Installing Gremlin on Kubernetes with Helm

The Gremlin Kubernetes agent lets you run experiments and reliability tests on standard Kubernetes resources, as well as Argo Rollouts. The Gremlin Helm Chart is the recommended way to install the Gremlin agent on Kubernetes. If you want to install Gremlin using plain YAML files and kubectl, see Install Gremlin on Kubernetes manually.

The Gremlin Helm Chart requires Helm v3 or later.

‍

Using the auto-generated Helm command (recommended)

Gremlin automatically generates a command that you can use to easily deploy and authenticate Gremlin via Helm. To use the auto-generated installation method:

1. Log into the Gremlin web app and navigate to the Getting Started page.
2. Under step 1 ("Install the Gremlin Agent"), click on the link to the values.yaml  file. This will prompt you to download the file to your PC.
3. Open a terminal and run the following command. Make sure to replace [your-cluster-ID] with the name of your cluster:

helm install gremlin gremlin/gremlin \
  --namespace gremlin \
  --create-namespace \
  --values values.yaml \
  --set gremlin.secret.clusterID=[your-cluster-ID]

‍

Manually deploying via Helm

The steps for deploying to Kubernetes with Helm are:

1. Gather your credentials
2. Deploy the Helm chart
3. Additional configuration
4. Verify your installation

Gather your credentials

All Gremlin integration installations require you to use one of Gremlin's authentication methods. With Helm, you can use either signature (i.e. certificate)-based authentication or secret authentication. Secret-based authentication is easier to implement, but we recommend using certificate-based authentication. We'll show both methods in this guide.

First, retrieve your Team ID from your team settings page. Next, if you're using secret-based authentication, create a new secret, copy the secret to a text file (or keep the pop-up open), then continue to the next step.

If you're using certificate-based authentication, generate a new certificate pair, or download an existing certificate pair if one exists. Unzip the folder containing your certificates, then continue to the next step.

Deploy the Helm chart

Now that we have our credentials, let's add the Gremlin Helm repo and create a new namespace for our deployment:

helm repo add gremlin https://helm.gremlin.com/
kubectl create namespace gremlin

‍

Now we'll deploy the Chart. With this command, your team ID and secrets/certificates are passed directly to the Helm Chart. If you'd prefer to manage these secrets outside of Helm, see our Helm Chart documentation.

‍

For secret-based authentication:

helm install gremlin gremlin/gremlin \
    --namespace gremlin \
    --set      gremlin.secret.type=secret \
    --set      gremlin.secret.managed=true \
    --set      gremlin.hostPID=true \
    --set      gremlin.hostNetwork=true \
    --set      gremlin.secret.teamID=YourGremlinTeamID \
    --set      gremlin.secret.clusterID=YourGremlinClusterID \
    --set      gremlin.secret.teamSecret=YourGremlinTeamSecret

‍

For certificate-based authentication:

helm install gremlin gremlin/gremlin \
    --namespace gremlin \
    --set      gremlin.secret.type=certificate \
    --set      gremlin.secret.managed=true \
    --set      gremlin.hostPID=true \
    --set      gremlin.hostNetwork=true \
    --set      gremlin.secret.teamID=YourGremlinTeamID \
    --set      gremlin.secret.clusterID=YourGremlinClusterID \
    --set-file gremlin.secret.certificate=/path/to/gremlin.cert \
    --set-file gremlin.secret.key=/path/to/gremlin.key

‍

Additional configuration

The Helm Chart includes several additional options for customizing to your environment. These options include:

• Specifying a container driver (Gremlin will try to auto-detect the current driver, but you can also tell Gremlin to use a specific driver).
• Enabling Gremlin on the Kubernetes Master so you can run tests on the Kubernetes Control Plane.
• Enabling AppArmor support for clusters that use AppArmor.
• Installing Gremlin's custom PodSecurityPolicy or seccomp policy for restrictive clusters.
• Configuring Gremlin to use a proxy.
• Allowing multiple Gremlin teams to target specific namespaces.
• Allow adding arbitrary environment variables.
• Allow adding arbitrary annotations to Kubernetes service account.

These are optional and are only necessary if Gremlin doesn't run using the default options. To learn more, see the Additional Configuration for Helm docs.

Verify your installation

On the cluster, you can run the following command to check that the Gremlin Agent was installed properly:

kubectl get pods -n gremlin

‍

This should list one Gremlin Agent for each node in your cluster, plus one Pod named <span class="code-class-custom">chao</span>. For example, this is the output for a three-node cluster:

kubectl get pods -n gremlin

NAME                    READY   STATUS    RESTARTS   AGE
chao-78bbc7cbf6-9hn7q   1/1     Running   0          5d20h
gremlin-9r4t7           1/1     Running   0          5d20h
gremlin-bwmtz           1/1     Running   1          126d
gremlin-bx6dn           1/1     Running   0          5d20h

‍

The following example shows 2 pending pods, which means the installation is incomplete. Contact your cluster administrator to debug why Gremlin is unable to run on those nodes.

kubectl get pods -n gremlin

NAME                    READY   STATUS    RESTARTS   AGE
chao-78bbc7cbf6-9hn7q   1/1     Running   0          5d20h
gremlin-c25ld           0/1     Pending   0          112d
gremlin-n5gt7           0/1     Pending   0          112d
gremlin-zn4kq           1/1     Running   0          126d

‍

‍

If you need additional troubleshooting help, see Gremlin Agent in the Gremlin Knowledge Base.

Uninstalling Gremlin from Kubernetes with Helm

To uninstall Gremlin via Helm, run the following commands:

helm uninstall gremlin -n gremlin
kubectl delete namespace gremlin

‍

‍