How to Install Apps on Kubernetes with Helm 3

Helm is a tool that assists with installing and managing applications on Kubernetes clusters. It is often referred to as “the package manager for Kubernetes,” and it provides functions that are similar to a package manager for an operating system:

• Helm prescribes a common format and directory structure for packaging your Kubernetes resources, known as a Helm chart.

• Helm provides a public repository of charts for popular software. You can also retrieve charts from third-party repositories, author and contribute your own charts to someone else’s repository, or run your own chart repository.

• The Helm client software offers commands for: listing and searching for charts by keyword, installing applications to your cluster from charts, upgrading those applications, removing applications, and other management functions.

Here are the biggest changes for Helm 3. For a complete list and more details, see the FAQ.

• The most notable change in Helm 3 was the removal of Tiller. With role-based access controls (RBAC) enabled by default in Kubernetes 1.6+, Tiller became unnecessary and was removed.

• Upgrading a chart is better than ever. Helm 3 introduces a 3-way merge patch, an improvement over Helm 2’s 2-way approach. Helm is now able to consider the old manifest, the current state, and the new manifest, instead of just the most recent manifest and the proposed changes. The 3-way merge patch helps to ensure that a user can roll back changes regardless of how they’re applied.

• Release names in Helm 3 are scoped to the namespace and have a sh.helm.release.v1 prefix.

• Secrets are used as the default storage driver for releases.

• The Go import path has changed from k8s.io/helm to helm.sh/helm/v3.

• requirements.yaml has been folded into Chart.yaml as the dependencies field.

• Helm 3 now supports Library charts. These are shared by other charts and are intended to be reused to avoid redundancy.

• Helm 3 has moved to XDG Base Directory Specification. This means instead of Helm 2’s $HELM_HOME location, you will find information stored in the following:

  • XDG_CACHE_HOME
  • XDG_CONFIG_HOME
  • XDG_DATA_HOME

• Helm Hub - Helm 3 does not come with chart repositories loaded out of the box. Instead there is now a central hub for charts called Helm Hub.

Helm has provided a plugin to migrate your projects from Helm 2 to Helm 3 called helm-2to3. This plugin works in three stages. First it migrates the configuration, then the release, then it cleans up the configuration, release data, and Tiller.

The components of a Kubernetes application–deployments, services, ingresses, and other objects–are listed in manifest files (in the YAML file format). Kubernetes does not tell you how you should organize those files, though the Kubernetes documentation does offer a general set of best practices.

Helm charts are the software packaging format for Helm. A chart specifies a file and directory structure that you follow when packaging your manifests. The structure looks as follows:

chart-name/
  Chart.yaml
  LICENSE
  README.md
  values.yaml
  charts/
  crds/
  templates/
  templates/NOTES.txt

When you tell Helm to install a chart, you can specify variable values to be inserted into the chart’s manifest templates. Helm will then compile those templates into manifests that can be applied to your cluster. When it does this, it creates a new release.

You can install a chart to the same cluster more than once. Each time you tell Helm to install a chart, it creates another release for that chart. A release can be upgraded when a new version of a chart is available, or even when you just want to supply new variable values to the chart. Helm tracks each upgrade to your release, and it allows you to roll back an upgrade. A release can be easily deleted from your cluster, and you can even roll back release deletions when configured to do so in advanced.

The Helm client software issues commands to your cluster. You run the client software on your computer, in your CI/CD environment, or anywhere else you’d like.

1. Install the Kubernetes CLI (kubectl) on your computer, if it is not already.

2. You should have a Kubernetes cluster running prior to starting this guide. One quick way to get a cluster up is with Linode’s Kubernetes Engine. This guide’s examples only require a cluster with one worker node. We recommend that you create cluster nodes that are at the Linode 4GB tier (g6-standard-2) or higher. This guide also assumes that your cluster has role-based access control (RBAC) enabled. This feature became available in Kubernetes 1.6 and later.

Caution

The k8s-alpha CLI is deprecated. On March 31st, 2020, it will be removed from the linode-cli. After March 31, 2020, you will no longer be able to create or manage clusters using the k8s-alpha CLI plugin.

However, you will still be able to create and manage these clusters using Terraform. The Terraform module used is a public project officially supported by Linode, and is currently used to power the k8s-alpha CLI.

Other alternatives for creating and managing clusters include:

• The Linode Kubernetes Engine (LKE), which creates clusters managed by Linode.
• Rancher, which provides a graphical user interface for managing clusters.

Note

This guide’s example instructions will also result in the creation of a Block Storage Volume and a NodeBalancer, which are also billable resources. If you do not want to keep using the example application after you finish reviewing your guide, make sure to delete these resources afterward.

1. You should also make sure that your Kubernetes CLI is using the right cluster context. Run the get-contexts subcommand to check:

   kubectl config get-contexts

2. You can set kubectl to use a certain cluster context with the use-context subcommand and the cluster name that was previously output from the get-contexts subcommand:

   kubectl config use-context your-cluster-name

3. It is beneficial to have a registered domain name for this guide’s example app, but it is not required.

Install the Helm client software on your computer:

Linux - Run the client installer script that Helm provides:

curl https://raw.githubusercontent.com/helm/helm/master/scripts/get-helm-3 > get_helm.sh
chmod 700 get_helm.sh
./get_helm.sh

macOS - Use Homebrew to install:

brew install helm

Windows - Use Chocolatey to install:

choco install kubernetes-helm

This guide will use the Ghost publishing platform as the example application.

1. Search the Helm Hub for the Ghost chart:

   helm search hub ghost
   

   URL                                     	CHART VERSION	APP VERSION	DESCRIPTION
   https://hub.helm.sh/charts/bitnami/ghost	9.0.3        	3.1.1      	A simple, powerful publishing platform that all...

   This gives you the URL where the chart is located in the central hub. Here you will find all the information about configuration and setup.

2. Add the stable repository:

   helm repo add stable https://charts.helm.sh/stable
   

3. Update the repo to ensure you get the latest chart version:

   helm repo update
   

4. The full name for the chart is stable/ghost. You can inspect the chart for more information:

   helm show readme stable/ghost
   

   This command’s output will resemble the README text available for the Ghost in the official central hub as linked above.

The helm install command is used to install a chart by name. It can be run without any other options, but some charts expect you to pass in configuration values for the chart:

1. Create a file named ghost-values.yaml on your computer for this snippet:

   File: ghost-values.yaml

   1 2 3 4 5

   ghostHost: "ghost.example.com" ghostEmail: "email@example.com" ghostUsername: "admin" ghostPassword: "mySecurePassword123!!" mariadb.mariadbRootPassword: "secretpassword"

   Replace the value for ghostHost with a domain or subdomain that you own and would like to assign to the app; the value for ghostEmail with your email; the values for ghostUsername and ghostPassword with the credentials you wish to use for logging into your site; and the value for mariabd.mariadbRootPassword for the password you wish to use for logging into the database.

   Note

   If you don’t own a domain name and won’t continue to use the Ghost website after finishing this guide, you can make up a domain for this configuration file.

2. Run the install command and pass in the configuration file:

   helm install --values=ghost-values.yaml stable/ghost --generate-name
   

3. The install command returns immediately and does not wait until the app’s cluster objects are ready. You will see output like the following snippet, which shows that the app’s pods are still in the “Pending” state. The text displayed is generated from the contents of the chart’s templates/NOTES.txt file:

   NAME: ghost-1576075187
   LAST DEPLOYED: Wed Dec 11 09:39:50 2019
   NAMESPACE: default
   STATUS: deployed
   REVISION: 1
   NOTES:
   1. Get the Ghost URL by running:
   
     echo Blog URL  : http://ghost.example.com/
     echo Admin URL : http://ghost.example.com/ghost
   
   2. Get your Ghost login credentials by running:
   
     echo Email:    email@example.com
     echo Password: $(kubectl get secret --namespace default ghost-1576075187 -o jsonpath="{.data.ghost-password}" | base64 --decode)

4. Helm has created a new release and assigned it a random name. Run the ls command to get a list of all of your releases:

   helm ls
   

   The output will look as follows:

   NAME            	NAMESPACE	REVISION	UPDATED                             	STATUS  	CHART      	APP VERSION
   ghost-1576075187	default  	1       	2019-12-11 09:39:50.168546 -0500 EST	deployed	ghost-9.0.3	3.1.1

5. You can check on the status of the release by running the status command:

   helm status ghost-1576075187
   

   This command will show the same output that was displayed after the helm install command, but the current state of the cluster objects will be updated.

1. Navigate to the NodeBalancer that was created in Cloud Manager and find the IP address.

2. The LoadBalancer that was created for the app will be displayed. Because this example uses a cluster created with Linode’s k8s-alpha CLI (which pre-installs the Linode CCM), the LoadBalancer will be implemented as a Linode NodeBalancer.

3. Copy the value under the IP Address column for the NodeBalancer and then paste it into your web browser. You should see the Ghost website:

4. Revisit the output from the status command. Instructions for logging into your Ghost website will be displayed:

   [...]
   1. Get the Ghost URL by running:
   
     echo Blog URL  : http://ghost.example.com/
     echo Admin URL : http://ghost.example.com/ghost
   
   2. Get your Ghost login credentials by running:
   
     echo Email:    email@example.com
     echo Password: $(kubectl get secret --namespace default ghost-1576075187 -o jsonpath="{.data.ghost-password}" | base64 --decode)

5. If you haven’t set up DNS for your site yet, you can instead access the admin interface by visiting the ghost URL on your LoadBalancer IP address (e.g. http://104.237.148.66/ghost). Visit this page in your browser and then follow the steps to complete admin account creation. You should be granted access to the administrative interface.

6. To set up DNS for your app, create an A record for your domain which is assigned to the external IP for your app’s LoadBalancer. Review Linode’s DNS Manager guide for instructions.

The upgrade command can be used to upgrade an existing release to a new version of a chart, or just to supply new chart values:

1. In your computer’s ghost-values.yaml file, add a line for the title of the website:

   File: ghost-values.yaml

   1 2 3 4 5 6

   ghostHost: "ghost.example.com" ghostEmail: "email@example.com" ghostUsername: "admin" ghostPassword: "mySecurePassword123!!" mariadb.mariadbRootPassword: "secretpassword" ghostBlogTitle: "Example Site Name"

2. Run the upgrade command, specifying the configuration file, release name, and chart name:

   helm upgrade --values=ghost-values.yaml ghost-1576075187 stable/ghost
   

Upgrades (and even deletions) can be rolled back if something goes wrong:

1. Run the helm ls command and observe the number under the “REVISION” column for your release:

   NAME            	NAMESPACE	REVISION	UPDATED                             	STATUS  	CHART      	APP VERSION
   ghost-1576075187	default  	2       	2019-12-11 11:54:49.136865 -0500 EST	deployed	ghost-9.0.3	3.1.1

2. Every time you perform an upgrade, the revision count is incremented by 1 (and the counter starts at 1 when you first install a chart). So, your current revision number is 2. To roll back the upgrade you just performed, enter the previous revision number:

   helm rollback ghost-1576075187 1
   

Caution

By default, Helm 3 does not keep any information about deleted releases, which will prevent you from rolling back. If you suspect that you may need to rollback your release following deletion, you will need to use the --keep-history flag.

1. Use the uninstall command with the name of a release to delete it:

   helm uninstall ghost-1576075187
   

   You should also confirm in the Linode Cloud Manager that the Volumes and NodeBalancer created for the app are removed as well.

   Note

   In Helm 2, deletions were performed using the delete command. This can still be entered to perform the same task, however in helm 3 delete aliases to uninstall.

2. If you wish to keep a history of past releases, you will want to use the --keep-history flag. This is a change from Helm 2.

   helm uninstall --keep-history
   

3. Helm will still save information about the uninstalled release. You can list releases including records where --keep-history was specified on uninstall:

   helm list --uninstalled
   

   Note

   You can no longer rollback a deleted or uninstalled release.

More Information

You may wish to consult the following resources for additional information on this topic. While these are provided in the hope that they will be useful, please note that we cannot vouch for the accuracy or timeliness of externally hosted materials.

• Helm Documentation
• Helm Security on Bitnami Engineering