Run SolarWinds Observability Agent on the Kubernetes platform

You can use the following options to run the Agent on the Kubernetes platform.

When you already have the Agent running, specify what you want to monitor. See Add entities for monitoring .

Run as a Kubernetes deployment

When running the Agent as a Kubernetes deployment, you can leverage Kubernetes secrets to hide the access token.

To create a secret for an access token, run the following command:

kubectl create secret generic access-token --from-literal=ACCESS_TOKEN="<token>"

To persist the Agent working directory between pod restarts, use a persistent volume. See Configure a Pod to Use a PersistentVolume for Storage in Kubernetes documentation.

For example, you can use the following file to create PersistentVolumeClaim that will be referenced later in the Deployment definition:

apiVersion: v1 
kind: PersistentVolumeClaim 
metadata: 
  name: uamsclient-pv-claim 
  labels: 
    app: router 
spec: 
  accessModes: 
    - ReadWriteOnce 
  resources: 
    requests: 
      storage: 1Gi

To apply the Agent as a Kubernetes deployment in your cluster, use the following file.

The restartPolicy setting must be set to OnFailure so that the Agent is able to perform restart procedures sent from the SolarWinds Observability SaaS.

apiVersion: apps/v1 
kind: Deployment 
metadata: 
  name: uamsclient-deployment 
  namespace: <namespace> 
  labels: 
    k8s-app: uamsclient 
spec: 
  replicas: 1 
  selector: 
    matchLabels: 
      app: uamsclient 
  template: 
    metadata: 
      labels: 
        app: uamsclient 
    spec:
      restartPolicy: OnFailure
      containers: 
      - name: uamsclient 
        image: solarwinds/swo-agent:latest 
        env: 
          - name: UAMS_ACCESS_TOKEN 
            value: <token> 
          - name: SWO_URL 
            value: <cluster endpoint> 
          - name: UAMS_ACCESS_TOKEN 
            valueFrom: 
              secretKeyRef: 
                name: access-token 
                key: ACCESS_TOKEN 
        volumeMounts: 
          - name: uamsclient-persistent-storage 
            mountPath: /uamsclient/workdir 
      volumes: 
        - name: uamsclient-persistent-storage 
          persistentVolumeClaim: 
            claimName: uamsclient-pv-claim

Run as a Kubernetes DaemonSet

To apply the Agent as a DaemonSet in your multi-node cluster, use the following file.

apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: uamsclient-daemonset
  labels:
    k8s-app: uamsclient
spec:
  selector:
    matchLabels:
      name: uamsclient
  template:
    metadata:
      labels:
        name: uamsclient
    spec:
      tolerations:
      # this toleration is to have the daemonset runnable on master nodes
      # remove it if your masters can't run pods
      - key: node-role.kubernetes.io/master
        operator: Exists
        effect: NoSchedule
      containers:
      - name: uamsclient
        image: solarwinds/swo-agent:latest
        resources:
          limits:
            memory: 200Mi
          requests:
            cpu: 100m
            memory: 200Mi
        env:
        - name: UAMS_ACCESS_TOKEN
          valueFrom:
            secretKeyRef:
              name: access-token
              key: ACCESS_TOKEN
        - name: SWO_URL
          value: <cluster endpoint>
        - name: UAMS_CONTAINER_HOSTNAME
          value: <optional custom container name>
        - name: UAMS_CLIENT_ID_OVERRIDE_SOURCE_NAME
          valueFrom:
            fieldRef:
              fieldPath: spec.nodeName
      terminationGracePeriodSeconds: 30

Run as a Kubernetes sidecar

To apply the Agent as a sidecar in a replicated workload, ensure that the Agent ID is not changing when the container is restarted or when a new pod for the same replica is started.

To achieve this, use StatefulSet which will always assign the same host name for a replica with a particular ordinal number. The hostname can then be used as UAMS_CLIENT_ID_OVERRIDE_SOURCE_NAME to ensure that each replica has always the same Agent ID.

The restartPolicy setting must be set to OnFailure so that the Agent is able to perform restart procedures sent from the SolarWinds Observability SaaS.

apiVersion: apps/v1 
kind: StatefulSet 
metadata: 
  name: <app>-with-uamsclient-sidecar 
spec: 
  selector: 
    matchLabels: 
      app: <app>-with-uamsclient-sidecar 
  serviceName: <app>-with-uamsclient-sidecar 
  replicas: 3 
  template: 
    metadata: 
      labels: 
        app: <app>-with-uamsclient-sidecar 
    spec: 
      restartPolicy: OnFailure 
      containers: 
      # - name: <app> # main app container 
      - name: uamsclient-sidecar 
        image: solarwinds/swo-agent:latest 
        env: 
          - name: UAMS_ACCESS_TOKEN 
            value: <token> 
          - name: SWO_URL 
            value: <cluster endpoint> 
          - name: UAMS_CLIENT_ID_OVERRIDE_SOURCE_NAME 
            valueFrom: 
              fieldRef: 
                fieldPath: metadata.name 
        volumeMounts: 
          - name: uamsclient-workdir 
            mountPath: /uamsclient/workdir 
  volumeClaimTemplates: 
  - metadata: 
      name: uamsclient-workdir 
    spec: 
      accessModes: [ "ReadWriteOnce" ] 
      resources: 
        requests: 
          storage: 1Gi