Deploy Helm Charts

Updated 1 day ago by Michael Cretzman

You can deploy Helm charts in standard Helm syntax in YAML from a remote Git repo, HTTP Helm Repository, or cloud storage service (Google Cloud Storage, AWS S3).

This process is also covered in the Helm CD Quickstart.

Before You Begin


  • Harness does not support AWS cross-account access for ChartMuseum and AWS S3. For example, if the Harness Delegate used to deploy charts is in AWS account A, and the S3 bucket is in AWS account B, the Harness Cloud Provider that uses this Delegate in A cannot assume the role for the B account.

Supported Platforms and Technologies

See Supported Platforms and Technologies.

ChartMuseum Binaries

Many Helm Chart users use ChartMuseum as their Helm Chart Repository server.

  • ChartMuseum binary v0.8.2: the default ChartMuseum binary used by Harness is v0.8.2.
  • ChartMuseum binary v0.12.0: to use ChartMuseum binary v0.12.0 you must enable the feature flag USE_LATEST_CHARTMUSEUM_VERSION. Contact Harness Support to enable the feature.

Review: Artifacts and Helm Charts

Harness supports image artifacts with Helm charts in the following ways.

Helm Chart with Hardcoded Artifact

The image artifact is identified in the Helm chart values.yaml file. For example:

- name: nginx

If the image is hardcoded then you do not use the Artifacts section of the Service. Any artifacts added here are ignored.

Helm Chart using Artifact Added to the Stage

You add an image artifact to the Artifacts section of the Service and then reference it in the Helm chart values.yaml file.

Artifacts in the Artifacts section are referenced using the <+artifact.image> expression. For example:

image: <+artifact.image>
pullPolicy: IfNotPresent
dockercfg: <+artifact.imagePullSecret>

This is the same method when using Artifacts with standard Kubernetes deployments. See Add Container Images as Artifacts for Kubernetes Deployments.

Step 1: Add the Helm Chart

Adding a Helm chart is a simple process of connecting Harness to the Git or HTTP Helm repo where your chart is located.

In your CD stage, click Service.

In Service Definition, select Kubernetes.

In Manifests, click Add Manifest.

In Specify Manifest Type, select Helm Chart, and click Continue.

In Specify Helm Chart Store, select the type of repo or or cloud storage service (Google Cloud Storage, AWS S3) you're using.

For the steps and settings of each option, see the Connect to an Artifact Repo How-tos.

If you are using Google Cloud Storage or Amazon S3, see Cloud Platform Connectors.

For all of the Helm Chart Store types (Git, GitHub, HTTP Helm, etc), you will need to provide the following Helm info:




Manifest Identifier

Enter a name that identifies this Helm chart. It doesn't have to be the chart name. It can be the name of the service you are deploying or another name.


Chart Name

Enter the name of the Helm chart for Harness to pull. Don't include the chart version. You will add that in the Chart Version setting.


Chart Version

Enter the version of the chart you want to deploy. This is found in the Chart.yaml version label in your chart.

You can list all available versions of a chart using the search repo command with the --versions option. See helm search repo from Helm.

If you leave Chart Version empty Harness gets the latest chart.

If you are going to use a Harness Trigger to run this Pipeline when a new version is added to your chart repo, select the Runtime Input option. When you set up the Trigger, you will select this chart and Harness will listen on the repo for new versions. See Trigger Pipelines on New Helm Chart.


Helm Version

Select the version of Helm used in your chart. See Helm Version Support Policy from Helm.

Version 2

Skip Resource Versioning

By default, Harness versions ConfigMaps and Secrets deployed into Kubernetes clusters.

In some cases, such as when using public manifests or Helm charts, you cannot add the annotation.

When you enable Skip Resource Versioning, Harness will not perform versioning of ConfigMaps and Secrets for the resource.

If you have enabled Skip Resource Versioning for a few deployments and then disable it, Harness will start versioning ConfigMaps and Secrets.

Helm Command Flags

You can use Helm command flags to extend the Helm commands that Harness runs when deploying your Helm chart.

Harness will run Helm-specific Helm commands and their flags as part of preprocessing. All the commands you select are run before helm install/upgrade.

Command Type

Select the Helm command type you want to use. For example:

  • Templatehelm template to render the helm template files.
  • Fetchhelm fetch (v1) or helm pull helm pull (v2) to get the Helm chart.



Add any options for the command.

If you haven't set up a Harness Delegate, you can add one as part of the Connector setup. This process is described in Helm CD Quickstart and Install a Kubernetes Delegate.

Once your Helm chart is added, it appears in the Manifests section. For example:

Option: Reference the Artifact

If the image artifact is not hardcoded in the Helm chart, add the artifact in Artifacts and use the expression <+artifact.image> in your values.yaml. For example:

image: <+artifact.image>
pullPolicy: IfNotPresent
dockercfg: <+artifact.imagePullSecret>

This is the same method when using Artifacts with standard Kubernetes deployments. See Add Container Images as Artifacts for Kubernetes Deployments.

Option: Override Chart Values YAML in Service

You can override the values YAML in the Helm chart by adding a values YAML in Manifests.

You add values YAML files in the same way you added your chart. You simply select Values YAML in Specify Manifest Type.

In Manifest Details, you enter the path to each values.yaml file. You can add multiple files.

If you use multiple files, priority is given from the last file to the first file. For example, if you have 3 files and the second and third files contain the same key:value as the first file, the third file's key:value overrides the second and first files.

Your values.yaml file can use Go templating and Harness built-in variable expressions in combination in your Manifests files.

See Example Kubernetes Manifests using Go Templating.

Option: Override Chart Values YAML in Environment

You can override the values YAML file for a stage's Environment by mapping the Environment name to the values file or folder. Next, you use the <> Harness expression in the values YAML path.

Let's look at an example.

Here is a repo with three values files, dev.yaml, qa.yaml. prod.yaml. In the File Path for the values file, you use the <> expression. Next, in the Environment setting, you add three Environments, one for each YAML file name.

When you select an Environment, such as qa, the name of the Environment is used in File Path and resolves to qa.yaml. At runtime, the qa.yaml values file is used, and it overrides the values.yaml file in the chart.

Instead of selecting the Environment in the Infrastructure each time, you can set the Environment as a Runtime Input and then enter dev, qa, or prod at runtime.

Option: Override Chart Values YAML at Runtime

You can make the values file path a Runtime Input and simply enter the name of the values file when you run the Pipeline.

In Manifest Details for the values file, in File Path, select Runtime Input. At runtime, you simply enter the name of the values file to use.

The values file you specify at runtime will override the values.yaml in the chart.

Step 2: Define the Infrastructure and Execution

There is nothing unique about defining the target cluster Infrastructure Definition for a Helm chart deployment. It is the same process as a typical Harness Kubernetes deployment.

See Define Your Kubernetes Target Infrastructure.

Helm charts can be deployed using any of the Execution steps and deployment strategies used in other Kubernetes deployments. See Kubernetes How-tos.

Step 3: Deploy

Each Helm chart deployment is treated as a release. During deployment, when Harness detects that there is a previous release for the chart, it upgrades the chart to the new release.

In your Pipeline, click Run.

The Helm chart deployment runs.

You will see Harness fetch the Helm chart. Here is an example:

Helm repository: Bitnami Helm Repo

Chart name: nginx

Chart version: 9.4.1

Helm version: V3

Repo url:

Successfully fetched values.yaml

Fetching files from helm chart repo

Helm repository: Bitnami Helm Repo

Chart name: nginx

Helm version: V3

Repo url:

Successfully fetched following files:

- nginx/.helmignore
- nginx/charts/common/.helmignore
- nginx/charts/common/templates/validations/_postgresql.tpl
- nginx/charts/common/templates/validations/_cassandra.tpl
- nginx/charts/common/templates/validations/_mongodb.tpl
- nginx/charts/common/templates/validations/_mariadb.tpl
- nginx/charts/common/templates/validations/_validations.tpl
- nginx/charts/common/templates/validations/_redis.tpl
- nginx/charts/common/templates/_ingress.tpl
- nginx/charts/common/templates/_names.tpl
- nginx/charts/common/templates/_affinities.tpl
- nginx/charts/common/templates/_storage.tpl
- nginx/charts/common/templates/_utils.tpl
- nginx/charts/common/templates/_errors.tpl
- nginx/charts/common/templates/_capabilities.tpl
- nginx/charts/common/templates/_secrets.tpl
- nginx/charts/common/templates/_warnings.tpl
- nginx/charts/common/templates/_tplvalues.tpl
- nginx/charts/common/templates/_images.tpl
- nginx/charts/common/templates/_labels.tpl
- nginx/charts/common/Chart.yaml
- nginx/charts/common/values.yaml
- nginx/charts/common/
- nginx/Chart.lock
- nginx/templates/svc.yaml
- nginx/templates/health-ingress.yaml
- nginx/templates/ldap-daemon-secrets.yaml
- nginx/templates/tls-secrets.yaml
- nginx/templates/NOTES.txt
- nginx/templates/pdb.yaml
- nginx/templates/ingress.yaml
- nginx/templates/server-block-configmap.yaml
- nginx/templates/serviceaccount.yaml
- nginx/templates/hpa.yaml
- nginx/templates/servicemonitor.yaml


Next, Harness will initialize and prepare the workloads, apply the Kubernetes manifests, and wait for steady state.

In Wait for Steady State you will see the workloads deployed and the pods scaled up and running (the release name has been shortened for readability):

kubectl --kubeconfig=config get events --namespace=default --output=custom-columns=KIND:involvedObject.kind,,NAMESPACE:.involvedObject.namespace,MESSAGE:.message,REASON:.reason --watch-only

kubectl --kubeconfig=config rollout status Deployment/ --namespace=default --watch=true

Status : Waiting for deployment spec update to be observed...

Event : Deployment default Scaled up replica set to 1 ScalingReplicaSet

Status : Waiting for deployment spec update to be observed...

Status : Waiting for deployment "" rollout to finish: 0 out of

Event : ReplicaSet default Created pod: SuccessfulCreate

Status : Waiting for deployment "" rollout to finish: 0 of 1 updated replicas are available...

Event : Pod default Successfully assigned default/ to gke-doc-account-default-pool-d910b20f-argz Scheduled

Event : Pod default Pulling image "" Pulling

Event : Pod default Successfully pulled image "" in 3.495150157s Pulled

Event : Pod default Created container nginx Created

Event : Pod default Started container nginx Started

Status : deployment "" successfully rolled out


You deployment is successful.

Versioning and Rollback

Helm chart deployments support versioning and rollback in the same way as standard Kubernetes deployments.

See Kubernetes Rollback.

Option: Trigger the Pipeline on a New Chart Version

You can set up a Harness Trigger to listen on the chart repo and execute the Pipeline when a new chart version appears. See Trigger Pipelines on New Helm Chart.

Please Provide Feedback