Managing with Helm

There is a community-supported Helm chart with extensive customization settings exposed via the values.yaml. The Helm installation assumes an existing Prometheus installation.

Get Repository Info

Start by giving your local helm installation access to the helm chart.

helm repo add opencost-charts https://opencost.github.io/opencost-helm-chart
helm repo update

Installing OpenCost with Helm

Create a values file (example) then install into your kubernetes cluster.

helm install opencost opencost-charts/opencost --namespace opencost --create-namespace -f values.yaml

Example Configuration

The full list of installation options can be found here.

opencost:
  prometheus:
    internal:
      namespaceName: monitoring
      serviceName: prometheus-kube-prometheus-prometheus
      port: 9090

  dataRetention:
    dailyResolutionDays: 30  # default: 15

  exporter:
    defaultClusterId: <cluster name>
    replicas: 1
    resources:
      requests:
        cpu: "10m"
        memory: "55Mi"
      limits:
        memory: "1Gi"
    persistence:
      enabled: true
      storageClass: "csi-cinder-high-speed"  # adjust based on your environment
      accessMode: ReadWriteOnce
      size: 5Gi

  customPricing:
    enabled: true
    provider: aws
    costModel:
      description: Modified pricing configuration.
      CPU: 0.031611
      spotCPU: 0.006655
      RAM: 0.004237
      spotRAM: 0.000892
      GPU: 0.95
      storage: 0.00005479452
      zoneNetworkEgress: 0.01
      regionNetworkEgress: 0.01
      internetNetworkEgress: 0.143
      spotLabel: ""
      spotLabelValue: ""
      awsServiceKeyName: ""
      awsServiceKeySecret: ""
      awsSpotDataRegion: ""
      awsSpotDataBucket: ""
      awsSpotDataPrefix: ""
      athenaBucketName: ""
      athenaRegion: ""
      athenaDatabase: ""
      athenaTable: ""
      projectID: "${ACCOUNT_ID}"

  metrics:
    kubeStateMetrics:
      emitKsmV1Metrics: false
      emitKsmV1MetricsOnly: false
    serviceMonitor:
      enabled: true
      additionalLabels:
        release: prometheus

  ui:
    enabled: true

Important Configuration Notes

1. Metrics Configuration: To avoid duplicate metrics from v2 KubeStateMetrics running in your cluster and v1 KubestateMetrics from OpenCost, configure the metrics settings as follows:

opencost:
  metrics:
    kubeStateMetrics:
      emitKsmV1Metrics: false
      emitKsmV1MetricsOnly: false

2. Persistence: For production deployments, it's recommended to enable persistence with appropriate storage class and size:

opencost:
  exporter:
    persistence:
      enabled: true
      storageClass: "your-storage-class"
      accessMode: ReadWriteOnce
      size: 5Gi

3. Resource Management: Configure appropriate resource requests and limits:

opencost:
  exporter:
    resources:
      requests:
        cpu: "10m"
        memory: "55Mi"
      limits:
        memory: "1Gi"

Pricing information is provided through the helm chart values. The example above uses AWS default configuration. Documentation can be found for: AWS, Azure, GCP, and Custom On Premise

Upgrading OpenCost with Helm

helm upgrade opencost opencost-charts/opencost --namespace opencost -f values.yaml

Side-grading with Helm

If you wish to override the version of OpenCost installed by the Helm chart (e.g. downgrading or alternative versions), update the values.yaml to set the tag of the downloaded image (you can also change the registry and repository if necessary).

opencost:
  exporter:
    image:
      registry: quay.io
      repository: kubecost1/kubecost-cost-model
      tag: prod-1.101.3
  ui:
    image:
      registry: quay.io
      repository: kubecost1/opencost-ui
      tag: prod-1.101.3

Deleting with Helm

helm uninstall opencost

Troubleshooting

• Sometimes you don't find cost reports from OpenCost for certain resources. In such scenarios, check whether Prometheus is scraping the metrics. If Prometheus is not scraping any metrics, then ensure whether Service monitors are enabled in that service.

example:

If the cert-manager namespace is not showing any cost values, and when Prometheus is not scraping any metrics, then enable ServiceMonitor as below:

  prometheus:
  enabled: true
  servicemonitor:
    enabled: true
    namespace: <prometheus namespace>
    honorLables: true

• Sometimes, the OpenCost report displays zero values for specific namespaces.

example:

kubectl cost --service-port 9003 --service-name opencost --kubecost-namespace opencost --allocation-path /allocation/compute  \
    namespace \
    --historical \
    --window 1d \
    --show-cpu \
    --show-memory \
    --show-pv \
    --show-efficiency=true


| CLUSTER | NAMESPACE     | CPU      | CPU EFF. | MEMORY   | MEMORY EFF. | PV       | TOTAL COST (ALL) | COST EFFICIENCY |
|---------|---------------|----------|----------|----------|-------------|----------|------------------|-----------------|
| dev  | kube-system   | 0.138740 | 0.001965 | 0.023890 | 0.386018    | 0.000000 | 0.162630         | 0.058381        |
|         | monitoring           | 0.000000 | 0.000000 | 0.000000 | 0.000000    | 0.000000 |         0.000000 |        0.000000 |

The reason for this could be low usage within a specific time window. You might need to wait for a period, possibly one or more days, to allow it to accumulate the metrics and display the data in the OpenCost report.