This tutorial shows you how to store the sensitive data that's used by your Google Kubernetes Engine (GKE) clusters in Secret Manager. You learn how to more securely access the data from your Pods by using Workload Identity Federation for GKE and the Google Cloud client libraries.
Storing your sensitive data outside your cluster storage reduces the risk of unauthorized access to the data if an attack occurs. Using Workload Identity Federation for GKE to access the data lets you avoid the risks associated with managing long-lived service account keys, and lets you control access to your secrets using Identity and Access Management (IAM) instead of in-cluster RBAC rules. You can use any external secret store provider, such as Secret Manager or HashiCorp Vault.
This page is for Security specialists who want to move sensitive data out of in-cluster storage. To learn more about common roles and example tasks that we reference in Google Cloud content, see Common GKE Enterprise user roles and tasks.
This tutorial uses a GKE Autopilot cluster. To perform the steps using GKE Standard, you must enable Workload Identity Federation for GKE manually.
You can use Workload Identity Federation for GKE to access any Google Cloud APIs from GKE workloads without having to use less secure approaches like static service account key files. This tutorial uses Secret Manager as an example, but you can use the same steps to access other Google Cloud APIs. To learn more, see Workload Identity Federation for GKE.
Objectives
- Create a secret in Google Cloud Secret Manager.
- Create a GKE Autopilot cluster, Kubernetes namespaces, and Kubernetes service accounts.
- Create IAM allow policies to grant access to your Kubernetes service accounts on the secret.
- Use test applications to verify service account access.
- Run a sample app that accesses the secret using the Secret Manager API.
Costs
In this document, you use the following billable components of Google Cloud:
To generate a cost estimate based on your projected usage,
use the pricing calculator.
When you finish the tasks that are described in this document, you can avoid continued billing by deleting the resources that you created. For more information, see Clean up.
Before you begin
- 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.
- Install the Google Cloud CLI.
-
To initialize the gcloud CLI, run the following command:
gcloud init
-
Create or select a Google Cloud project.
-
Create a Google Cloud project:
gcloud projects create PROJECT_ID
Replace
PROJECT_ID
with a name for the Google Cloud project you are creating. -
Select the Google Cloud project that you created:
gcloud config set project PROJECT_ID
Replace
PROJECT_ID
with your Google Cloud project name.
-
-
Make sure that billing is enabled for your Google Cloud project.
-
Enable the Kubernetes Engine and Secret Manager APIs:
gcloud services enable container.googleapis.com
secretmanager.googleapis.com - Install the Google Cloud CLI.
-
To initialize the gcloud CLI, run the following command:
gcloud init
-
Create or select a Google Cloud project.
-
Create a Google Cloud project:
gcloud projects create PROJECT_ID
Replace
PROJECT_ID
with a name for the Google Cloud project you are creating. -
Select the Google Cloud project that you created:
gcloud config set project PROJECT_ID
Replace
PROJECT_ID
with your Google Cloud project name.
-
-
Make sure that billing is enabled for your Google Cloud project.
-
Enable the Kubernetes Engine and Secret Manager APIs:
gcloud services enable container.googleapis.com
secretmanager.googleapis.com -
Grant roles to your user account. Run the following command once for each of the following IAM roles:
roles/secretmanager.admin, roles/container.clusterAdmin
gcloud projects add-iam-policy-binding PROJECT_ID --member="user:USER_IDENTIFIER" --role=ROLE
- Replace
PROJECT_ID
with your project ID. -
Replace
USER_IDENTIFIER
with the identifier for your user account. For example,user:myemail@example.com
. - Replace
ROLE
with each individual role.
- Replace
Prepare the environment
Clone the GitHub repository that contains the sample files for this tutorial:
git clone https://github.com/GoogleCloudPlatform/kubernetes-engine-samples
cd ~/kubernetes-engine-samples/security/wi-secrets
Create a secret in Secret Manager
The following example shows the data you'll use to create a secret:
Create a secret to store the sample data:
gcloud secrets create bq-readonly-key \ --data-file=manifests/bq-readonly-key \ --ttl=3600s
This command does the following:
- Creates a new Secret Manager secret with the sample key
in the
us-central1
Google Cloud region. - Sets the secret to expire one hour after you run the command.
- Creates a new Secret Manager secret with the sample key
in the
Create the cluster and Kubernetes resources
Create a GKE cluster, Kubernetes namespaces, and Kubernetes service accounts. You create two namespaces, one for read-only access and one for read-write access to the secret. You also create a Kubernetes service account in each namespace to use with Workload Identity Federation for GKE.
Create a GKE Autopilot cluster:
gcloud container clusters create-auto secret-cluster \ --region=us-central1
The cluster might take about five minutes to deploy. Autopilot clusters always have Workload Identity Federation for GKE enabled. If you want to use a GKE Standard cluster instead, you must manually enable Workload Identity Federation for GKE before you continue.
Create a
readonly-ns
namespace and anadmin-ns
namespace:kubectl create namespace readonly-ns kubectl create namespace admin-ns
Create a
readonly-sa
Kubernetes service account and anadmin-sa
Kubernetes service account:kubectl create serviceaccount readonly-sa --namespace=readonly-ns kubectl create serviceaccount admin-sa --namespace=admin-ns
Create IAM allow policies
Grant the
readonly-sa
service account read-only access to the secret:gcloud secrets add-iam-policy-binding bq-readonly-key \ --member=principal://iam.googleapis.com/projects/PROJECT_NUMBER/locations/global/workloadIdentityPools/PROJECT_ID.svc.id.goog/subject/ns/readonly-ns/sa/readonly-sa \ --role='roles/secretmanager.secretAccessor' \ --condition=None
Replace the following:
PROJECT_NUMBER
: your numerical Google Cloud project number.PROJECT_ID
: your Google Cloud project ID.
Grant the
admin-sa
service account read-write access to the secret:gcloud secrets add-iam-policy-binding bq-readonly-key \ --member=principal://iam.googleapis.com/projects/PROJECT_NUMBER/locations/global/workloadIdentityPools/PROJECT_ID.svc.id.goog/subject/ns/admin-ns/sa/admin-sa \ --role='roles/secretmanager.secretAccessor' \ --condition=None gcloud secrets add-iam-policy-binding bq-readonly-key \ --member=principal://iam.googleapis.com/projects/PROJECT_NUMBER/locations/global/workloadIdentityPools/PROJECT_ID.svc.id.goog/subject/ns/admin-ns/sa/admin-sa \ --role='roles/secretmanager.secretVersionAdder' \ --condition=None
Verify secret access
Deploy test Pods in each namespace to verify the read-only and read-write access.
Review the read-only Pod manifest:
This Pod uses the
readonly-sa
service account in thereadonly-ns
namespace.Review the read-write Pod manifest:
This Pod uses the
admin-sa
service account in theadmin-ns
namespace.Deploy the test Pods:
kubectl apply -f manifests/admin-pod.yaml kubectl apply -f manifests/readonly-pod.yaml
The Pods might take a few minutes to start running. To monitor progress, run the following command:
watch kubectl get pods -n readonly-ns
When the Pod status changes to
RUNNING
, pressCtrl+C
to return to the command-line.
Test read-only access
Open a shell in the
readonly-test
Pod:kubectl exec -it readonly-test --namespace=readonly-ns -- /bin/bash
Try to read the secret:
gcloud secrets versions access 1 --secret=bq-readonly-key
The output is
key=my-api-key
.Try to write new data to the secret:
printf "my-second-api-key" | gcloud secrets versions add bq-readonly-key --data-file=-
The output is similar to the following:
ERROR: (gcloud.secrets.versions.add) PERMISSION_DENIED: Permission 'secretmanager.versions.add' denied for resource 'projects/PROJECT_ID/secrets/bq-readonly-key' (or it may not exist).
The Pod using the read-only service account can only read the secret, and can't write new data.
Exit the Pod:
exit
Test read-write access
Open a shell in the
admin-test
Pod:kubectl exec -it admin-test --namespace=admin-ns -- /bin/bash
Try to read the secret:
gcloud secrets versions access 1 --secret=bq-readonly-key
The output is
key=my-api-key
.Try to write new data to the secret:
printf "my-second-api-key" | gcloud secrets versions add bq-readonly-key --data-file=-
The output is similar to the following:
Created version [2] of the secret [bq-readonly-key].
Read the new secret version:
gcloud secrets versions access 2 --secret=bq-readonly-key
The output is
my-second-api-key
.Exit the Pod:
exit
The Pods only get the level of access you granted to the Kubernetes
service account used in the Pod manifest. Any Pods that use the admin-sa
Kubernetes account in the admin-ns
namespace can write new versions of the
secret, but any Pods in the readonly-ns
namespace that use the readonly-sa
Kubernetes service account can only read the secret.
Access secrets from your code
In this section, you do the following:
Deploy a sample application that reads your secret in Secret Manager using client libraries.
Check that the application can access your secret.
You should access Secret Manager secrets from your application code whenever possible, using the Secret Manager API.
Review the sample application source code:
This application calls the Secret Manager API to try and read the secret.
Review the sample application Pod manifest:
This manifest does the following:
- Creates a Pod in the
readonly-ns
namespace that uses thereadonly-sa
service account. - Pulls a sample application from a Google image registry. This
application calls the Secret Manager API using the
Google Cloud client libraries. You can view the application code
in
/main.go
in the repository. - Sets environment variables for the sample application to use.
- Creates a Pod in the
Replace environment variables in the sample application:
sed -i "s/YOUR_PROJECT_ID/PROJECT_ID/g" "manifests/secret-app.yaml"
Deploy the sample app:
kubectl apply -f manifests/secret-app.yaml
The Pod might take a few minutes to start working. If the Pod needs a new node in your cluster, you might notice
CrashLoopBackOff
type events while GKE provisions the node. The crashes stop when the node provisions successfully.Verify the secret access:
kubectl logs readonly-secret-test -n readonly-ns
The output is
my-second-api-key
. If the output is blank, the Pod might not be running yet. Wait a few minutes and try again.
Alternative approaches
If you need to mount your sensitive data to your Pods, use the Secret Manager add-on for GKE. This add-on deploys and manages the Google Cloud Secret Manager provider for the Kubernetes Secret Store CSI driver in your GKE clusters. For instructions, see Use Secret Manager add-on with GKE.
Providing secrets as mounted volumes has the following risks:
- Mounted volumes are susceptible to directory traversal attacks.
- Environment variables can be compromised due to misconfigurations such as opening a debug endpoint.
Whenever possible, we recommend that you programmatically access secrets through the Secret Manager API. For instructions, use the sample application in this tutorial or refer to Secret Manager client libraries.
Clean up
To avoid incurring charges to your Google Cloud account for the resources used in this tutorial, either delete the project that contains the resources, or keep the project and delete the individual resources.
Delete individual resources
Delete the cluster:
gcloud container clusters delete secret-cluster \ --region=us-central1
Optional: Delete the secret in Secret Manager:
gcloud secrets delete bq-readonly-key
If you don't do this step, the secret automatically expires because you set the
--ttl
flag during creation.
Delete the project
Delete a Google Cloud project:
gcloud projects delete PROJECT_ID
What's next
- Learn more about how Workload Identity Federation for GKE works.
- Explore reference architectures, diagrams, and best practices about Google Cloud. Take a look at our Cloud Architecture Center.