Helm guide – Feldera Test

Last Updated: 5/8/2026

These instructions explain how to install and run Feldera Enterprise on a Kubernetes cluster. It requires a valid Feldera Enterprise license — please request one at learnmore@feldera.com if you are interested.

Prerequisites

kubectl: kubectl version

Used to interact with the Kubernetes cluster
helm: helm version

Package manager to conveniently manage deployments on Kubernetes
Feldera license, notably the account id and license key. They are required in the installation. Please reach out to us at learnmore@feldera.com if you are interested.
The Feldera Helm chart and Docker images are hosted on AWS ECR public registry . We will use the chart and multi-arch images by referring to their online repository naming public.ecr.aws/feldera/<image>:<version>.

Kubernetes Cluster Configuration

Node Sizing

Resource	Recommendation
CPU	32–48 CPUs per node
RAM	At least 32 GiB per node

By default, the Feldera control plane and pipelines run on the same nodes. Use nodeSelectors to run them on different nodes or node types.

RAM requirements vary by workload. See the memory usage documentation for detailed guidance.

Persistent Volume Sizing

Persistent volumes should meet the following specifications:

Volume Size: At least the size of the input data. Feldera stores state to perform incremental computations, so the volume must be large enough to hold it. Users can expand pipeline storage on demand.
IOPS: At least 16,000
Throughput: At least 1,000 MB/s

Recommended disk types by cloud provider:

Cloud Provider	Disk Type
AWS	gp3
Azure	Ultra Disk
GCP	Hyperdisk Balanced

Installing Feldera Enterprise

Kubernetes access: check that your kubectl is configured to your Kubernetes cluster.
```
kubectl cluster-info
kubectl get namespace
```
Version: Visit https://gallery.ecr.aws/feldera/feldera-chart , click on the “Image tags” tab for a list of versions, and choose a version. The most recent version should be at the top of the list and is normally the right choice.

Installation: we use our Helm chart, which internally refers to the other images, to perform the installation in namespace feldera with release name feldera.


ACCOUNT_ID="00000000-0000-0000-0000-000000000000"  # Set to own
LICENSE_KEY="00000000-0000-0000-0000-000000000000"  # Set to own
VERSION=0.88.0  # Replace with previously chosen version
 
helm upgrade --install feldera \
    oci://public.ecr.aws/feldera/feldera-chart --version "${VERSION}" \
    --namespace feldera --create-namespace \
    --set felderaVersion="${VERSION}" \
    --set felderaAccountId="${ACCOUNT_ID}" \
    --set felderaLicenseKey="${LICENSE_KEY}" \
    --set felderaDatabaseSecretRef="feldera-db-insecure-secret"

Note: the license verification is done by interacting with our online license server at cloud.feldera.com.

Note: how to configure your own database credentials is explained in a further section.

Check: check the status of the deployment.


kubectl get pods -n feldera

… which should output approximately the following:


NAME                                         READY   STATUS    RESTARTS   AGE
feldera-db-0                                 1/1     Running   0          3m9s
feldera-kubernetes-runner-6447b8f56d-86j4w   1/1     Running   0          3m9s
feldera-compiler-server-0                    1/1     Running   0          3m9s
feldera-api-server-c546499bc-wdpkm           1/1     Running   0          3m9s

Usage:

Interaction with Feldera happens through the API server, which has an associated service named <release name>-api-server.<namespace>.svc.cluster.local:8080. With kubectl, this can be port-forwarded to be accessible locally:
```
kubectl port-forward -n feldera svc/feldera-api-server 8080:8080
```
… after which you can (leaving the forwarding running in a terminal):
- Visit Web Console in browser: http://127.0.0.1:8080
- Interact with the API: http://127.0.0.1:8080/v0/…
Note: access through kubectl port-forwarding is mostly useful for test and development. In other cases, setting up an ingress (e.g., in EKS) is likely preferable.

Extra

Retrieving the Feldera values file

The values.yaml file can be retrieved for an existing Feldera release or from the remote Feldera registry.

Remote Feldera registry: Run helm show values oci://public.ecr.aws/feldera/feldera-chart --version {version} >> values.yaml where version is the Feldera platform version for which you want to retrieve values.

Existing Feldera release: Run helm get values {feldera-release-name} --all -o yaml >> values.yaml where feldera-release-name is the name of your release. The value is usually feldera, unless the value has been configured during the installation process.

Configure custom database credentials

Instead of the insecure default DB credentials, you can supply your own custom database credentials.

Secret configuration file: create a file feldera-db-secret.yaml with custom database credentials.


# Filename: feldera-db-secret.yaml
apiVersion: v1
kind: Secret
type: Opaque
metadata:
  name: feldera-db-secret
stringData:
  .user: "..."  # Fill in
  .password: "..."  # Fill in

Create secret: with the configuration file, create the secret:


#kubectl create namespace feldera  # If the namespace does not exist yet
kubectl apply -n feldera -f feldera-db-secret.yaml

Specify secret during installation: in the helm installation command, set the following:
```
--set felderaDatabaseSecretRef="feldera-db-secret"
```
Note: it must be a new installation.

Connect to an external database service

You can connect to an external PostgreSQL database with the following steps:

Secret configuration file: create a file feldera-db-secret.yaml with the .connection_url


 # Filename: feldera-db-secret.yaml
 apiVersion: v1
 kind: Secret
 type: Opaque
 metadata:
   name: feldera-db-secret
 stringData:
   .connection_url: "postgresql://username:password@db-name.cluster-name.region.rds.amazonaws.com/db_name?sslmode=require"

Create secret: with the configuration file, create the secret:


#kubectl create namespace feldera  # If the namespace does not exist yet
kubectl apply -n feldera -f feldera-db-secret.yaml

Optional: Add the TLS Certificate (.pem file) in a configmap:
```
wget https://example.com/your-bundle.pem
kubectl create configmap postgres-pem -n feldera --from-file=postgresql-ca.pem=your-bundle.pem
```
This step is optional, if no certificate is provided feldera will use the system certificates (ca-certificates package from Ubuntu 24.04 and the Amazon RDS global-bundle.pem file) to validate the TLS connection.

Note: If necessary, you can override the file (postgresql-ca.pem) and configmap (postgres-pem) names by adjusting postgresTlsCertificateFile and postgresTlsConfigMapRef in values.yaml.
Specify secret and external postgres configuration during installation: in the helm installation command, set the following:
```
--set postgresExternal=true
--set felderaDatabaseSecretRef="feldera-db-secret"
--set postgresTlsConfigMapRef="postgres-pem"
```
Note: it must be a new installation. If you skip step 3, you must not set the postgresTlsConfigMapRef variable.

Feldera installation overview

Services:


$ kubectl get svc -n feldera
NAME                        TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)     AGE
feldera-kubernetes-runner   ClusterIP   10.43.175.228   <none>        8080/TCP    5m10s
feldera-api-server          ClusterIP   10.43.176.120   <none>        8080/TCP    5m10s
feldera-db                  ClusterIP   10.43.60.190    <none>        5431/TCP    5m10s
feldera-compiler-server     ClusterIP   10.43.60.125    <none>        8085/TCP    5m10s

Pods:


$ kubectl get pods -n feldera
NAME                                         READY   STATUS    RESTARTS   AGE
feldera-db-0                                 1/1     Running   0          5m31s
feldera-kubernetes-runner-6447b8f56d-86j4w   1/1     Running   0          5m31s
feldera-compiler-server-0                    1/1     Running   0          5m31s
feldera-api-server-c546499bc-wdpkm           1/1     Running   0          5m31s

StatefulSets:


$ kubectl get sts -n feldera
NAME                      READY   AGE
feldera-db                1/1     5m41s
feldera-compiler-server   1/1     5m41s

Deployments:


$ kubectl get deployments -n feldera
NAME                        READY   UP-TO-DATE   AVAILABLE   AGE
feldera-kubernetes-runner   1/1     1            1           5m55s
feldera-api-server          1/1     1            1           5m55s

Persistent volume claims:


$ kubectl get pvc -n feldera
NAME                                         STATUS   VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS   AGE
db-storage-feldera-db-0                      Bound    pvc-a5acb4ef-1d3c-4773-ad42-528657f11c94   20Gi       RWO            local-path     6m11s
compiler-storage-feldera-compiler-server-0   Bound    pvc-07369993-c20a-4f8a-abd1-52bf3be413c3   20Gi       RWO            local-path     6m11s

Uninstallation

If you installed Feldera in its own dedicated namespace, the most thorough way to uninstall Feldera is to delete the entire namespace: kubectl delete namespace feldera

If you wish to preserve the namespace, it can be uninstalled through the usual helm method:


helm uninstall -n feldera feldera

This does not delete the persistent volumes, which need to be deleted separately:


kubectl get pvc -n feldera
kubectl delete pvc -n feldera db-storage-feldera-db-0
kubectl delete pvc -n feldera compiler-storage-feldera-compiler-server-0
kubectl get pvc -n feldera

Troubleshooting

Incorrect license

If an incorrect account id and/or license key was provided, the kubernetes-runner will fail to start:


$ kubectl get pods -n feldera
NAME                                         READY   STATUS             RESTARTS     AGE
feldera-db-0                                 1/1     Running            0            2m34s
feldera-compiler-server-0                    1/1     Running            0            2m34s
feldera-kubernetes-runner-854446fb84-rg66z   0/1     CrashLoopBackOff   1 (9s ago)   2m34s
feldera-api-server-7bb757f685-jwn5v          1/1     Running            0            2m34s

… and an explanation will be visible in its log:


$ kubectl logs -n feldera deployment/feldera-kubernetes-runner
...
FAIL: license verification failed: a valid license is required for the Enterprise version of Feldera. Please contact Feldera support (support@feldera.com) to resolve this issue.
...

To resolve this, provide a correct account id and license key during the helm installation. If you do not have a license, please reach out to us at support@feldera.com.