Installing Apps on Kubernetes with Helm 2

What is Helm?

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.

Charts

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
  requirements.yaml
  values.yaml
  charts/
  templates/
  templates/NOTES.txt

Releases

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.

Helm Client and Helm Tiller

Helm operates with two components:

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

• A server component runs on your cluster and receives commands from the Helm client software. This component is called Tiller. Tiller is responsible for directly interacting with the Kubernetes API (which the client software does not do). Tiller maintains the state for your Helm releases.

Before You Begin

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 k8s-alpha CLI command. 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 or higher.

   This guide also assumes that your cluster has role-based access control (RBAC) enabled. This feature became available in Kubernetes 1.6. It is enabled on clusters created via the k8s-alpha Linode CLI.

   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.

3. 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

4. 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

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

Install Helm

Install the Helm Client

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 > get_helm.sh
  chmod 700 get_helm.sh
  ./get_helm.sh

• macOS. Use Homebrew to install:

  brew install kubernetes-helm

• Windows. Use Chocolatey to install:

  choco install kubernetes-helm

Install Tiller on your Cluster

Tiller’s default installation instructions will attempt to install it without adequate permissions on a cluster with RBAC enabled, and it will fail. Alternative instructions are available which grant Tiller the appropriate permissions:

1. Create a file on your computer named rbac-config.yaml with the following snippet:

   File: rbac-config.yaml

   1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

   apiVersion: v1 kind: ServiceAccount metadata: name: tiller namespace: kube-system --- apiVersion: rbac.authorization.k8s.io/v1 kind: ClusterRoleBinding metadata: name: tiller roleRef: apiGroup: rbac.authorization.k8s.io kind: ClusterRole name: cluster-admin subjects: - kind: ServiceAccount name: tiller namespace: kube-system

   This configuration creates a Kubernetes Service Account for Tiller, and then binds it to the cluster-admin role.

2. Apply this configuration to your cluster:

   kubectl create -f rbac-config.yaml

   serviceaccount "tiller" created
   clusterrolebinding "tiller" created

3. Initialize Tiller on the cluster:

   helm init --service-account tiller --history-max 200
   Note

   The --history-max option prevents Helm’s historical record of the objects it tracks from growing too large.

4. You should see output like:

   $HELM_HOME has been configured at /Users/your-user/.helm.
   
   Tiller (the Helm server-side component) has been installed into your Kubernetes Cluster.
   
   Please note: by default, Tiller is deployed with an insecure 'allow unauthenticated users' policy.
   To prevent this, run `helm init` with the --tiller-tls-verify flag.
   For more information on securing your installation see: https://docs.helm.sh/using_helm/#securing-your-helm-installation
   Happy Helming!

5. The Pod for Tiller will be running in the kube-system namespace:

   kubectl get pods --namespace kube-system | grep tiller

   tiller-deploy-b6647fc9d-vcdms                1/1       Running   0          1m

Use Helm Charts to Install Apps

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

Search for a Chart

1. Run the repo update subcommand to make sure you have a full list of available charts:

   helm repo update
   Note

   Run helm repo list to see which repositories are registered with your client.

2. Run the search command with a keyword to search for a chart by name:

   helm search ghost

   The output will look like:

   NAME        	CHART VERSION	APP VERSION	DESCRIPTION
   stable/ghost	6.7.7        	2.19.4     	A simple, powerful publishing platform that allows you to...

3. The full name for the chart is stable/ghost. Inspect the chart for more information:

   helm inspect stable/ghost

   This command’s output will resemble the README text available for the Ghost chart in the official Helm chart repository on GitHub.

Install the Chart

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-config.yaml on your computer from this snippet:

   File: ghost-config.yaml

   1 2	ghostHost: ghost.example.com ghostEmail: email@example.com

   Replace the value for ghostHost with a domain or subdomain that you own and would like to assign to the app, and the value for ghostEmail with your email.

   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 -f ghost-config.yaml stable/ghost

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:

   Full output of helm install

   NAME:   oldfashioned-cricket
   LAST DEPLOYED: Tue Apr 16 09:15:41 2019
   NAMESPACE: default
   STATUS: DEPLOYED
   
   RESOURCES:
   ==> v1/ConfigMap
   NAME                      DATA  AGE
   oldfashioned-cricket-mariadb        1     1s
   oldfashioned-cricket-mariadb-tests  1     1s
   
   ==> v1/PersistentVolumeClaim
   NAME              STATUS   VOLUME                CAPACITY  ACCESS MODES  STORAGECLASS  AGE
   oldfashioned-cricket-ghost  Pending  linode-block-storage  1s
   
   ==> v1/Pod(related)
   NAME                               READY  STATUS   RESTARTS  AGE
   oldfashioned-cricket-ghost-64ff89b9d6-9ngjs  0/1    Pending  0         1s
   oldfashioned-cricket-mariadb-0               0/1    Pending  0         1s
   
   ==> v1/Secret
   NAME                TYPE    DATA  AGE
   oldfashioned-cricket-ghost    Opaque  1     1s
   oldfashioned-cricket-mariadb  Opaque  2     1s
   
   ==> v1/Service
   NAME                TYPE          CLUSTER-IP      EXTERNAL-IP  PORT(S)       AGE
   oldfashioned-cricket-ghost    LoadBalancer  10.110.3.191    <pending>    80:32658/TCP  1s
   oldfashioned-cricket-mariadb  ClusterIP     10.107.128.144  <none>       3306/TCP      1s
   
   ==> v1beta1/Deployment
   NAME              READY  UP-TO-DATE  AVAILABLE  AGE
   oldfashioned-cricket-ghost  0/1    1           0          1s
   
   ==> v1beta1/StatefulSet
   NAME                READY  AGE
   oldfashioned-cricket-mariadb  0/1    1s
   
   
   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 oldfashioned-cricket-ghost -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      	REVISION	UPDATED                 	STATUS  	CHART      	APP VERSION	NAMESPACE
   oldfashioned-cricket	1       	Tue Apr 16 09:15:41 2019	DEPLOYED	ghost-6.7.7	2.19.4     	default

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

   helm status oldfashioned-cricket

   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.

Access your App

1. Run the helm status command again and observe the “Service” section:

   ==> v1/Service
   NAME                TYPE          CLUSTER-IP      EXTERNAL-IP     PORT(S)       AGE
   oldfashioned-cricket-ghost    LoadBalancer  10.110.3.191    104.237.148.15  80:32658/TCP  11m
   oldfashioned-cricket-mariadb  ClusterIP     10.107.128.144  <none>          3306/TCP      11m

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 EXTERNAL-IP column for the LoadBalancer 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 oldfashioned-cricket-ghost -o jsonpath="{.data.ghost-password}" | base64 --decode)

5. Retrieve the auto-generated password for your app:

   echo Password: $(kubectl get secret --namespace default oldfashioned-cricket-ghost -o jsonpath="{.data.ghost-password}" | base64 --decode)

6. You haven’t set up DNS for your site yet, but you can instead access the admin interface by visiting the ghost URL on your LoadBalancer IP address (e.g. http://104.237.148.15/ghost). Visit this page in your browser and then enter your email and password. You should be granted access to the administrative interface.

7. Set up DNS for your app. You can do this by creating 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.

Upgrade your App

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-config.yaml file, add a line for the title of the website:

   File: ghost-config.yaml

   1 2 3

   ghostHost: ghost.example.com ghostEmail: email@example.com ghostBlogTitle: Example Site Name

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

   helm upgrade -f ghost-config.yaml oldfashioned-cricket stable/ghost

Roll Back a Release

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      	REVISION	UPDATED                 	STATUS  	CHART      	APP VERSION	NAMESPACE
   oldfashioned-cricket	2       	Tue Apr 16 10:02:58 2019	DEPLOYED	ghost-6.7.7	2.19.4     	default

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 oldfashioned-cricket 1

Delete a Release

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

   helm delete oldfashioned-cricket

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

2. Helm will still save information about the deleted release. You can list deleted releases:

   helm list --deleted

   You can use the revision number of a deleted release to roll back the deletion.

3. To fully remove a release, use the --purge option with the delete command:

   helm delete oldfashioned-cricket --purge

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 2 Documentation
• Helm Security on Bitnami Engineering