Redpanda Helm Chart

The Redpanda Helm chart deploys a Redpanda cluster in Kubernetes, and provides the following features:

Schema registry (enabled by default)
REST (aka PandaProxy, enabled by default)
TLS
SASL
External access

Requirements

Helm version 3.6.0 or later
Kubernetes version 1.21.0 or later
Cert-manager 1.9.0 or later (required for TLS support only)

Installation

To get started, see the Redpanda documentation.

For examples of using the Helm chart, see the examples/ directory. Each example focuses on a specific feature.

Configuration

The Redpanda Helm chart is configured in the values.yaml file. To customize your deployment, you can override the default values in your own YAML file with the --values option or in the command line with the --set option. For example, you can do the following:

Specify which Kubernetes components to deploy.
Configure the deployed Kubernetes components.

To learn how to override the default values in the values.yaml file, see the Helm documentation.

All configuration options for the Redpanda Helm chart are documented in the values.yaml file.

Upgrading Chart

helm upgrade [RELEASE_NAME] redpanda/redpanda

From 2.6.x onwards

In order to enable dedicated persistent volume for tiered storage cache, the storage.tieredStoragePersistentVolume.enabled need to be set to true. The helm upgrade will fail with the following error.

helm upgrade --namespace redpanda redpanda/redpanda \
  --set storage.tieredStoragePersistentVolume.enabled=true \
  --set storage.tieredConfig.cloud_storage_enabled=true \
  --set storage.tieredConfig.cloud_storage_cache_directory=/some/path/for-tiered-storage
Error: UPGRADE FAILED: cannot patch "redpanda" with kind StatefulSet: StatefulSet.apps "redpanda" is invalid: spec: Forbidden: updates to statefulset spec for fields other than 'replicas', 'template', 'updateStrategy', 'persistentVolumeClaimRetentionPolicy' and 'minReadySeconds' are forbidden

To be able to add PersistentVolume for tiered storage cache please delete StatefulSet with cascade orphan option to leave Pods behind.

kubectl delete sts redpanda --cascade=orphan

The helm upgrade should be able to succeed, but you need to manually do rolling update starting from ordinal 0 to ordinal N.

kubectl delete pod redpanda-0

Please wait for deleted Pod to be restarted and become ready in order to move to next Pod.

Contributing

If you have improvements that can be made to this Helm chart, please consider becoming a contributor. To contribute to the Helm chart, see our contribution guidelines.