Exercise

In this exercise, you will manipulate a DaemonSet.

In the log-ds.yaml file, create a DaemonSet specification with the following characteristics:

the DaemonSet has the name log
each Pod has a single container based on the alpine image
each container has access to the /var/log/ directory on the filesystem of the node on which it runs, this directory being mounted at the location /var/log/node in the container’s filesystem
each container writes the content of *.log files (located in /var/log/node) to its standard output

Create the DaemonSet defined in this specification and look at the logs of the created Pods.

Note: you can use the following documentation for help: https://kubernetes.io/docs/concepts/workloads/controllers/daemonset/.

Then delete the DaemonSet.

Solution

The following specification defines the DaemonSet log where each Pod has a single container based on the alpine image:

apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: log
spec:
  selector:
    matchLabels:
      app: log
  template:
    metadata:
      labels:
        app: log
    spec:
      containers:
      - name: log
        image: alpine

⚠️

We don’t launch this DaemonSet yet, it would result in an error as the alpine container would continuously restart.

We add the definition of a hostPath volume to access the /var/log directory of the host machine and we make its content available in the /var/log/node directory of the container:

apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: log
spec:
  selector:
    matchLabels:
      app: log
  template:
    metadata:
      labels:
        app: log
    spec:
      containers:
      - name: log
        image: alpine
        volumeMounts:
        - name: varlog
          mountPath: /var/log/node
      volumes:
      - name: varlog
        hostPath:
          path: /var/log

We add a command to make the container write the content of *.log files, present in the /var/log/nodes directory, to its standard output:

apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: log
spec:
  selector:
    matchLabels:
      app: log
  template:
    metadata:
      labels:
        app: log
    spec:
      containers:
      - name: log
        image: alpine
        command: ["/bin/sh", "-c"]
        args: ["tail -f /var/log/node/*.log"]
        volumeMounts:
        - name: varlog
          mountPath: /var/log/node
      volumes:
      - name: varlog
        hostPath:
          path: /var/log

Once saved in the log-ds.yaml file, we can create this DaemonSet:

kubectl apply -f log-ds.yaml

We list the associated Pods:

kubectl get pods -l app=log

We can then see the logs of one of the Pods associated with the DaemonSet, which displays the content of the log files existing on the Node the Pod is running on.

kubectl logs ds/log

In a production context, a DaemonSet could be used to deploy Pods that will read logs from each of the nodes and send them to a centralized log management solution (such as ElasticSearch, Splunk, …)

We then delete the DaemonSet

kubectl delete ds log