Skip to main content

Running Kurtosis in Kubernetes

This guide assumes that you have Kurtosis installed.

If you would like more information on Kubernetes and how to set up, run and manage a cluster check out these official docs.

Please note that in order to ensure Kurtosis works the same way over Kubernetes as it does over Docker locally, service names must be a valid RFC-1035 Label Name. This means service names must contain: at most 63 characters, only lowercase alphanumeric characters or '-', start with an alphabetic character, and end with an alphanumeric character.

I. Create a Kubernetes Cluster

There are many different ways to get a Kubernetes cluster (roughly ordered easiest to hardest):

  • Use Kubernetes provided with Docker Desktop
  • Install Minikube
  • Use k3s
  • Deploy it on an onprem cluster and manage the machine provisioning yourself
  • Deploy it on the cloud, managing the Kubernetes nodes on cloud instances yourself (e.g. EC2, AVM, GCE, etc)
  • Deploy it on a managed Kuberenetes cluster, managing scaling and configurations yourself (e.g. EKS, AKS, GKE)
Kurtosis Kloud Early Access

If you're looking to run a stress-free "Kurtosis on Kubernetes in the cloud", look no further! Check out Kurtosis Cloud.

II. Add you Kubernetes Cluster credentials to your kubeconfig

This step will depend highly on how your cluster was created. But generally you will need to either:

  • Manually edit the kubeconfig file to contain cluster and authentication data. For more information, see Kubernetes docs.
  • Use your cloud provider's CLI to automatically edit the kubeconfig file so that it contains your cluster and authentication data. For example, you if you are using Amazon's managed Kubernetes service (called EKS), this tutorial is comprehensive.

III. Add your cluster information to kurtosis-config.yml

  1. Open the file located at "$(kurtosis config path)". This should look like /Users/<YOUR_USER>/Library/Application Support/kurtosis/kurtosis-config.yml on MacOS.
  2. Paste the following contents, changing NAME-OF-YOUR-CLUSTER and STORAGE-CLASS-TO-USE as per the cluster you created and save:
config-version: 2
should-send-metrics: true
type: "docker"
type: "kubernetes"
kubernetes-cluster-name: "minikube"
storage-class: "standard"
enclave-size-in-megabytes: 10
type: "kubernetes"
kubernetes-cluster-name: "NAME-OF-YOUR-CLUSTER"
storage-class: "STORAGE-CLASS-TO-USE"
enclave-size-in-megabytes: 10
Storage Class

The Storage Class specified in the configuration above will be used for spinning up persistent volumes. Make sure you have the right value in case you are using persistent directories.

We support storage classes that support dynamic provisioning; here are some of them:

  1. For AWS we recommend the aws-ebs-csi-driver
  2. For DigitalOcean we recommend do-block-storage but your cluster should have this out of the box
  3. K3s the default provisioner local-path should just work out of the box
  4. For minikube the default provisioner standard should just work out of the box
  5. On Docker Desktop Kubernetes the default provisioner is hostpath

For any other cloud setup please reach out to us by creating an issue on our GitHub

IV. Configure Kurtosis

  1. Run kurtosis cluster set cloud. This will start the engine remotely. See the CLI reference for more information about kurtosis cluster commands here.
  2. In another terminal, run kurtosis gateway. This will act as a middle man between your computer's ports and your services deployed on Kubernetes ports and has to stay running as a separate process.

Done! Now you can run any Kurtosis command or package just like if you were doing it locally.

Kurtosis Kloud Early Access

To switch back to using Kurtosis locally, simply use: kurtosis cluster set docker

V. [Optional] Activate the enclave pool to accelerate the enclave creation time

This step is optional, but we recommend taking it as it improves the user experience during the enclave creation, specifically regarding speed.

Creating a new enclave from scratch demands several time-consuming engine tasks and the creation of resources.

The enclave pool feature was introduced to reduce the time it takes for a user to run a Kurtosis package in the cloud by spinning up the enclaves before they are needed.

The enclave pool is a functionality of the Kurtosis engine that automatically creates idle enclaves, when the engine is started, that are then used whenever users need to create a new enclave (e.g: when running the kurtosis enclave add command).

This mechanism reduces enclave creation time by using a running idle enclave when a new enclave is requested from the engine.

To enable this feature you have to run the following:

  1. Run kurtosis engine restart --enclave-pool-size {pool-size-number}. If you already follow the previous step and replace the {pool-size-number} with an integer


  1. Run kurtosis engine start --enclave-pool-size {pool-size-number}. If the engine has not been started yet.