rancher-partner-charts/charts/kubecost/cost-analyzer

Kubecost Helm chart

This is the official Helm chart for Kubecost, an enterprise-grade application to monitor and manage Kubernetes spend. Please see the website for more details on what Kubecost can do for you and the official documentation here, or contact team@kubecost.com for assistance.

To install via Helm, run the following command.

helm upgrade --install kubecost -n kubecost --create-namespace \
  --repo https://kubecost.github.io/cost-analyzer/ cost-analyzer \
  --set kubecostToken="aGVsbUBrdWJlY29zdC5jb20=xm343yadf98"

Alternatively, add the Helm repository first and scan for updates.

helm repo add kubecost https://kubecost.github.io/cost-analyzer/
helm repo update

Next, install the chart.

helm install kubecost kubecost/cost-analyzer -n kubecost --create-namespace \
  --set kubecostToken="aGVsbUBrdWJlY29zdC5jb20=xm343yadf98"

While Helm is the recommended install path for Kubecost especially in production, Kubecost can alternatively be deployed with a single-file manifest using the following command. Keep in mind when choosing this method, Kubecost will be installed from a development branch and may include unreleased changes.

kubectl apply -f https://raw.githubusercontent.com/kubecost/cost-analyzer-helm-chart/develop/kubecost.yaml

The following table lists commonly used configuration parameters for the Kubecost Helm chart and their default values. Please see the values file for the complete set of definable values.

Parameter	Description	Default
global.prometheus.enabled	If false, use an existing Prometheus install. More info.	true
prometheus.kube-state-metrics.disabled	If false, deploy kube-state-metrics for Kubernetes metrics	false
prometheus.kube-state-metrics.resources	Set kube-state-metrics resource requests and limits.	{}
prometheus.server.persistentVolume.enabled	If true, Prometheus server will create a Persistent Volume Claim.	true
prometheus.server.persistentVolume.size	Prometheus server data Persistent Volume size. Default set to retain ~6000 samples per second for 15 days.	32Gi
prometheus.server.retention	Determines when to remove old data.	15d
prometheus.server.resources	Prometheus server resource requests and limits.	{}
prometheus.nodeExporter.resources	Node exporter resource requests and limits.	{}
prometheus.nodeExporter.enabled prometheus.serviceAccounts.nodeExporter.create	If false, do not crate NodeExporter daemonset.	true
prometheus.alertmanager.persistentVolume.enabled	If true, Alertmanager will create a Persistent Volume Claim.	true
prometheus.pushgateway.persistentVolume.enabled	If true, Prometheus Pushgateway will create a Persistent Volume Claim.	true
persistentVolume.enabled	If true, Kubecost will create a Persistent Volume Claim for product config data.	true
persistentVolume.size	Define PVC size for cost-analyzer	32.0Gi
persistentVolume.dbSize	Define PVC size for cost-analyzer's flat file database	32.0Gi
ingress.enabled	If true, Ingress will be created	false
ingress.annotations	Ingress annotations	{}
ingress.className	Ingress class name	{}
ingress.paths	Ingress paths	["/"]
ingress.hosts	Ingress hostnames	[cost-analyzer.local]
ingress.tls	Ingress TLS configuration (YAML)	[]
networkPolicy.enabled	If true, create a NetworkPolicy to deny egress	false
networkPolicy.costAnalyzer.enabled	If true, create a newtork policy for cost-analzyer	false
networkPolicy.costAnalyzer.annotations	Annotations to be added to the network policy	{}
networkPolicy.costAnalyzer.additionalLabels	Additional labels to be added to the network policy	{}
networkPolicy.costAnalyzer.ingressRules	A list of network policy ingress rules	null
networkPolicy.costAnalyzer.egressRules	A list of network policy egress rules	null
networkCosts.enabled	If true, collect network allocation metrics More info	false
networkCosts.podMonitor.enabled	If true, a PodMonitor for the network-cost daemonset is created	false
serviceMonitor.enabled	Set this to true to create ServiceMonitor for Prometheus operator	false
serviceMonitor.additionalLabels	Additional labels that can be used so ServiceMonitor will be discovered by Prometheus	{}
serviceMonitor.relabelings	Sets Prometheus metric_relabel_configs on the scrape job	[]
serviceMonitor.metricRelabelings	Sets Prometheus relabel_configs on the scrape job	[]
prometheusRule.enabled	Set this to true to create PrometheusRule for Prometheus operator	false
prometheusRule.additionalLabels	Additional labels that can be used so PrometheusRule will be discovered by Prometheus	{}
grafana.resources	Grafana resource requests and limits.	{}
grafana.sidecar.datasources.defaultDatasourceEnabled	Set this to false to disable creation of Prometheus datasource in Grafana	true
serviceAccount.create	Set this to false if you want to create the service account kubecost-cost-analyzer on your own	true
tolerations	node taints to tolerate	[]
affinity	pod affinity	{}
kubecostProductConfigs.productKey.mountPath	Use instead of kubecostProductConfigs.productKey.secretname to declare the path at which the product key file is mounted (eg. by a secrets provisioner)	N/A
kubecostFrontend.api.fqdn	Customize the upstream api FQDN	computed in terms of the service name and namespace
kubecostFrontend.model.fqdn	Customize the upstream model FQDN	computed in terms of the service name and namespace
clusterController.fqdn	Customize the upstream cluster controller FQDN	computed in terms of the service name and namespace
global.grafana.fqdn	Customize the upstream grafana FQDN	computed in terms of the release name and namespace

Adjusting Log Output

The log output can be customized during deployment by using the LOG_LEVEL and/or LOG_FORMAT environment variables.

Adjusting Log Level

Adjusting the log level increases or decreases the level of verbosity written to the logs. To set the log level to trace, the following flag can be added to the helm command.

--set 'kubecostModel.extraEnv[0].name=LOG_LEVEL,kubecostModel.extraEnv[0].value=trace'

Adjusting Log Format

Adjusting the log format changes the format in which the logs are output making it easier for log aggregators to parse and display logged messages. The LOG_FORMAT environment variable accepts the values JSON, for a structured output, and pretty for a nice, human-readable output.

Value	Output
JSON	{"level":"info","time":"2006-01-02T15:04:05.999999999Z07:00","message":"Starting cost-model (git commit \"1.91.0-rc.0\")"}
pretty	2006-01-02T15:04:05.999999999Z07:00 INF Starting cost-model (git commit "1.91.0-rc.0")

Testing

To perform local testing do next:

• install locally kind according to documentation.
• install locally ct according to documentation.
• create local cluster using kind
  use image version from https://github.com/kubernetes-sigs/kind/releases e.g. kindest/node:v1.25.11@sha256:227fa11ce74ea76a0474eeefb84cb75d8dad1b08638371ecf0e86259b35be0c8

kind create cluster --image kindest/node:v1.25.11@sha256:227fa11ce74ea76a0474eeefb84cb75d8dad1b08638371ecf0e86259b35be0c8

• perform ct execution

ct install  --chart-dirs="." --charts="."

• perform ct StatefulSet execution

# create multiple nodes kind config
cat > kind-config.yaml <<EOF
kind: Cluster
apiVersion: kind.x-k8s.io/v1alpha4
nodes:
- role: control-plane
- role: worker
- role: worker
EOF
# creaet kind cluster with kind config
kind create cluster --name kubecost-statefulset --config kind-config.yaml --image kindest/node:v1.25.11@sha256:227fa11ce74ea76a0474eeefb84cb75d8dad1b08638371ecf0e86259b35be0c8
# deploy an object storage for our testing purpose (https://min.io/docs/minio/kubernetes/upstream/index.html)
curl --silent https://raw.githubusercontent.com/minio/docs/master/source/extra/examples/minio-dev.yaml  | sed -e "s/kubealpha.local/kubecost-statefulset-worker/" -e "s%minio server /data%mkdir -p /data/kubecost; minio server /data%" | kubectl apply -f -
# create a headless service to the minio S3 API port
kubectl create service clusterip -n minio-dev minio --tcp=9000:9000 --clusterip="None"
# create our testing namespace
kubectl create namespace kubecost-statefulset
# create the bucket config 
cat > etlBucketConfigSecret.yaml <<EOF
type: s3
config:
  bucket: kubecost
  endpoint: minio.minio-dev:9000
  insecure: true
  access_key: minioadmin
  secret_key: minioadmin
EOF
# create the secret with the object-store.yaml
kubectl create secret generic -n kubecost-statefulset object-store --from-file=object-store.yaml=etlBucketConfigSecret.yaml
# start our chart-testing
ct install --namespace kubecost-statefulset --chart-dirs="." --charts="." --helm-extra-set-args="--set=global.prometheus.enabled=true --set=global.grafana.enabled=true --set=kubecostDeployment.leaderFollower.enabled=true --set=kubecostDeployment.statefulSet.enabled=true --set=kubecostDeployment.replicas=2 --set=kubecostModel.etlBucketConfigSecret=object-store"
# cleanup
kind delete cluster --name kubecost-statefulset