Install Kf on Google Cloud

This document describes how to set up a GKE cluster, and then install Kf and its dependencies.

Before you begin

Note: Autopilot GKE clusters are not currently able to run Kf. Use a Standard GKE cluster. For more information about these modes, and to learn more about Autopilot, see the Autopilot overview.

GKE cluster requirements

• Optional, but recommended, the cluster should be dedicated to Kf. We recommend that you only install Kf and its dependencies to ensure that the compatibility matrix is maintained.

• At least four nodes. If you need to add nodes, see Resizing a cluster.

• The minimum machine type that has at least four vCPUs, such as e2-standard-4. If the machine type for your cluster doesn't have at least four vCPUs, change the machine type as described in Migrating workloads to different machine types.

• Optional, but recommended, enroll the cluster in a release channel. Follow the instructions in Enrolling an existing cluster in a release channel if you have a static GKE version.

• Workload Identity enabled.

Kf requirements

Review and understand the access permissions of components in Kf in the Kf dependencies and architecture page.

The Dependency matrix lists the specific versions.

• Artifact Registry

• Cloud Service Mesh

• Config Connector

• Tekton for use by Kf; this is not a user facing service

• A dedicated Google Service Account

Note: The Clean up section has instructions on deleting the cluster.

Enable support for Compute Engine

1. Sign in to your Google Cloud account. If you're new to Google Cloud, create an account to evaluate how our products perform in real-world scenarios. New customers also get $300 in free credits to run, test, and deploy workloads.

2. In the Google Cloud console, on the project selector page, select or create a Google Cloud project.

   Note: If you don't plan to keep the resources that you create in this procedure, create a project instead of selecting an existing project. After you finish these steps, you can delete the project, removing all resources associated with the project.

   Go to project selector

3. Make sure that billing is enabled for your Google Cloud project.

4. In the Google Cloud console, on the project selector page, select or create a Google Cloud project.

   Note: If you don't plan to keep the resources that you create in this procedure, create a project instead of selecting an existing project. After you finish these steps, you can delete the project, removing all resources associated with the project.

   Go to project selector

5. Make sure that billing is enabled for your Google Cloud project.

6. Enable the Compute Engine API.

   Enable the API

Enable support for Artifact Registry

1. Enable the Artifact Registry API.

   Enable the Artifact Registry API

Enable and configure GKE

Before you start, make sure you have performed the following tasks:

• Enable the Google Kubernetes Engine API.
Enable Google Kubernetes Engine API
• If you want to use the Google Cloud CLI for this task, install and then initialize the gcloud CLI. If you previously installed the gcloud CLI, get the latest version by running gcloud components update.
  Note: For existing gcloud CLI installations, make sure to set the compute/region and compute/zone properties. By setting default locations, you can avoid errors in gcloud CLI like the following: One of [--zone, --region] must be supplied: Please specify location.

Prepare a new GKE cluster and related services

Set up environment variables

Note: Export these environment variables in your terminal or Cloud Shell to make subsequent commands work. If you disconnect from your shell environment and reconnect later, you need to set up the variables again.

Linux and Mac

export PROJECT_ID=YOUR_PROJECT_ID
export CLUSTER_PROJECT_ID=YOUR_PROJECT_ID
export CLUSTER_NAME=kf-cluster
export COMPUTE_ZONE=us-central1-a
export COMPUTE_REGION=us-central1
export CLUSTER_LOCATION=${COMPUTE_ZONE} # Replace ZONE with REGION to switch
export NODE_COUNT=4
export MACHINE_TYPE=e2-standard-4
export NETWORK=default

Windows PowerShell

Set-Variable -Name PROJECT_ID -Value YOUR_PROJECT_ID
Set-Variable -Name CLUSTER_PROJECT_ID -Value YOUR_PROJECT_ID
Set-Variable -Name CLUSTER_NAME -Value kf-cluster
Set-Variable -Name COMPUTE_ZONE -Value us-central1-a
Set-Variable -Name COMPUTE_REGION -Value us-central1
Set-Variable -Name CLUSTER_LOCATION -Value $COMPUTE_ZONE # Replace ZONE with REGION to switch
Set-Variable -Name NODE_COUNT -Value 4
Set-Variable -Name MACHINE_TYPE -Value e2-standard-4
Set-Variable -Name NETWORK -Value default

Set up service account

Create a Google Cloud service account that will be associated with a Kubernetes Service Account via Workload Identity. This prevents the need to create and inject a service account key.

Note: Read through and understand the security implications of using Workload Identity in the Security overview document.

1. Create the service account that Kf will use.

   gcloud iam service-accounts create ${CLUSTER_NAME}-sa \
   --project=${CLUSTER_PROJECT_ID} \
   --description="GSA for Kf ${CLUSTER_NAME}" \
   --display-name="${CLUSTER_NAME}"

2. Create a new custom IAM role.

   gcloud iam roles create serviceAccountUpdater \
   --project=${CLUSTER_PROJECT_ID} \
   --title "Service Account Updater" \
   --description "This role only updates members on a GSA" \
   --permissions iam.serviceAccounts.get,iam.serviceAccounts.getIamPolicy,iam.serviceAccounts.list,iam.serviceAccounts.setIamPolicy

3. Allow the service account to modify its own policy. The Kf controller will use this to add new (name)spaces to the policy, allowing reuse for Workload Identity.

   gcloud projects add-iam-policy-binding ${CLUSTER_PROJECT_ID} \
     --member="serviceAccount:${CLUSTER_NAME}-sa@${CLUSTER_PROJECT_ID}.iam.gserviceaccount.com" \
     --role="projects/${CLUSTER_PROJECT_ID}/roles/serviceAccountUpdater"

4. Give monitoring metrics role for write access to Cloud Monitoring.

   gcloud projects add-iam-policy-binding ${CLUSTER_PROJECT_ID} \
     --member="serviceAccount:${CLUSTER_NAME}-sa@${CLUSTER_PROJECT_ID}.iam.gserviceaccount.com" \
     --role="roles/monitoring.metricWriter"

5. Give logging role for write access to Cloud Logging.

   gcloud projects add-iam-policy-binding ${CLUSTER_PROJECT_ID} \
     --member="serviceAccount:${CLUSTER_NAME}-sa@${CLUSTER_PROJECT_ID}.iam.gserviceaccount.com" \
     --role="roles/logging.logWriter"

Create GKE cluster

Note: For demonstration purposes, these instructions use a zonal cluster. For production deployments, a regional cluster is recommended.

gcloud container clusters create ${CLUSTER_NAME} \
  --project=${CLUSTER_PROJECT_ID} \
  --zone=${CLUSTER_LOCATION} \
  --num-nodes=${NODE_COUNT} \
  --machine-type=${MACHINE_TYPE} \
  --disk-size "122" \
  --network=${NETWORK} \
  --addons HorizontalPodAutoscaling,HttpLoadBalancing,GcePersistentDiskCsiDriver \
  --enable-dataplane-v2 \
  --enable-stackdriver-kubernetes \
  --enable-ip-alias \
  --enable-autorepair \
  --enable-autoupgrade \
  --scopes cloud-platform \
  --release-channel=regular \
  --workload-pool="${CLUSTER_PROJECT_ID}.svc.id.goog" \
  --service-account="${CLUSTER_NAME}-sa@${CLUSTER_PROJECT_ID}.iam.gserviceaccount.com"

Set firewall rules

Kf requires some firewall ports to be open. The master node needs to be able to communicate with pods on ports 80, 443, 8080, 8443 and 6443.

Enable Workload Identity

Now that you have a service account and GKE cluster, associate the cluster's identity namespace with the cluster.

gcloud iam service-accounts add-iam-policy-binding \
  --project=${CLUSTER_PROJECT_ID} \
  --role roles/iam.workloadIdentityUser \
  --member "serviceAccount:${CLUSTER_PROJECT_ID}.svc.id.goog[kf/controller]" \
  "${CLUSTER_NAME}-sa@${CLUSTER_PROJECT_ID}.iam.gserviceaccount.com"

gcloud iam service-accounts add-iam-policy-binding \
  --project=${CLUSTER_PROJECT_ID} \
  --role roles/iam.workloadIdentityUser \
  --member "serviceAccount:${CLUSTER_PROJECT_ID}.svc.id.goog[cnrm-system/cnrm-controller-manager]" \
  "${CLUSTER_NAME}-sa@${CLUSTER_PROJECT_ID}.iam.gserviceaccount.com"

Target GKE cluster

Configure kubectl command line access by running the following command.

gcloud container clusters get-credentials ${CLUSTER_NAME} \
    --project=${CLUSTER_PROJECT_ID} \
    --zone=${CLUSTER_LOCATION}

Create an Artifact Registry repository

1. Create an Artifact Registry for container images to be stored.

   gcloud artifacts repositories create ${CLUSTER_NAME} \
     --project=${CLUSTER_PROJECT_ID} \
     --repository-format=docker \
     --location=${COMPUTE_REGION}

2. Grant the service account permission on the Artifact Registry repository.

   gcloud artifacts repositories add-iam-policy-binding ${CLUSTER_NAME} \
     --project=${CLUSTER_PROJECT_ID} \
     --location=${COMPUTE_REGION} \
     --member="serviceAccount:${CLUSTER_NAME}-sa@${CLUSTER_PROJECT_ID}.iam.gserviceaccount.com" \
     --role='roles/artifactregistry.writer'

Install software dependencies on a cluster

Note: The dependency matrix on the Kf Downloads page has pinned versions for the given Kf version.

1. Install Cloud Service Mesh v1.12.

   Note: when installing gateways following the ASM document, please select 'Revision label' on step 2 as that is the way Kf supports.
   1. Follow the Cloud Service Mesh install guide, including steps to create an ingress gateway.

2. Install Config Connector.

   Note: Config Connector is required starting with Kf v2.4.0.
   Warning: Do not use the GKE add-on method to install Config Connector as it is not supported by Kf. Please use these custom installation steps.

   1. Download the required Config Connector Operator tar file.

      Note: Keep a copy of the installation bundle as you need this bundle to uninstall Config Connector.

   2. Extract the tar file.

      tar zxvf release-bundle.tar.gz
      

   3. Install the Config Connector operator on your cluster.

      kubectl apply -f operator-system/configconnector-operator.yaml
      

   4. Configure the Config Connector operator.

      1. Copy the following YAML into a file named configconnector.yaml:

         Warning: Replace the KF_SERVICE_ACCOUNT_NAME variable in the YAML below with the full service account resolved from ${CLUSTER_NAME}-sa@${CLUSTER_PROJECT_ID}.iam.gserviceaccount.com

         # configconnector.yaml
         apiVersion: core.cnrm.cloud.google.com/v1beta1
         kind: ConfigConnector
         metadata:
           # the name is restricted to ensure that there is only one
           # ConfigConnector resource installed in your cluster
           name: configconnector.core.cnrm.cloud.google.com
         spec:
           mode: cluster
           googleServiceAccount: "KF_SERVICE_ACCOUNT_NAME" # Replace with the full service account resolved from ${CLUSTER_NAME}-sa@${CLUSTER_PROJECT_ID}.iam.gserviceaccount.com
         

      2. Apply the configuration to your cluster.

         kubectl apply -f configconnector.yaml

   5. Verify Config Connector is fully installed before continuing.

      • Config Connector runs all of its components in a namespace named cnrm-system. Verify the Pods are ready by running the following command:

        Note: The controller Pod can take several minutes to start, so this command might not immediately work.

        kubectl wait -n cnrm-system --for=condition=Ready pod --all

      • If Config Connector is installed correctly, expect to see output similar to the following:

        pod/cnrm-controller-manager-0 condition met
        pod/cnrm-deletiondefender-0 condition met
        pod/cnrm-resource-stats-recorder-86858dcdc5-6lqzb condition met
        pod/cnrm-webhook-manager-58c799b8fb-kcznq condition met
        pod/cnrm-webhook-manager-58c799b8fb-n2zpx condition met

   6. Setup Workload Identity.

      kubectl annotate serviceaccount \
      --namespace cnrm-system \
      --overwrite \
      cnrm-controller-manager \
      iam.gke.io/gcp-service-account=${CLUSTER_NAME}-sa@${CLUSTER_PROJECT_ID}.iam.gserviceaccount.com

3. Install Tekton:

   kubectl apply -f "https://storage.googleapis.com/tekton-releases/pipeline/previous/v0.32.1/release.yaml"

Install Kf

Note: Ensure your cluster meets the requirements found in Create and prepare a GKE cluster for Kf.

1. Install the Kf CLI:

   Linux

   This command installs the Kf CLI for all users on the system. Follow the instructions in the Cloud Shell tab to install it just for yourself.

   gcloud storage cp gs://kf-releases/v2.9.0/kf-linux /tmp/kf
   chmod a+x /tmp/kf
   sudo mv /tmp/kf /usr/local/bin/kf
   

   Mac

   This command installs kf for all users on the system.

   gcloud storage cp gs://kf-releases/v2.9.0/kf-darwin /tmp/kf
   chmod a+x /tmp/kf
   sudo mv /tmp/kf /usr/local/bin/kf
   

   Cloud Shell

   This command installs kf on your Cloud Shell instance if you use bash, the instructions may need to be modified for other shells.

   mkdir -p ~/bin
   gcloud storage cp gs://kf-releases/v2.9.0/kf-linux ~/bin/kf
   chmod a+x ~/bin/kf
   echo "export PATH=$HOME/bin:$PATH" >> ~/.bashrc
   source ~/.bashrc
   

   Windows

   This command downloads kf to current directory. Add it to the path if you want to call if from anywhere other than the current directory.

   gcloud storage cp gs://kf-releases/v2.9.0/kf-windows.exe kf.exe
   

2. Install the operator:

   kubectl apply -f "https://storage.googleapis.com/kf-releases/v2.9.0/operator.yaml"

3. Configure the operator for Kf:

   kubectl apply -f "https://storage.googleapis.com/kf-releases/v2.9.0/kfsystem.yaml"

4. Setup secrets and defaults:

   export CONTAINER_REGISTRY=${COMPUTE_REGION}-docker.pkg.dev/${CLUSTER_PROJECT_ID}/${CLUSTER_NAME}
   
   kubectl patch \
   kfsystem kfsystem \
   --type='json' \
   -p="[{'op': 'replace', 'path': '/spec/kf', 'value': {'enabled': true, 'config': {'spaceContainerRegistry': '${CONTAINER_REGISTRY}', 'secrets':{'workloadidentity':{'googleserviceaccount':'${CLUSTER_NAME}-sa', 'googleprojectid':'${CLUSTER_PROJECT_ID}'}}}}}]"
   

Validate installation

  kf doctor --retries=20

Note: This command will inspect the cluster to ensure Kf and its dependencies are installed and working. It will run up to 20 times to give the Kubernetes cluster time to reconcile, and should result in a series PASS results. If the status for any component results in FAIL after the retries, please see the Kf Troubleshooting page.

Clean up

These steps should remove all the components created in the Create and prepare a new GKE cluster section.

1. Delete Google Service Account:

   gcloud iam service-accounts delete ${CLUSTER_NAME}-sa@${CLUSTER_PROJECT_ID}.iam.gserviceaccount.com

2. Delete IAM policy bindings:

   gcloud projects remove-iam-policy-binding ${CLUSTER_PROJECT_ID} \
       --member="serviceAccount:${CLUSTER_NAME}-sa@${CLUSTER_PROJECT_ID}.iam.gserviceaccount.com" \
       --role="roles/iam.serviceAccountAdmin"
   gcloud projects remove-iam-policy-binding ${CLUSTER_PROJECT_ID} \
       --member="serviceAccount:${CLUSTER_NAME}-sa@${CLUSTER_PROJECT_ID}.iam.gserviceaccount.com" \
       --role="roles/monitoring.metricWriter"
   

3. Delete container image repository:

   gcloud artifacts repositories delete ${CLUSTER_NAME} \
     --location=${COMPUTE_REGION}

4. Uninstall Kf:

   kubectl patch kfsystem kfsystem \
     --type='json' \
     -p="[{'op': 'replace', 'path': '/spec/kf', 'value': {'enabled': false, }}]"
   

5. (Optional) Delete GKE cluster:

   gcloud container clusters delete ${CLUSTER_NAME} --zone ${CLUSTER_LOCATION}