These steps walk through deploying Vespa using the official Helm chart.
The following tools are required for a smooth deployment.
These instructions assume that your kubeconfig is pointing to an active Kubernetes cluster. Refer to the Getting Started guide to create a Kubernetes cluster. For instructions on deploying Vespa
locally on MiniKube, refer to the Deploy Vespa Locally guide.
Vespa on Kubernetes uses a Custom Resource Definition (CRD) called a VespaSet. Users intending to manage
the CRD definition by themselves should apply it to the cluster before installation.
The permissions that are needed to run Vespa are listed on the Permissions page. The Helm Chart will automatically apply a default set of RBAC API resources onto the cluster.
Note: Vespa on Kubernetes is an enterprise feature. You will need access to the images below. Contact us through our support portal to receive an authentication ID and token. For production use, we recommend mirroring these images into your own registry or a well-known internal repository appropriate to your infrastructure.
VESPA_IMAGE=images.ves.pa/kubernetes/vespaVESPA_OPERATOR_IMAGE=images.ves.pa/kubernetes/operatorHELM_CHART_REF=oci://images.ves.pa/helm/vespa-operator
We will use this naming convention throughout this guide. The tags for all three images conform to the Vespa Version release semantics.
We recommend using the latest Vespa release as the default. We will refer to it as VESPA_VERSION.
The Vespa Operator and all Vespa components are local to a namespace. We will refer to the namespace as NAMESPACE in this guide.
Authenticate to the Helm Chart OCI registry. The credentials will be provided by our support team.
$ helm registry login images.ves.pa -u $USER -p $TOKEN
Install the Helm Chart onto the namespace. This will deploy the Vespa Operator and apply
the VespaSet resource definition. Set image.repository to VESPA_OPERATOR_IMAGE as provided by our support team.
The image.tag refers to the VESPA_VERSION.
$ helm install vespa-operator $HELM_CHART_REF --namespace $NAMESPACE --create-namespace --set image.repository=$VESPA_OPERATOR_IMAGE --set image.tag=$VESPA_VERSION
The lifecycle of the CRD definition can be managed separately. However, the CRD specification must
be manually applied to the Kubernetes cluster before installing the Helm Chart. Our support team can provide this specification if necessary.
To do this, use the --skip-crds option in Helm.
$ kubectl apply vespasets.k8s.ai.vespa-v1.yaml $ helm install vespa-operator $HELM_CHART_REF --namespace $NAMESPACE --create-namespace --skip-crds --set image.repository $VESPA_OPERATOR_IMAGE --set image.tag $VESPA_VERSION
Ensure that the Deployment resource was successfully created, and that the Vespa Operator Pod is running.
To set up a dev environment in Vespa on Kubernetes, refer to the example on the
Setup Dev Environment page.
A VespaSet is a quorum of ConfigServer Pods that manage the lifecycle of Vespa applications. Several examples of
VespaSet resources are provided in the Helm Chart samples directory.
An example of a VespaSet for an archetypical Amazon Elastic Kubernetes Service (EKS) setup is shown below.
# vespaset sample for EKS
$ cat > vespaset.yaml <<EOF
apiVersion: k8s.ai.vespa/v1
kind: VespaSet
metadata:
name: vespaset-sample
namespace: ${NAMESPACE}
spec:
version: "${VESPA_VERSION}"
configServer:
image: "${VESPA_OPERATOR_IMAGE}"
storageClass: "gp3"
generateRbac: false
application:
image: "${VESPA_IMAGE}"
storageClass: "gp3"
ingress:
endpointType: "LOAD_BALANCER"
EOF
$ kubectl apply -f vespaset.yaml
An example for an archetypical local deployment on MiniKube is shown below.
# vespaset sample for MiniKube
$ cat > vespaset.yaml <<EOF
apiVersion: k8s.ai.vespa/v1
kind: VespaSet
metadata:
name: vespaset-sample
namespace: ${NAMESPACE}
spec:
version: "${VESPA_VERSION}"
configServer:
image: "${VESPA_OPERATOR_IMAGE}"
storageClass: "local-storage"
generateRbac: false
application:
image: "${VESPA_IMAGE}"
storageClass: "local-storage"
ingress:
endpointType: "NONE"
EOF
$ kubectl apply -f vespaset.yaml
Once a VespaSet is applied, the operator will automatically detect the newly created resource and create a quorum of
ConfigServers.
Once applied, the ConfigServers will bootstrap themselves until meeting a quorum. This process takes roughly a minute. The bootstrap process is complete once
the VespaSet resource shows the status as RUNNING for all ConfigServer Pods. For example:
$ kubectl describe vespaset vespaset-sample -n $NAMESPACE
Name: vespaset-sample
Namespace: $NAMESPACE
API Version: k8s.ai.vespa/v1
Kind: VespaSet
Spec:
Application:
Image: 192.168.49.2:5000/localhost/vespaai/kubernetes
Storage Class: gp3
Config Server:
Generate Rbac: false
Image: 192.168.49.2:5000/localhost/vespaai/kubernetes
Storage Class: gp3
Ingress:
Endpoint Type: NONE
Version: 8.643.16
Status:
Bootstrap Status:
Pods:
cfg-1:
Last Updated: 2026-01-29T21:38:45Z
Message: Pod is running
Phase: RUNNING
Converged Version: 8.643.16
cfg-2:
Last Updated: 2026-01-29T21:38:09Z
Message: Pod is running
Phase: RUNNING
Converged Version: 8.643.16
cfg-3:
Last Updated: 2026-01-29T21:36:32Z
Message: Pod is running
Phase: RUNNING
Converged Version: 8.643.16
Last Transition Time: 2026-01-29T21:33:55Z
Message: All configservers running
Phase: RUNNING
Events: <none>
A Vespa application can be deployed through the ConfigServers' ingress endpoint once a quorum has been met. Refer to the Vespa Sample Applications to get started. In the following example, we will use the Album Recommendation sample application.
Set up the Vespa CLI to download the Album Recommendation sample to a directory.
$ vespa clone album-recommendation myapp && cd myapp
The Node resources must be specified for any application package is deployed on Vespa on Kubernetes.
These will directly translate to Kubernetes container resource requests and limits.
In a default deployment without any PodTemplate overrides, the requests will equal the limits for a container.
Modify the container and content cluster specifications in the application package, as shown below:
<?xml version="1.0" encoding="utf-8" ?>
<services version="1.0" xmlns:deploy="vespa" xmlns:preprocess="properties">
<container id="default" version="1.0">
<document-api/>
<search/>
<nodes count="2">
<resources vcpu="2" memory="2Gb" disk="20Gb" />
</nodes>
</container>
<content id="music" version="1.0">
<min-redundancy>2</min-redundancy>
<documents>
<document type="music" mode="index" />
</documents>
<nodes count="2">
<resources vcpu="2" memory="2Gb" disk="20Gb" />
</nodes>
</content>
</services>
Enable port-forwarding from the ConfigServer's ingress port 19071 to your local port 19071. Any ConfigServer
Pod can be used.
$ vespa config set target local $ kubectl -n $NAMESPACE port-forward pod/cfg-1 19071:19071
Deploy and activate the application.
$ vespa prepare --target local $ while ! vespa --target local activate; do sleep 1; done
The ConfigServers will create the Container, Content, and Cluster-Controller Pods as specified in the application package. The deployment
is considered complete once all Pods show the phase RUNNING in the VespaSet status.
Port-forwarding provides a simple way to ingress to the ConfigServer locally. For other ingress options, refer to the Configuring the External Access Layer page.
Feed and query documents by port-forwarding the ConfigServer ingress port and the Dataplane ingress port, then using the Vespa CLI.
$ kubectl -n $NAMESPACE port-forward pod/cfg-1 19071:19071 $ kubectl -n $NAMESPACE port-forward pod/default-100 8080:8080 $ vespa feed dataset/A-Head-Full-of-Dreams.json $ vespa query 'yql=select * from music where true limit 1'
Refer to the Vespa CLI documentation for the full list of available commands.