Quickstart Guide

This guide covers how you can quickly get started using Helm.


  • You must have Kubernetes installed. We recommend version 1.4.1 or later.
  • You should also have a local configured copy of kubectl.

Helm will figure out where to install Tiller by reading your Kubernetes configuration file (usually $HOME/.kube/config). This is the same file that kubectl uses.

To find out which cluster Tiller would install to, you can run kubectl config current-context or kubectl cluster-info.

$ kubectl config current-context

Install Helm

Download a binary release of the Helm client. You can use tools like homebrew, or look at the official releases page.

For more details, or for other options, see the installation guide.

Initialize Helm and Install Tiller

Once you have Helm ready, you can initialize the local CLI and also install Tiller into your Kubernetes cluster in one step:

$ helm init

This will install Tiller into the Kubernetes cluster you saw with kubectl config current-context.

TIP: Want to install into a different cluster? Use the --kube-context flag.

TIP: When you want to upgrade Tiller, just run helm init --upgrade.

Install an Example Chart

To install a chart, you can run the helm install command. Helm has several ways to find and install a chart, but the easiest is to use one of the official stable charts.

$ helm repo update              # Make sure we get the latest list of charts
$ helm install stable/mysql
Released smiling-penguin

In the example above, the stable/mysql chart was released, and the name of our new release is smiling-penguin. You get a simple idea of the features of this MySQL chart by running helm inspect stable/mysql.

Whenever you install a chart, a new release is created. So one chart can be installed multiple times into the same cluster. And each can be independently managed and upgraded.

The helm install command is a very powerful command with many capabilities. To learn more about it, check out the Using Helm Guide

Learn About Releases

It’s easy to see what has been released using Helm:

$ helm ls
NAME           	VERSION	 UPDATED                       	STATUS         	CHART
smiling-penguin	 1      	Wed Sep 28 12:59:46 2016      	DEPLOYED       	mysql-0.1.0

The helm list function will show you a list of all deployed releases.

Uninstall a Release

To uninstall a release, use the helm delete command:

$ helm delete smiling-penguin
Removed smiling-penguin

This will uninstall smiling-penguin from Kubernetes, but you will still be able to request information about that release:

$ helm status smiling-penguin

Because Helm tracks your releases even after you’ve deleted them, you can audit a cluster’s history, and even undelete a release (with helm rollback).

Reading the Help Text

To learn more about the available Helm commands, use helm help or type a command followed by the -h flag:

$ helm get -h

Installing Helm

There are two parts to Helm: The Helm client (helm) and the Helm server (Tiller). This guide shows how to install the client, and then proceeds to show two ways to install the server.

Installing the Helm Client

The Helm client can be installed either from source, or from pre-built binary releases.

From the Binary Releases

Every release of Helm provides binary releases for a variety of OSes. These binary versions can be manually downloaded and installed.

  1. Download your desired version
  2. Unpack it (tar -zxvf helm-v2.0.0-linux-amd64.tgz)
  3. Find the helm binary in the unpacked directory, and move it to its desired destination (mv linux-amd64/helm /usr/local/bin/helm)

From there, you should be able to run the client: helm help.

From Homebrew (macOS)

Members of the Kubernetes community have contributed a Helm formula build to Homebrew. This formula is generally up to date.

brew install kubernetes-helm

(Note: There is also a formula for emacs-helm, which is a different project.)

From Script

Helm now has an installer script that will automatically grab the latest version of the Helm client and install it locally.

You can fetch that script, and then execute it locally. It’s well documented so that you can read through it and understand what it is doing before you run it.

$ curl https://raw.githubusercontent.com/kubernetes/helm/master/scripts/get > get_helm.sh
$ chmod 700 get_helm.sh
$ ./get_helm.sh

Yes, you can curl ...| bash that if you want to live on the edge.

From Canary Builds

“Canary” builds are versions of the Helm software that are built from the latest master branch. They are not official releases, and may not be stable. However, they offer the opportunity to test the cutting edge features.

Canary Helm binaries are stored in the Kubernetes Helm GCS bucket. Here are links to the common builds:

From Source (Linux, macOS)

Building Helm from source is slightly more work, but is the best way to go if you want to test the latest (pre-release) Helm version.

You must have a working Go environment with glide and Mercurial installed.

$ cd $GOPATH
$ mkdir -p src/k8s.io
$ cd src/k8s.io
$ git clone https://github.com/kubernetes/helm.git
$ cd helm
$ make bootstrap build

The bootstrap target will attempt to install dependencies, rebuild the vendor/ tree, and validate configuration.

The build target will compile helm and place it in bin/helm. Tiller is also compiled, and is placed in bin/tiller.

Installing Tiller

Tiller, the server portion of Helm, typically runs inside of your Kubernetes cluster. But for development, it can also be run locally, and configured to talk to a remote Kubernetes cluster.

Easy In-Cluster Installation

The easiest way to install tiller into the cluster is simply to run helm init. This will validate that helm’s local environment is set up correctly (and set it up if necessary). Then it will connect to whatever cluster kubectl connects to by default (kubectl config view). Once it connects, it will install tiller into the kube-system namespace.

After helm init, you should be able to run kubectl get po --namespace kube-system and see Tiller running.

You can explicitly tell helm init to…

  • Install the canary build with the --canary-image flag
  • Install a particular image (version) with --tiller-image
  • Install to a particular cluster with --kube-context
  • Install into a particular namespace with --tiller-namespace

Once Tiller is installed, running helm version should show you both the client and server version. (If it shows only the client version, helm cannot yet connect to the server. Use kubectl to see if any tiller pods are running.)

If Helm will look for Tiller in the kube-system namespace unless --tiller-namespace or TILLER_NAMESPACE is set.

Installing Tiller Canary Builds

Canary images are built from the master branch. They may not be stable, but they offer you the chance to test out the latest features.

The easiest way to install a canary image is to use helm init with the --canary-image flag:

$ helm init --canary-image

This will use the most recently built container image. You can always uninstall Tiller by deleting the Tiller deployment from the kube-system namespace using kubectl.

Running Tiller Locally

For development, it is sometimes easier to work on Tiller locally, and configure it to connect to a remote Kubernetes cluster.

The process of building Tiller is explained above.

Once tiller has been built, simply start it:

$ bin/tiller
Tiller running on :44134

When Tiller is running locally, it will attempt to connect to the Kubernetes cluster that is configured by kubectl. (Run kubectl config view to see which cluster that is.)

You must tell helm to connect to this new local Tiller host instead of connecting to the one in-cluster. There are two ways to do this. The first is to specify the --host option on the command line. The second is to set the $HELM_HOST environment variable.

$ export HELM_HOST=localhost:44134
$ helm version # Should connect to localhost.
Client: &version.Version{SemVer:"v2.0.0-alpha.4", GitCommit:"db...", GitTreeState:"dirty"}
Server: &version.Version{SemVer:"v2.0.0-alpha.4", GitCommit:"a5...", GitTreeState:"dirty"}

Importantly, even when running locally, Tiller will store release configuration in ConfigMaps inside of Kubernetes.

Upgrading Tiller

As of Helm 2.2.0, Tiller can be upgraded using helm init --upgrade.

For older versions of Helm, or for manual upgrades, you can use kubectl to modify the Tiller image:

$ export TILLER_TAG=v2.0.0-beta.1        # Or whatever version you want
$ kubectl --namespace=kube-system set image deployments/tiller-deploy tiller=gcr.io/kubernetes-helm/tiller:$TILLER_TAG
deployment "tiller-deploy" image updated

Setting TILLER_TAG=canary will get the latest snapshot of master.

Deleting or Reinstalling Tiller

Because Tiller stores its data in Kubernetes ConfigMaps, you can safely delete and re-install Tiller without worrying about losing any data. The recommended way of deleting Tiller is with kubectl delete deployment tiller-deploy --namespace kube-system, or more concisely helm reset.

Tiller can then be re-installed from the client with:

$ helm init


In most cases, installation is as simple as getting a pre-built helm binary and running helm init. This document covers additional cases for those who want to do more sophisticated things with Helm.

Once you have the Helm Client and Tiller successfully installed, you can move on to using Helm to manage charts.

Kubernetes Distribution Guide

This document captures information about using Helm in specific Kubernetes environments.

We are trying to add more details to this document. Please contribute via Pull Requests if you can.


Helm is tested and known to work with minikube. It requires no additional configuration.

scripts/local-cluster and Hyperkube

Hyperkube configured via scripts/local-cluster.sh is known to work. For raw Hyperkube you may need to do some manual configuration.


Google’s GKE hosted Kubernetes platform is known to work with Helm, and requires no additional configuration.

Ubuntu with ‘kubeadm’

Kubernetes bootstrapped with kubeadm is known to work on the following Linux distributions:

  • Ubuntu 16.04

Some versions of Helm (v2.0.0-beta2) require you to export KUBECONFIG=/etc/kubernetes/admin.conf or create a ~/.kube/config.

Container Linux by CoreOS

Helm requires that kubelet have access to a copy of the socat program to proxy connections to the Tiller API. On Container Linux the Kubelet runs inside of a hyperkube container image that has socat. So, even though Container Linux doesn’t ship socat the container filesystem running kubelet does have socat. To learn more read the Kubelet Wrapper docs.

Installation: Frequently Asked Questions

This section tracks some of the more frequently encountered issues with installing or getting started with Helm.

We’d love your help making this document better. To add, correct, or remove information, file an issue or send us a pull request.


I want to know more about my downloading options.

Q: I can’t get to GitHub releases of the newest Helm. Where are they?

A: We no longer use GitHub releases. Binaries are now stored in a GCS public bucket.

Q: Why aren’t there Debian/Fedora/… native packages of Helm?

We’d love to provide these or point you toward a trusted provider. If you’re interested in helping, we’d love it. This is how the Homebrew formula was started.

Q: Why do you provide a curl ...|bash script?

A: There is a script in our repository (scripts/get) that can be executed as a curl ..|bash script. The transfers are all protected by HTTPS, and the script does some auditing of the packages it fetches. However, the script has all the usual dangers of any shell script.

We provide it because it is useful, but we suggest that users carefully read the script first. What we’d really like, though, are better packaged releases of Helm.


I’m trying to install Helm/Tiller, but something is not right.

Q: How do I put the Helm client files somewhere other than ~/.helm?

Set the $HELM_HOME environment variable, and then run helm init:

export HELM_HOME=/some/path
helm init --client-only

Note that if you have existing repositories, you will need to re-add them with helm repo add....

Q: How do I configure Helm, but not install Tiller?

A: By default, helm init will ensure that the local $HELM_HOME is configured, and then install Tiller on your cluster. To locally configure, but not install Tiller, use helm init --client-only.

Q: How do I manually install Tiller on the cluster?

A: Tiller is installed as a Kubernetes deployment. You can get the manifest by running helm init --dry-run --debug, and then manually install it with kubectl. It is suggested that you do not remove or change the labels on that deployment, as they are sometimes used by supporting scripts and tools.

Q: Why do I get Error response from daemon: target is unknown during Tiller install?

A: Users have reported being unable to install Tiller on Kubernetes instances that are using Docker 1.13.0. The root cause of this was a bug in Docker that made that one version incompatible with images pushed to the Docker registry by earlier versions of Docker.

This issue was fixed shortly after the release, and is available in Docker 1.13.1-RC1 and later.

Getting Started

I successfully installed Helm/Tiller but I can’t use it.

Q: Trying to use Helm, I get the error “client transport was broken”

E1014 02:26:32.885226   16143 portforward.go:329] an error occurred forwarding 37008 -> 44134: error forwarding port 44134 to pod tiller-deploy-2117266891-e4lev_kube-system, uid : unable to do port forwarding: socat not found.
2016/10/14 02:26:32 transport: http2Client.notifyError got notified that the client transport was broken EOF.
Error: transport is closing

A: This is usually a good indication that Kubernetes is not set up to allow port forwarding.

Typically, the missing piece is socat. If you are running CoreOS, we have been told that it may have been misconfigured on installation. The CoreOS team recommends reading this:

Here are a few resolved issues that may help you get started:

Q: Trying to use Helm, I get the error “lookup XXXXX on no such host”

Error: Error forwarding ports: error upgrading connection: dial tcp: lookup kube-4gb-lon1-02 on no such host

A: We have seen this issue with Ubuntu and Kubeadm in multi-node clusters. The issue is that the nodes expect certain DNS records to be obtainable via global DNS. Until this is resolved upstream, you can work around the issue as follows:

1) Add entries to /etc/hosts on the master mapping your hostnames to their public IPs 2) Install dnsmasq on the master (e.g. apt install -y dnsmasq) 3) Kill the k8s api server container on master (kubelet will recreate it) 4) Then systemctl restart docker (or reboot the master) for it to pick up the /etc/resolv.conf changes

See this issue for more information: https://github.com/kubernetes/helm/issues/1455

Q: On GKE (Google Container Engine) I get “No SSH tunnels currently open”

Error: Error forwarding ports: error upgrading connection: No SSH tunnels currently open. Were the targets able to accept an ssh-key for user "gke-[redacted]"?

Another variation of the error message is:

Unable to connect to the server: x509: certificate signed by unknown authority

A: The issue is that your local Kubernetes config file must have the correct credentials.

When you create a cluster on GKE, it will give you credentials, including SSL certificates and certificate authorities. These need to be stored in a Kubernetes config file (Default: ~/.kube/config so that kubectl and helm can access them.

Q: When I run a Helm command, I get an error about the tunnel or proxy

A: Helm uses the Kubernetes proxy service to connect to the Tiller server. If the command kubectl proxy does not work for you, neither will Helm. Typically, the error is related to a missing socat service.

Q: Tiller crashes with a panic

When I run a command on Helm, Tiller crashes with an error like this:

Tiller is listening on :44134
Probes server is listening on :44135
Storage driver is ConfigMap
Cannot initialize Kubernetes connection: the server has asked for the client to provide credentials 2016-12-20 15:18:40.545739 I | storage.go:37: Getting release "bailing-chinchilla" (v1) from storage
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x8053d5]

goroutine 77 [running]:
panic(0x1abbfc0, 0xc42000a040)
        /usr/local/go/src/runtime/panic.go:500 +0x1a1
k8s.io/helm/vendor/k8s.io/kubernetes/pkg/client/unversioned.(*ConfigMaps).Get(0xc4200c6200, 0xc420536100, 0x15, 0x1ca7431, 0x6, 0xc42016b6a0)
        /home/ubuntu/.go_workspace/src/k8s.io/helm/vendor/k8s.io/kubernetes/pkg/client/unversioned/configmap.go:58 +0x75
k8s.io/helm/pkg/storage/driver.(*ConfigMaps).Get(0xc4201d6190, 0xc420536100, 0x15, 0xc420536100, 0x15, 0xc4205360c0)
        /home/ubuntu/.go_workspace/src/k8s.io/helm/pkg/storage/driver/cfgmaps.go:69 +0x62
k8s.io/helm/pkg/storage.(*Storage).Get(0xc4201d61a0, 0xc4205360c0, 0x12, 0xc400000001, 0x12, 0x0, 0xc420200070)
        /home/ubuntu/.go_workspace/src/k8s.io/helm/pkg/storage/storage.go:38 +0x160
k8s.io/helm/pkg/tiller.(*ReleaseServer).uniqName(0xc42002a000, 0x0, 0x0, 0xc42016b800, 0xd66a13, 0xc42055a040, 0xc420558050, 0xc420122001)
        /home/ubuntu/.go_workspace/src/k8s.io/helm/pkg/tiller/release_server.go:577 +0xd7
k8s.io/helm/pkg/tiller.(*ReleaseServer).prepareRelease(0xc42002a000, 0xc42027c1e0, 0xc42002a001, 0xc42016bad0, 0xc42016ba08)
        /home/ubuntu/.go_workspace/src/k8s.io/helm/pkg/tiller/release_server.go:630 +0x71
k8s.io/helm/pkg/tiller.(*ReleaseServer).InstallRelease(0xc42002a000, 0x7f284c434068, 0xc420250c00, 0xc42027c1e0, 0x0, 0x31a9, 0x31a9)
        /home/ubuntu/.go_workspace/src/k8s.io/helm/pkg/tiller/release_server.go:604 +0x78
k8s.io/helm/pkg/proto/hapi/services._ReleaseService_InstallRelease_Handler(0x1c51f80, 0xc42002a000, 0x7f284c434068, 0xc420250c00, 0xc42027c190, 0x0, 0x0, 0x0, 0x0, 0x0)
        /home/ubuntu/.go_workspace/src/k8s.io/helm/pkg/proto/hapi/services/tiller.pb.go:747 +0x27d
k8s.io/helm/vendor/google.golang.org/grpc.(*Server).processUnaryRPC(0xc4202f3ea0, 0x28610a0, 0xc420078000, 0xc420264690, 0xc420166150, 0x288cbe8, 0xc420250bd0, 0x0, 0x0)
        /home/ubuntu/.go_workspace/src/k8s.io/helm/vendor/google.golang.org/grpc/server.go:608 +0xc50
k8s.io/helm/vendor/google.golang.org/grpc.(*Server).handleStream(0xc4202f3ea0, 0x28610a0, 0xc420078000, 0xc420264690, 0xc420250bd0)
        /home/ubuntu/.go_workspace/src/k8s.io/helm/vendor/google.golang.org/grpc/server.go:766 +0x6b0
k8s.io/helm/vendor/google.golang.org/grpc.(*Server).serveStreams.func1.1(0xc420124710, 0xc4202f3ea0, 0x28610a0, 0xc420078000, 0xc420264690)
        /home/ubuntu/.go_workspace/src/k8s.io/helm/vendor/google.golang.org/grpc/server.go:419 +0xab
created by k8s.io/helm/vendor/google.golang.org/grpc.(*Server).serveStreams.func1
        /home/ubuntu/.go_workspace/src/k8s.io/helm/vendor/google.golang.org/grpc/server.go:420 +0xa3

A: Check your security settings for Kubernetes.

A panic in Tiller is almost always the result of a failure to negotiate with the Kubernetes API server (at which point Tiller can no longer do anything useful, so it panics and exits).

Often, this is a result of authentication failing because the Pod in which Tiller is running does not have the right token.

To fix this, you will need to change your Kubernetes configuration. Make sure that --service-account-private-key-file from controller-manager and --service-account-key-file from apiserver point to the same x509 RSA key.


My Helm used to work, then I upgrade. Now it is broken.

Q: After upgrade, I get the error “Client version is incompatible”. What’s wrong?

Tiller and Helm have to negotiate a common version to make sure that they can safely communicate without breaking API assumptions. That error means that the version difference is too great to safely continue. Typically, you need to upgrade Tiller manually for this.

The Installation Guide has definitive information about safely upgrading Helm and Tiller.

The rules for version numbers are as follows:

  • Pre-release versions are incompatible with everything else. Alpha.1 is incompatible with Alpha.2.
  • Patch revisions are compatible: 1.2.3 is compatible with 1.2.4
  • Minor revisions are not compatible: 1.2.0 is not compatible with 1.3.0, though we may relax this constraint in the future.
  • Major revisions are not compatible: 1.0.0 is not compatible with 2.0.0.


I am trying to remove stuff.

Q: When I delete the Tiller deployment, how come all the releases are still there?

Releases are stored in ConfigMaps inside of the kube-system namespace. You will have to manually delete them to get rid of the record.

Q: I want to delete my local Helm. Where are all its files?

Along with the helm binary, Helm stores some files in $HELM_HOME, which is located by default in ~/.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
home: https://mariadb.org
- 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/k8s.io/helm/mariadb-0.3.0.tgz
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/k8s.io/helm/mariadb-0.3.0.tgz
## Bitnami MariaDB image version
## ref: https://hub.docker.com/r/bitnami/mariadb/tags/
## Default: none
imageTag: 10.1.14-r3

## Specify a imagePullPolicy
## Default to 'Always' if imageTag is 'latest', else set to 'IfNotPresent'
## ref: http://kubernetes.io/docs/user-guide/images/#pre-pulling-images
# imagePullPolicy:

## Specify password for root user
## ref: https://github.com/bitnami/bitnami-docker-mariadb/blob/master/README.md#setting-the-root-password-on-first-run
# mariadbRootPassword:

## Create a database user
## ref: https://github.com/bitnami/bitnami-docker-mariadb/blob/master/README.md#creating-a-database-user-on-first-run
# mariadbUser:
# mariadbPassword:

## Create a database
## ref: https://github.com/bitnami/bitnami-docker-mariadb/blob/master/README.md#creating-a-database-on-first-run
# mariadbDatabase:

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

$ echo 'mariadbUser: user0' > config.yaml
$ helm install -f config.yaml stable/mariadb

The above will set the default MariaDB user to user0, but 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: Specify overrides on the command line.

If both are used, --set values are merged into --values with higher precedence.

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

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:

  kubernetes.io/role: master

The --set syntax is not as expressive as YAML, especially when it comes to collections. And there is currently no method for expressing things such as “set the third item in a list to…“.

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 https://example.com/charts/foo-1.2.3.tgz)

‘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/k8s.io/helm/mariadb-0.3.0.tgz
happy-panda has been upgraded. Happy Helming!
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 and 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
stable         	https://kubernetes-charts.storage.googleapis.com
local          	http://localhost:8879/charts
mumoshu        	https://mumoshu.github.io/charts

And new repositories can be added with helm repo add:

$ helm repo add dev https://example.com/dev-charts

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 Kubernetes Charts GitHub repository. That project accepts chart source code, and (after audit) packages those for you.


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.

The Helm Plugins Guide

Helm 2.1.0 introduced the concept of a client-side Helm plugin. A plugin is a tool that can be accessed through the helm CLI, but which is not part of the built-in Helm codebase.

Existing plugins can be found on related section or by searching Github.

This guide explains how to use and create plugins.

An Overview

Helm plugins are add-on tools that integrate seamlessly with Helm. They provide a way to extend the core feature set of Helm, but without requiring every new feature to be written in Go and added to the core tool.

Helm plugins have the following features:

  • They can be added and removed from a Helm installation without impacting the core Helm tool.
  • They can be written in any programming language.
  • They integrate with Helm, and will show up in helm help and other places.

Helm plugins live in $(helm home)/plugins.

The Helm plugin model is partially modeled on Git’s plugin model. To that end, you may sometimes hear helm referred to as the porcelain layer, with plugins being the plumbing. This is a shorthand way of suggesting that Helm provides the user experience and top level processing logic, while the plugins do the “detail work” of performing a desired action.

Installing a Plugin

A Helm plugin management system is in the works. But in the short term, plugins are installed by copying the plugin directory into $(helm home)/plugins.

$ cp -a myplugin/ $(helm home)/plugins/

If you have a plugin tar distribution, simply untar the plugin into the $(helm home)/plugins directory.

Building Plugins

In many ways, a plugin is similar to a chart. Each plugin has a top-level directory, and then a plugin.yaml file.

$(helm home)/plugins/
  |- keybase/
      |- plugin.yaml
      |- keybase.sh

In the example above, the keybase plugin is contained inside of a directory named keybase. It has two files: plugin.yaml (required) and an executable script, keybase.sh (optional).

The core of a plugin is a simple YAML file named plugin.yaml. Here is a plugin YAML for a plugin that adds support for Keybase operations:

name: "keybase"
version: "0.1.0"
usage: "Integreate Keybase.io tools with Helm"
description: |-
  This plugin provides Keybase services to Helm.
ignoreFlags: false
useTunnel: false
command: "$HELM_PLUGIN_DIR/keybase.sh"

The name is the name of the plugin. When Helm executes it plugin, this is the name it will use (e.g. helm NAME will invoke this plugin).

name should match the directory name. In our example above, that means the plugin with name: keybase should be contained in a directory named keybase.

Restrictions on name:

  • name cannot duplicate one of the existing helm top-level commands.
  • name must be restricted to the characters ASCII a-z, A-Z, 0-9, _ and -.

version is the SemVer 2 version of the plugin. usage and description are both used to generate the help text of a command.

The ignoreFlags switch tells Helm to not pass flags to the plugin. So if a plugin is called with helm myplugin --foo and ignoreFlags: true, then --foo is silently discarded.

The useTunnel switch indicates that the plugin needs a tunnel to Tiller. This should be set to true anytime a plugin talks to Tiller. It will cause Helm to open a tunnel, and then set $TILLER_HOST to the right local address for that tunnel. But don’t worry: if Helm detects that a tunnel is not necessary because Tiller is running locally, it will not create the tunnel.

Finally, and most importantly, command is the command that this plugin will execute when it is called. Environment variables are interpolated before the plugin is executed. The pattern above illustrates the preferred way to indicate where the plugin program lives.

There are some strategies for working with plugin commands:

  • If a plugin includes an executable, the executable for a command: should be packaged in the plugin directory.
  • The command: line will have any environment variables expanded before execution. $HELM_PLUGIN_DIR will point to the plugin directory.
  • The command itself is not executed in a shell. So you can’t oneline a shell script.
  • Helm injects lots of configuration into environment variables. Take a look at the environment to see what information is available.
  • Helm makes no assumptions about the language of the plugin. You can write it in whatever you prefer.
  • Commands are responsible for implementing specific help text for -h and --help. Helm will use usage and description for helm help and helm help myplugin, but will not handle helm myplugin --help.

Environment Variables

When Helm executes a plugin, it passes the outer environment to the plugin, and also injects some additional environment variables.

Variables like KUBECONFIG are set for the plugin if they are set in the outer environment.

The following variables are guaranteed to be set:

  • HELM_PLUGIN: The path to the plugins directory
  • HELM_PLUGIN_NAME: The name of the plugin, as invoked by helm. So helm myplug will have the short name myplug.
  • HELM_PLUGIN_DIR: The directory that contains the plugin.
  • HELM_BIN: The path to the helm command (as executed by the user).
  • HELM_HOME: The path to the Helm home.
  • HELM_PATH_*: Paths to important Helm files and directories are stored in environment variables prefixed by HELM_PATH.
  • TILLER_HOST: The domain:port to Tiller. If a tunnel is created, this will point to the local endpoint for the tunnel. Otherwise, it will point to $HELM_HOST, --host, or the default host (according to Helm’s rules of precedence).

While HELM_HOST may be set, there is no guarantee that it will point to the correct Tiller instance. This is done to allow plugin developer to access HELM_HOST in its raw state when the plugin itself needs to manually configure a connection.

A Note on useTunnel

If a plugin specifies useTunnel: true, Helm will do the following (in order):

  1. Parse global flags and the environment
  2. Create the tunnel
  4. Execute the plugin
  5. Close the tunnel

The tunnel is removed as soon as the command returns. So, for example, a command cannot background a process and assume that that process will be able to use the tunnel.

A Note on Flag Parsing

When executing a plugin, Helm will parse global flags for its own use. Some of these flags are not passed on to the plugin.

  • --debug: If this is specified, $HELM_DEBUG is set to 1
  • --home: This is converted to $HELM_HOME
  • --host: This is converted to $HELM_HOST
  • --kube-context: This is simply dropped. If your plugin uses useTunnel, this is used to set up the tunnel for you.

Plugins should display help text and then exit for -h and --help. In all other cases, plugins may use flags as appropriate.