker nodes, and to deploy pods that do maintenance and provide support services for nodes.

Overview of DaemonSets

A DaemonSet is a Kubernetes object that ensures a copy of a pod that's defined in the configuration is always available on every worker node in a cluster. When a new node is added to a cluster, the DaemonSet automatically allocates the pod on that node.

Similarly, when a node is deleted, the pod that's running on the node is also deleted and isn't rescheduled on another node (for example, as happens with ReplicaSets). This enables you to overcome Kubernetes scheduling limitations and make sure a specific application is deployed on all nodes within the cluster.

DaemonSets can improve the overall cluster performance. For example, you can use them to deploy pods to perform maintenance tasks and support services to every node:

  • Run a log collection daemon, such as Fluentd and Logstash.
  • Run a node monitoring daemon, such as Prometheus.
  • Run a cluster storage daemon, such as glusterd or ceph.

Although DaemonSets create a pod on every node by default, you can limit the number of acceptable nodes by predefining the node selector field in the YAML file. The DaemonSet controller only creates pods on nodes that match the node selector.

Usually, one DaemonSet deploys one daemon type across all nodes, but it's possible to have multiple DaemonSets control one daemon type by using different labels. A Kubernetes label specifies deployment rules based on the characteristics of individual nodes.

For more information about how to use DaemonSets, see Kubernetes DaemonSets.

Create a DaemonSet

You describe a DaemonSet by using a YAML file, and then create it using the kubectl create or kubectl apply commands (for example, kubectl create –f example-daemon.yaml).

The following example describes the features of a DaemonSet configuration file using an nginx image:

---
apiVersion: apps/v1
kind: DaemonSet
metadata:
  labels:
    app: nginx
  name: example-daemon
spec:
  template:
    metadata: null
    labels:
      app: nginx
containers: null
name: nginx
image: nginx

To view the current state of the DaemonSet, use the kubectl describe command (for example, kubectl describe daemonset example-daemon).

Limit DaemonSet to specific nodes

By default, DaemonSets create pods on all the nodes in a cluster, but with node selectors, you can configure them to create pods only in specific nodes. If you want to limit a DaemonSet to specific nodes, use the kubectl label command.

Update a DaemonSet

You can update a DaemonSet using the kubectl edit ds<NAME> command. However, it's recommended that you edit the original configuration file, and then use the kubectl apply command when it was initially created. After you apply an update, you can view the update status using the kubectl rollout status ds <daemonset-name> command.

Delete a DaemonSet

To remove a DaemonSet, use the kubectl delete command (for example, kubectl delete –f example-daemon.yaml -n monitoring). You should be cautious when specifying the name of the DaemonSet file, as deleting a DaemonSet cleans up all the pods it deployed.

Next steps