Skip to content
Start free

Introduction

Introduction

What Kubeadapt is

Kubeadapt is Kubernetes cost monitoring and optimization. An agent in your cluster reads resource state and usage and sends it to Kubeadapt over HTTPS. Kubeadapt attributes cost down to the container, recommends right-sized CPU and memory requests for each workload, and finds idle nodes and orphaned volumes.

The agent does not change your workloads. You review each recommendation and apply it through your own deployment process. The FAQ lists the exact permissions the agent holds.

About 5 minutes to install. Recommendations after about 7 days of usage data.

Supported platforms

The agent runs on managed and self-hosted Kubernetes, versions 1.24 to 1.35.

Self-hosted

  • kubeadm, kops, RancherFull support
  • Red Hat OpenShiftFull support
  • Bare-metal and on-premisesCustom pricing

    Override the default per-core and per-GiB rates from the dashboard.

Every cluster needs Kubernetes 1.24 to 1.35, metrics-server for usage data, and outbound HTTPS to ingest.kubeadapt.io.

Your first week

Three steps, in the order we suggest.

01

Detect

Connect your first cluster and wait about 24 hours for usage data. Then open the dashboard and start with total spend, idle nodes and orphaned volumes.

Cost figures appear within 30 minutes.

02

Optimize

Open Recommendations. Review CPU and memory rightsizing for non-production workloads first, copy the suggested YAML and apply it through your own GitOps flow. Check spot migration candidates next.

Recommendations need about 7 days of usage data.

Read the Rightsizing guide
03

Prevent

Review Best Practices scores and set up Smart Alerting rules for cost changes. Network cost from the optional eBPF agent is in early access: request early access.

Alerts go to Slack, email or a webhook.

Install in about five minutes

  1. 1

    Add the Helm repository.

    helm repo add kubeadapt …

  2. 2

    Get an agent token.

    Open app.kubeadapt.io → Clusters → Add Cluster. Pick your provider, name the cluster and choose Production or Non-Production.

  3. 3

    Install with Helm.

    Copy the install command the dashboard shows, or use the one below.

  4. 4

    Verify the agent.

    Check that the pod is running: kubectl get pods -n kubeadapt

bash
1# 1. Add the Helm repository
2helm repo add kubeadapt https://kubeadapt.github.io/kubeadapt-helm
3helm repo update
4
5# 2. Get an agent token at app.kubeadapt.io → Clusters → Add cluster
6
7# 3. Install
8helm install kubeadapt kubeadapt/kubeadapt \
9  --namespace kubeadapt --create-namespace \
10  --set agent.config.token="<your-token>"
11
12# 4. Check that the agent pod is running
13kubectl get pods -n kubeadapt

The cluster appears in the dashboard about a minute after the agent starts. Cost figures appear within 30 minutes; rightsizing recommendations need about 7 days of usage data.

Read the full Quick Start

Where to get help

FAQ

Permissions, supported clusters, data handling, pricing.

Open the FAQ