An alternative to running the CostGraph operator is to run the CostGraph agent as a DaemonSet. This is useful if you want to monitor all nodes in your Kubernetes cluster.

​
Prerequisites

​
Installation Steps

1

Create Configuration

Create a ConfigMap with your configuration and API key.
cat <<EOF | kubectl apply -f -
apiVersion: v1
kind: ConfigMap
metadata:
  name: costgraph-agent-config
  namespace: costgraph
data:
  conf.yaml: |
    api_key: "your costgraph api key"
    prometheus:
      enabled: true
      url: "prometheus.example.com"
      write_path: "/api/v1/write"
      timeout: 10s
      interval: 60s
      tlsInsecure: false
      labels:
        env: "daemonset"
    expected_utilisation:
      cpu: 60
      memory: 75
EOF
2

Deploy DaemonSet

Apply the DaemonSet manifest to your cluster.
cat <<EOF | kubectl apply -f -
apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: costgraph-agent
  namespace: costgraph
spec:
  selector:
    matchLabels:
      app: costgraph-agent
  template:
    metadata:
      labels:
        app: costgraph-agent
    spec:
      hostPID: true
      hostNetwork: true
      containers:
      - name: agent
        image: ghcr.io/baselinehq/costgraph-agent:v0.0.7
        command: ["costgraph-agent", "-conf", "/etc/costgraph-agent/conf.yaml"]
        imagePullPolicy: Always
        env:
        - name: DEBUG
          value: "true"
        volumeMounts:
        - name: state
          mountPath: /var/lib/costgraph-agent
        - name: proc
          mountPath: /host/proc
          readOnly: true
        - name: config
          mountPath: /etc/costgraph-agent/conf.yaml
          subPath: conf.yaml
          readOnly: true
      volumes:
      - name: state
        emptyDir: {}
      - name: proc
        hostPath:
          path: /proc
      - name: config
        configMap:
          name: costgraph-agent-config
EOF
3

Verify Installation

Check that the agent pods are running on all nodes.
kubectl get pods -n costgraph
kubectl logs -n costgraph -l app=costgraph-agent