Using Helm

This guide explains the basics of using Helm (and Tiller) to manage packages on your Kubernetes cluster. It assumes that you have already installed the Helm client and the Tiller server (typically by helm init).

If you are simply interested in running a few quick commands, you may wish to begin with the Quickstart Guide. This chapter covers the particulars of Helm commands, and explains how to use Helm.

Three Big Concepts

A Chart is a Helm package. It contains all of the resource definitions necessary to run an application, tool, or service inside of a Kubernetes cluster. Think of it like the Kubernetes equivalent of a Homebrew formula, an Apt dpkg, or a Yum RPM file.

A Repository is the place where charts can be collected and shared. It’s like Perl’s CPAN archive or the Fedora Package Database, but for Kubernetes packages.

A Release is an instance of a chart running in a Kubernetes cluster. One chart can often be installed many times into the same cluster. And each time it is installed, a new release is created. Consider a MySQL chart. If you want two databases running in your cluster, you can install that chart twice. Each one will have its own release, which will in turn have its own release name.

With these concepts in mind, we can now explain Helm like this:

Helm installs charts into Kubernetes, creating a new release for each installation. And to find new charts, you can search Helm chart repositories.

‘helm search’: Finding Charts

When you first install Helm, it is preconfigured to talk to the official Kubernetes charts repository. This repository contains a number of carefully curated and maintained charts. This chart repository is named stable by default.

You can see which charts are available by running helm search:

$ helm search
stable/drupal   	0.3.2   	One of the most versatile open source content m...
stable/jenkins  	0.1.0   	A Jenkins Helm chart for Kubernetes.
stable/mariadb  	0.5.1   	Chart for MariaDB
stable/mysql    	0.1.0   	Chart for MySQL

With no filter, helm search shows you all of the available charts. You can narrow down your results by searching with a filter:

$ helm search mysql
stable/mysql  	0.1.0  	Chart for MySQL
stable/mariadb	0.5.1  	Chart for MariaDB

Now you will only see the results that match your filter.

Why is mariadb in the list? Because its package description relates it to MySQL. We can use helm inspect chart to see this:

$ helm inspect stable/mariadb
Fetched stable/mariadb to mariadb-0.5.1.tgz
description: Chart for MariaDB
engine: gotpl
- mariadb
- mysql
- database
- sql

Search is a good way to find available packages. Once you have found a package you want to install, you can use helm install to install it.

‘helm install’: Installing a Package

To install a new package, use the helm install command. At its simplest, it takes only one argument: The name of the chart.

$ helm install stable/mariadb
Fetched stable/mariadb-0.3.0 to /Users/mattbutcher/Code/Go/src/
NAME: happy-panda
LAST DEPLOYED: Wed Sep 28 12:32:28 2016
NAMESPACE: default

==> extensions/Deployment
happy-panda-mariadb   1         0         0            0           1s

==> v1/Secret
NAME                     TYPE      DATA      AGE
happy-panda-mariadb   Opaque    2         1s

==> v1/Service
NAME                     CLUSTER-IP   EXTERNAL-IP   PORT(S)    AGE
happy-panda-mariadb    <none>        3306/TCP   1s

MariaDB can be accessed via port 3306 on the following DNS name from within your cluster:

To connect to your database run the following command:

   kubectl run happy-panda-mariadb-client --rm --tty -i --image bitnami/mariadb --command -- mysql -h happy-panda-mariadb

Now the mariadb chart is installed. Note that installing a chart creates a new release object. The release above is named happy-panda. (If you want to use your own release name, simply use the --name flag on helm install.)

During installation, the helm client will print useful information about which resources were created, what the state of the release is, and also whether there are additional configuration steps you can or should take.

Helm does not wait until all of the resources are running before it exits. Many charts require Docker images that are over 600M in size, and may take a long time to install into the cluster.

To keep track of a release’s state, or to re-read configuration information, you can use helm status:

$ helm status happy-panda
Last Deployed: Wed Sep 28 12:32:28 2016
Namespace: default

==> v1/Service
NAME                     CLUSTER-IP   EXTERNAL-IP   PORT(S)    AGE
happy-panda-mariadb    <none>        3306/TCP   4m

==> extensions/Deployment
happy-panda-mariadb   1         1         1            1           4m

==> v1/Secret
NAME                     TYPE      DATA      AGE
happy-panda-mariadb   Opaque    2         4m

MariaDB can be accessed via port 3306 on the following DNS name from within your cluster:

To connect to your database run the following command:

   kubectl run happy-panda-mariadb-client --rm --tty -i --image bitnami/mariadb --command -- mysql -h happy-panda-mariadb

The above shows the current state of your release.

Customizing the Chart Before Installing

Installing the way we have here will only use the default configuration options for this chart. Many times, you will want to customize the chart to use your preferred configuration.

To see what options are configurable on a chart, use helm inspect values:

helm inspect values stable/mariadb
Fetched stable/mariadb-0.3.0.tgz to /Users/mattbutcher/Code/Go/src/
## Bitnami MariaDB image version
## ref:
## Default: none
imageTag: 10.1.14-r3

## Specify a imagePullPolicy
## Default to 'Always' if imageTag is 'latest', else set to 'IfNotPresent'
## ref:
# imagePullPolicy:

## Specify password for root user
## ref:
# mariadbRootPassword:

## Create a database user
## ref:
# mariadbUser:
# mariadbPassword:

## Create a database
## ref:
# mariadbDatabase:

You can then override any of these settings in a YAML formatted file, and then pass that file during installation.

$ cat << EOF > config.yaml
mariadbUser: user0
mariadbDatabase: user0db
$ helm install -f config.yaml stable/mariadb

The above will create a default MariaDB user with the name user0, and grant this user access to a newly created user0db database, but will accept all the rest of the defaults for that chart.

There are two ways to pass configuration data during install:

  • --values (or -f): Specify a YAML file with overrides. This can be specified multiple times and the rightmost file will take precedence
  • --set (and its variants --set-string and --set-file): Specify overrides on the command line.

If both are used, --set values are merged into --values with higher precedence. Overrides specified with --set are persisted in a configmap. Values that have been --set can be viewed for a given release with helm get values <release-name>. Values that have been --set can be cleared by running helm upgrade with --reset-values specified.

The Format and Limitations of --set

The --set option takes zero or more name/value pairs. At its simplest, it is used like this: --set name=value. The YAML equivalent of that is:

name: value

Multiple values are separated by , characters. So --set a=b,c=d becomes:

a: b
c: d

More complex expressions are supported. For example, --set outer.inner=value is translated into this:

  inner: value

Lists can be expressed by enclosing values in { and }. For example, --set name={a, b, c} translates to:

  - a
  - b
  - c

As of Helm 2.5.0, it is possible to access list items using an array index syntax. For example, --set servers[0].port=80 becomes:

  - port: 80

Multiple values can be set this way. The line --set servers[0].port=80,servers[0].host=example becomes:

  - port: 80
    host: example

Sometimes you need to use special characters in your --set lines. You can use a backslash to escape the characters; --set name="value1\,value2" will become:

name: "value1,value2"

Similarly, you can escape dot sequences as well, which may come in handy when charts use the toYaml function to parse annotations, labels and node selectors. The syntax for --set nodeSelector."kubernetes\.io/role"=master becomes:

nodeSelector: master

Deeply nested data structures can be difficult to express using --set. Chart designers are encouraged to consider the --set usage when designing the format of a values.yaml file.

Helm will cast certain values specified with --set to integers. For example, --set foo=true results Helm to cast true into an int64 value. In case you want a string, use a --set’s variant named --set-string. --set-string foo=true results in a string value of "true".

--set-file key=filepath is another variant of --set. It reads the file and use its content as a value. An example use case of it is to inject a multi-line text into values without dealing with indentation in YAML. Say you want to create a brigade project with certain value containing 5 lines JavaScript code, you might write a values.yaml like:

defaultScript: |
  const { events, Job } = require("brigadier")
  function run(e, project) {
    console.log("hello default script")
  events.on("run", run)

Being embedded in a YAML, this makes it harder for you to use IDE features and testing framework and so on that supports writing code. Instead, you can use --set-file defaultScript=brigade.js with brigade.js containing:

const { events, Job } = require("brigadier")
function run(e, project) {
  console.log("hello default script")
events.on("run", run)

More Installation Methods

The helm install command can install from several sources:

  • A chart repository (as we’ve seen above)
  • A local chart archive (helm install foo-0.1.1.tgz)
  • An unpacked chart directory (helm install path/to/foo)
  • A full URL (helm install

‘helm upgrade’ and ‘helm rollback’: Upgrading a Release, and Recovering on Failure

When a new version of a chart is released, or when you want to change the configuration of your release, you can use the helm upgrade command.

An upgrade takes an existing release and upgrades it according to the information you provide. Because Kubernetes charts can be large and complex, Helm tries to perform the least invasive upgrade. It will only update things that have changed since the last release.

$ helm upgrade -f panda.yaml happy-panda stable/mariadb
Fetched stable/mariadb-0.3.0.tgz to /Users/mattbutcher/Code/Go/src/
happy-panda has been upgraded.
Last Deployed: Wed Sep 28 12:47:54 2016
Namespace: default

In the above case, the happy-panda release is upgraded with the same chart, but with a new YAML file:

mariadbUser: user1

We can use helm get values to see whether that new setting took effect.

$ helm get values happy-panda
mariadbUser: user1

The helm get command is a useful tool for looking at a release in the cluster. And as we can see above, it shows that our new values from panda.yaml were deployed to the cluster.

Now, if something does not go as planned during a release, it is easy to roll back to a previous release using helm rollback [RELEASE] [REVISION].

$ helm rollback happy-panda 1

The above rolls back our happy-panda to its very first release version. A release version is an incremental revision. Every time an install, upgrade, or rollback happens, the revision number is incremented by 1. The first revision number is always 1. And we can use helm history [RELEASE] to see revision numbers for a certain release.

Helpful Options for Install/Upgrade/Rollback

There are several other helpful options you can specify for customizing the behavior of Helm during an install/upgrade/rollback. Please note that this is not a full list of cli flags. To see a description of all flags, just run helm <command> --help.

  • --timeout: A value in seconds to wait for Kubernetes commands to complete This defaults to 300 (5 minutes)
  • --wait: Waits until all Pods are in a ready state, PVCs are bound, Deployments have minimum (Desired minus maxUnavailable) Pods in ready state and Services have an IP address (and Ingress if a LoadBalancer) before marking the release as successful. It will wait for as long as the --timeout value. If timeout is reached, the release will be marked as FAILED. Note: In scenario where Deployment has replicas set to 1 and maxUnavailable is not set to 0 as part of rolling update strategy, --wait will return as ready as it has satisfied the minimum Pod in ready condition.
  • --no-hooks: This skips running hooks for the command
  • --recreate-pods (only available for upgrade and rollback): This flag will cause all pods to be recreated (with the exception of pods belonging to deployments)

‘helm delete’: Deleting a Release

When it is time to uninstall or delete a release from the cluster, use the helm delete command:

$ helm delete happy-panda

This will remove the release from the cluster. You can see all of your currently deployed releases with the helm list command:

$ helm list
NAME           	VERSION	UPDATED                        	STATUS         	CHART
inky-cat       	1      	Wed Sep 28 12:59:46 2016       	DEPLOYED       	alpine-0.1.0

From the output above, we can see that the happy-panda release was deleted.

However, Helm always keeps records of what releases happened. Need to see the deleted releases? helm list --deleted shows those, and helm list --all shows all of the releases (deleted and currently deployed, as well as releases that failed):

⇒  helm list --all
NAME           	VERSION	UPDATED                        	STATUS         	CHART
happy-panda   	2      	Wed Sep 28 12:47:54 2016       	DELETED        	mariadb-0.3.0
inky-cat       	1      	Wed Sep 28 12:59:46 2016       	DEPLOYED       	alpine-0.1.0
kindred-angelf 	2      	Tue Sep 27 16:16:10 2016       	DELETED        	alpine-0.1.0

Because Helm keeps records of deleted releases, a release name cannot be re-used. (If you really need to re-use a release name, you can use the --replace flag, but it will simply re-use the existing release and replace its resources.)

Note that because releases are preserved in this way, you can rollback a deleted resource, and have it re-activate.

‘helm repo’: Working with Repositories

So far, we’ve been installing charts only from the stable repository. But you can configure helm to use other repositories. Helm provides several repository tools under the helm repo command.

You can see which repositories are configured using helm repo list:

$ helm repo list
NAME           	URL
local          	http://localhost:8879/charts

And new repositories can be added with helm repo add:

$ helm repo add dev

Because chart repositories change frequently, at any point you can make sure your Helm client is up to date by running helm repo update.

Creating Your Own Charts

The Chart Development Guide explains how to develop your own charts. But you can get started quickly by using the helm create command:

$ helm create deis-workflow
Creating deis-workflow

Now there is a chart in ./deis-workflow. You can edit it and create your own templates.

As you edit your chart, you can validate that it is well-formatted by running helm lint.

When it’s time to package the chart up for distribution, you can run the helm package command:

$ helm package deis-workflow

And that chart can now easily be installed by helm install:

$ helm install ./deis-workflow-0.1.0.tgz

Charts that are archived can be loaded into chart repositories. See the documentation for your chart repository server to learn how to upload.

Note: The stable repository is managed on the Helm Charts GitHub repository. That project accepts chart source code, and (after audit) packages those for you.

Tiller, Namespaces and RBAC

In some cases you may wish to scope Tiller or deploy multiple Tillers to a single cluster. Here are some best practices when operating in those circumstances.

  1. Tiller can be installed into any namespace. By default, it is installed into kube-system. You can run multiple Tillers provided they each run in their own namespace.
  2. Limiting Tiller to only be able to install into specific namespaces and/or resource types is controlled by Kubernetes RBAC roles and rolebindings. You can add a service account to Tiller when configuring Helm via helm init --service-account <NAME>. You can find more information about that here.
  3. Release names are unique PER TILLER INSTANCE.
  4. Charts should only contain resources that exist in a single namespace.
  5. It is not recommended to have multiple Tillers configured to manage resources in the same namespace.


This chapter has covered the basic usage patterns of the helm client, including searching, installation, upgrading, and deleting. It has also covered useful utility commands like helm status, helm get, and helm repo.

For more information on these commands, take a look at Helm’s built-in help: helm help.

In the next chapter, we look at the process of developing charts.