kubernetes-retired / k8s-gsm-tools

Controllers to sync and rotate kubernetes secrets with google secret manager

k8s-gsm-tools

Secret rotation and synchronization integrating Google Secret Manager and Kubernetes.

Current Functions

Parse configuration file into source and destination secrets.

Fetch the latest versions of from Secret Manager secret and Kubernetes secrets.

Prerequisites

• Create a gke cluster.

• Create service account for app

• Enable Identity and Access Management (IAM) API for project.

    gcloud services enable iam.googleapis.com --project=<gcloud-project-id>
  

• Grant required permissions to the service account gsa-name.

  • Permission to manage service account keys:

      gcloud projects add-iam-policy-binding <gcloud-project-id> --member "serviceAccount:<gsa-name>@<gcloud-project-id>.iam.gserviceaccount.com" --role "roles/iam.serviceAccountKeyAdmin"
    

  • Permission to get clusters:

      gcloud projects add-iam-policy-binding <gcloud-project-id> --member "serviceAccount:<gsa-name>@<gcloud-project-id>.iam.gserviceaccount.com" --role "roles/container.clusterViewer"
    

  • Permission to manage secrets:

      gcloud projects add-iam-policy-binding <gcloud-project-id> --member "serviceAccount:<gsa-name>@<gcloud-project-id>.iam.gserviceaccount.com" --role "roles/secretmanager.admin"
    

  • Permission to manage secrets within containers:

    • Create a custom iam role iam-role-id with container.secrets.* permissions and add the role to service account gsa-name:

      • service-secret-role.yaml

          title: Kubernetes Engine Secret Admin
          description: Provides access to management of Kubernetes Secrets
          stage: GA
          includedPermissions:
          - container.secrets.create
          - container.secrets.list
          - container.secrets.get
          - container.secrets.delete
          - container.secrets.update
        

      • Create a custom iam role

          gcloud iam roles create <iam-role-id> --project=<gcloud-project-id> --file=service-secret-role.yaml
        

      • Add the role to service account gsa-name:

          gcloud projects add-iam-policy-binding <gcloud-project-id> --member "serviceAccount:<gsa-name>@<gcloud-project-id>.iam.gserviceaccount.com" --role "roles/<iam-role-id>"
        

    • Or just add [Kubernetes Engine Developer] role to service account gsa-name:

        gcloud projects add-iam-policy-binding <gcloud-project-id> --member "serviceAccount:<gsa-name>@<gcloud-project-id>.iam.gserviceaccount.com" --role "roles/container.developer"
      

• Modify the cluster to enable Workload Identity

gcloud container clusters update <cluster-name> \
  --workload-pool=<gcloud-project-id>.svc.id.goog

• Modify an existing node pool to enable GKE_METADATA

gcloud container node-pools update <nodepool-name> \
  --cluster=<cluster-name> \
  --workload-metadata=GKE_METADATA

• Create Kubernetes service account

kubectl apply -f service-account/serviceaccount.yaml

• Set up Workload Identity binding

gcloud iam service-accounts add-iam-policy-binding \
  --role roles/iam.workloadIdentityUser \
  --member "serviceAccount:<gcloud-project-id>.svc.id.goog[<k8s_namespace>/<ksa_name>]" \
  <gsa-name>@<gcloud-project-id>.iam.gserviceaccount.com

• Annotate the KSA to complete the binding between the KSA and GSA

kubectl annotate serviceaccount \
  --namespace <k8s_namespace> \
   <ksa_name> \
   iam.gke.io/gcp-service-account=<gsa-name>@<gcloud-project-id>.iam.gserviceaccount.com

• Set up Kubernetes service account role and binding (action might require container.roles.create and container.roles.bind permissions if using gke cluster)

kubectl apply -f service-account/role.yaml

Usage

• secret-sync-controller

  • create ConfigMap config with key syncConfig.

  • deploy controller in continuous mode

      kubectl apply -f cmd/secret-sync-controller/deployment.yaml
    

  • run testing job

      kubectl apply -f cmd/secret-sync-controller/test-job.yaml
    

• secret-rotator

  • create ConfigMap config with key rotConfig.

  • deploy rotator in continuous mode

      kubectl apply -f cmd/secret-rotator/deployment.yaml
    

• test-svc-consumer

  • build image locally and push

      docker build --pull \
      --build-arg "cmd=consumer" \
      -t "gcr.io/<gcloud-project-id>/consumer:latest" \
      -f "./images/default/Dockerfile" .
      docker push gcr.io/<gcloud-project-id>/consumer
    

  • run consumer as a job

      kubectl apply -f experiment/cmd/consumer/job.yaml
    

Demo for rotating service account keys

• create Secret Manager secret and Kubernetes namespace

    gcloud secrets create secret-1
    kubectl create namespace ns-a
  

• deploy secret-sync-controller

    kubectl apply -f cmd/secret-sync-controller/deployment.yaml
  

• deploy secret-rotator

    kubectl apply -f cmd/secret-rotator/deployment.yaml
  

• deploy svc-consumer

    kubectl apply -f experiment/cmd/consumer/job.yaml
  

• get logs

    kubectl logs -n ns-a <svc-consumer-pod>
  

• cleanup

    kubectl delete -f cmd/secret-sync-controller/deployment.yaml
    kubectl delete -f cmd/secret-rotator/deployment.yaml
    kubectl delete -f experiment/cmd/consumer/job.yaml
    gcloud secrets delete secret-1
    kubectl create namespace ns-a
  

Building / pushing images

To build images locally:

make images

If you have access to a GCP project that has Google Cloud Build enabled:

gcloud builds submit --config=./images/cloudbuild.yaml .

This file can be used by a prow image-pushing job to push to the project's repository