This guide lists the steps necessary to install Kubeflow on any conformant Kubernetes, including AKS, EKS, GKE, Openshift and any kubeadm-deployed cluster, provided that you have access to it via kubectl. We recommend microk8s for local deployments.
Kubeflow 1.6 supports Kubernetes version 1.22, which you can install with the command
snap install microk8s --classic --channel=1.22/stable
The previous release Kubeflow 1.4 requires 1.21 version of Kubernetes.
Note that at minimum the following addons should be enabled:
microk8s enable dns storage ingress metallb:10.64.140.43-10.64.140.49
Contents:
- Install the Juju client
- Connect Juju to your Kubernetes cluster
- Create a controller
- Create a model
- Deploy Kubeflow
- Set URL in authentication methods
- Access the dashboard
Install the Juju client
On Linux, installjuju via snap with the following command:
snap install juju --classic
Alternatively, brew install juju on macOS or download the Windows installer.
Connect Juju to your Kubernetes cluster
In order to operate workloads in your Kubernetes cluster with Juju, you have to add your cluster to the list of clouds in juju via theadd-k8s command.
If your Kubernetes config file is in the standard location (~/.kube/config on Linux), and you only have one cluster, you can simply run:
juju add-k8s myk8s
If your kubectl config file contains multiple clusters, you can specify the appropriate one by name:
juju add-k8s myk8s --cluster-name=foo
Finally, to use a different config file, you can set the KUBECONFIG environment variable to point to the relevant file. For example:
KUBECONFIG=path/to/file juju add-k8s myk8s
For more details, see the Juju documentation.
Create a controller
To operate workloads on your Kubernetes cluster, Juju uses controllers. You can create a controller with the bootstrap command:
juju bootstrap myk8s my-controller
This command will create a couple of pods under the my-controller namespace. You can see your controllers with the juju controllers command.
You can read more about controllers in the Juju documentation.
Create a model
A model in Juju is a blank canvas where your operators will be deployed, and it holds a 1:1 relationship with a Kubernetes namespace.You need to create a model and give it the name kubeflow, with the add-model command:
juju add-model kubeflow
You can list your models with the juju models command.
Deploy Kubeflow
The minimum resources required to deploy kubeflow are: 50Gb disk space, 16Gb RAM and 2 CPUs available to your Linux machine or VM.
If you have fewer resources, please deploy kubeflow-lite.
Once you have a model, you can simply juju deploy any of the provided Kubeflow bundles into your cluster.
For example, for the Kubeflow lite bundle, run:
juju deploy kubeflow --trust --channel=1.6/stable
Congratulations, Kubeflow is now installing !
You can observe your Kubeflow deployment getting spun-up with the command:
watch -c juju status --color
In the previous 1.4 release of Kubeflow, in order to set up Istio correctly when RBAC is enabled, you need to provide the istio-ingressgateway operator access to Kubernetes resources. The following command will create the appropriate role:
microk8s kubectl patch role -n kubeflow istio-ingressgateway-operator -p '{"apiVersion":"rbac.authorization.k8s.io/v1","kind":"Role","metadata":{"name":"istio-ingressgateway-operator"},"rules":[{"apiGroups":["*"],"resources":["*"],"verbs":["*"]}]}'
Set URL in authentication methods
A final step to enable your Kubeflow dashboard access is to provide the istio-ingressgateway public URL to dex-auth and oidc-gatekeeper via the following commands:juju config dex-auth public-url=http://<URL>
juju config oidc-gatekeeper public-url=http://<URL>
Where <URL> is the hostname that the Kubeflow dashboard responds to. For example, in a typical MicroK8s installation with MetalLB addon enabled, this URL is http://10.64.140.43.nip.io. Note that when you have set up DNS, you should use the resolvable address used by istio-ingressgateway.