Skip to content
This project demonstrates a series of best practices for improving the security of containerized applications deployed to Kubernetes Engine. You will deploy multiple instances of the same container image with a variety of security settings to illustrate the use of RBAC, security contexts, and AppArmor policies.
Branch: master
Clone or download
chrislovecnm Merge pull request #16 from alekhyal/master
Latest commit 97aa96b Mar 14, 2019
Type Name Latest commit message Commit time
Failed to load latest commit information.
images/custom-nginx Initial commit Jul 24, 2018
terraform Updated Readme, fixed terraform code for private cluster Feb 13, 2019
test updated boilerplate files and Makefile Jan 11, 2019
.gitignore Initial commit Jul 24, 2018
LICENSE Initial commit Jul 24, 2018
OWNERS Creating initial owners file Nov 8, 2018 Adding README-QWIKLABS file Mar 14, 2019 Updated Readme, fixed terraform code for private cluster Feb 13, 2019

Three examples of securing applications in Kubernetes Engine

Table of Contents


This tutorial demonstrates how Kubernetes Engine security features can be used to grant varying levels of privilege to applications, based on their particular requirements.

When configuring security, applications should be granted the smallest set of privileges that still allows them to operate correctly. When applications have more privileges than they need then they are more dangerous when compromised. In a Kubernetes cluster, these privileges can be grouped into the following broad levels:

  • Host access: describes what permissions an application has on it's host node, outside of its container. This is controlled via Pod and Container security contexts, as well as app armor profiles.
  • Network access: describes what other resources or workloads an application can access via the network. This is controlled with NetworkPolicies.
  • Kubernetes API access: describes which API calls an application is allowed to make against. API access is controlled using the Role Based Access Control (RBAC) model via Role and RoleBinding definitions.

In the sections below, you will explore three scenarios with distinct security requirements at each of these levels:

Scenario Host Access Network Access Kubernetes API Access
Hardened Web Server (NGINX) Minimal Must be able to serve port 80 None
Kubernetes Controller (Demo) Minimal Must be able to access the API Server List and Patch Pods
Privileged Daemonset (AppArmor Loader) Privileged None None


The tutorial uses three applications to illustrate the scenarios described above:

Hardened Web Server (nginx)

Creates an nginx deployment whose pods have their host-level access restricted by an AppArmor profile and whose network connectivity is restricted by a NetworkPolicy.

System Daemonset (AppArmor Loader)

Creates a daemonset responsible for loading (installing) the app armor profile applied to the nginx pods on each node. Loading profiles requires more privileges than can be provided via Capabilities, so it's containers are given full privileges via their SecurityContexts.

Simplified Kubernetes Controller (Pod Labeler)

The pod-labeler deployment creates a single pod that watches all other pods in the default namespace and periodically labels them. This requires access to the Kubernetes API server - which is configured via RBAC using a ServiceAccount, Role, and RoleMapping.


Run Demo in a Google Cloud Shell

Click the button below to run the demo in a Google Cloud Shell.

Open in Cloud Shell

All the tools for the demo are installed. When using Cloud Shell execute the following command in order to setup gcloud cli. When executing this command please setup your region and zone.

gcloud init


  1. Terraform >= 0.11.7
  2. Google Cloud SDK version >= 204.0.0
  3. kubectl matching the latest GKE version
  4. bash or bash compatible shell
  5. GNU Make 3.x or later
  6. A Google Cloud Platform project where you have permission to create networks

Install Cloud SDK

The Google Cloud SDK is used to interact with your GCP resources. Installation instructions for multiple platforms are available online.

Install kubectl CLI

The kubectl CLI is used to interteract with both Kubernetes Engine and kubernetes in general. Installation instructions for multiple platforms are available online.

Install Terraform

Terraform is used to automate the manipulation of cloud infrastructure. Its installation instructions are also available online.


The steps below will walk you through using terraform to deploy a Kubernetes Engine cluster that you will then use for installing a pod that uses an AppArmor profile and another pod that securely interacts with the Kubernetes API.

Authenticate gcloud

Prior to running this demo, ensure you have authenticated your gcloud client by running the following command:

gcloud auth application-default login

Setup this project

This project requires the following Google Cloud Service APIs to be enabled:


In addition, the terraform configuration takes three parameters to determine where the Kubernetes Engine cluster should be created:

  • project
  • region
  • zone

For simplicity, these parameters are to be specified in a file named terraform.tfvars, in the terraform directory. To ensure the appropriate APIs are enabled and to generate the terraform/terraform.tfvars file based on your gcloud defaults, run:

make setup-project

This will enable the necessary Service APIs, and it will also generate a terraform/terraform.tfvars file with the following keys. The values themselves will match the output of gcloud config list:

cat terraform/terraform.tfvars


If you need to override any of the defaults, simply replace the desired value(s) to the right of the equals sign(s). Be sure your replacement values are still double-quoted.

Provisioning the Kubernetes Engine Cluster

Next, apply the terraform configuration and create the environment with:

# From within the project root, use make to apply the terraform
make create


Once complete, terraform will output a message indicating successful creation of the cluster.

google_container_cluster.primary: Still creating... (3m20s elapsed)
google_container_cluster.primary: Still creating... (3m30s elapsed)
google_container_cluster.primary: Creation complete after 3m33s (ID: gke-security-demo)

Apply complete! Resources: 7 added, 0 changed, 0 destroyed.

Once the apply completes, ssh to the bastion host:

gcloud compute ssh gke-tutorial-bastion

and execute:

kubectl get pods --all-namespaces

NAMESPACE     NAME                                                          READY     STATUS        RESTARTS   AGE
kube-system   calico-node-g7mlz                                             2/2       Running       0          56s
kube-system   calico-node-rzbbc                                             2/2       Running       0          56s
kube-system   calico-node-vertical-autoscaler-8b959b949-tnwhm               1/1       Running       0          2m
kube-system   calico-node-vnw26                                             2/2       Running       0          56s
kube-system   calico-typha-dbf7c7b6d-bpm2r                                  1/1       Running       0          1m
kube-system   calico-typha-horizontal-autoscaler-5545fbd5d6-4wbgg           1/1       Running       0          2m
kube-system   calico-typha-vertical-autoscaler-54d8f88b84-2pwdz             1/1       Running       0          2m
kube-system   event-exporter-v0.2.1-5f5b89fcc8-vb5bc                        2/2       Running       0          2m
kube-system   fluentd-gcp-scaler-7c5db745fc-hcpr5                           1/1       Running       0          1m
kube-system   fluentd-gcp-v3.0.0-ks89p                                      2/2       Running       0          1m
kube-system   fluentd-gcp-v3.0.0-mfbg4                                      2/2       Running       0          1m
kube-system   heapster-v1.5.3-7ff6c8d4cc-tbf8d                              3/3       Running       0          1m
kube-system   ip-masq-agent-f8kdp                                           1/1       Running       0          1m
kube-system   ip-masq-agent-h7fbh                                           1/1       Running       0          1m
kube-system   ip-masq-agent-nlkl9                                           1/1       Running       0          1m
kube-system   kube-dns-788979dc8f-c6zv4                                     4/4       Running       0          1m
kube-system   kube-dns-788979dc8f-xt97h                                     4/4       Running       0          2m
kube-system   kube-dns-autoscaler-79b4b844b9-cmhns                          1/1       Running       0          1m
kube-system   kube-proxy-gke-gke-security-demo-default-pool-a55f1c91-6lxr   1/1       Running       0          1m
kube-system   kube-proxy-gke-gke-security-demo-default-pool-a55f1c91-7smw   1/1       Running       0          1m
kube-system   kube-proxy-gke-gke-security-demo-default-pool-a55f1c91-9khs   1/1       Running       0          1m
kube-system   l7-default-backend-5d5b9874d5-57mb5                           1/1       Running       0          2m
kube-system   metrics-server-v0.2.1-7486f5bd67-f68bq                        2/2       Running       0          1m

Set up Nginx

Again on the bastion host, execute:

kubectl apply -f manifests/nginx.yaml

deployment.apps/nginx created
service/nginx-lb created created

Then execute:

kubectl get pods

NAME                     READY     STATUS    RESTARTS   AGE
nginx-775877bc7b-5hk65   0/1       Blocked   0          13s
nginx-775877bc7b-jwcv9   0/1       Blocked   0          13s
nginx-775877bc7b-xc7p2   0/1       Blocked   0          13s

You should see that while the pods have been created, they're in a blocked state. The nginx pods are blocked because the manifest includes an AppArmor profile that doesn't exist on the nodes:

      annotations: localhost/k8s-nginx

We can confirm this by executing the following command showing the Message field:

kubectl describe pod -l app=nginx

Labels:         app=nginx
       plugin set: cpu request for container nginx
Status:         Pending
Reason:         AppArmor
Message:        Cannot enforce AppArmor: profile "k8s-nginx" is not loaded
Controlled By:  ReplicaSet/nginx-775877bc7b
    Container ID:

Set up AppArmor-loader

In order to resolve this, the relevant AppArmor profile must be loaded. because we don't know on which nodes the nginx pods will be allocated, we must deploy the AppArmor profile to all nodes. The way you'll deploy this, ensuring all nodes are covered, is via a daemonset

The included apparmor-loader.yaml is the manifest that is highly privileged, as it needs to write a file onto each node. The contents of that file are in the configmap included in the same; in order to deploy it, execute:

# On the bastion host
kubectl apply -f manifests/apparmor-loader.yaml

namespace/apparmor created
configmap/apparmor-profiles created
daemonset.apps/apparmor-loader created created

The nginx nodes are in backoff by this point, so the fastest way to resolve it is by deleting the existing nginx pods and allowing kubernetes to re-create them:

# On the bastion host
kubectl delete pods -l app=nginx

pod "nginx-775877bc7b-5hk65" deleted
pod "nginx-775877bc7b-jwcv9" deleted
pod "nginx-775877bc7b-xc7p2" deleted

After which, we can confirm that nginx is running successfully with:

kubectl get pods

NAME                     READY     STATUS    RESTARTS   AGE
nginx-775877bc7b-n2rxx   1/1       Running   0          28s
nginx-775877bc7b-q7qfb   1/1       Running   0          28s
nginx-775877bc7b-zzbdl   1/1       Running   0          28s

Another test that will show the successful deployment of nginx is by executing:

kubectl get services

NAME         TYPE           CLUSTER-IP      EXTERNAL-IP     PORT(S)        AGE
kubernetes   ClusterIP      <none>          443/TCP        7m
nginx-lb     LoadBalancer   [NGINX_IP]      80:30033/TCP   54s

the external IP listed by the nginx-lb is publicly accessible and can reached with curl, wget, or a web browser. Navigating to the IP displayed in place of [NGINX_IP] above, you will see:

nginx browser

Set up Pod-labeler

In this scenario you'll go through the process of deploying an application that requires access to the Kubernetes API. This pod will update the metadata of other pods.

First execute:

kubectl get pods --show-labels

NAME                          READY     STATUS              RESTARTS   AGE       LABELS
nginx-d99fc9b6c-ff6hw         1/1       Running             0          4m        app=nginx,pod-template-hash=855975627
nginx-d99fc9b6c-m7gc2         1/1       Running             0          4m        app=nginx,pod-template-hash=855975627
nginx-d99fc9b6c-zbkf9         1/1       Running             0          4m        app=nginx,pod-template-hash=855975627

Now execute:

kubectl apply -f manifests/pod-labeler.yaml created
serviceaccount/pod-labeler created created
deployment.apps/pod-labeler created

This can take up to two minutes to finish the creation of the new pod, but once this has finished, re-execute:

kubectl get pods --show-labels

NAME                          READY     STATUS    RESTARTS   AGE       LABELS
nginx-d99fc9b6c-ff6hw         1/1       Running   0          5m        app=nginx,pod-template-hash=855975627,updated=1530656545.33
nginx-d99fc9b6c-m7gc2         1/1       Running   0          5m        app=nginx,pod-template-hash=855975627,updated=1530656545.35
nginx-d99fc9b6c-zbkf9         1/1       Running   0          5m        app=nginx,pod-template-hash=855975627,updated=1530656545.36
pod-labeler-8845f6488-px9hl   1/1       Running   0          55s       app=pod-labeler,pod-template-hash=440192044,updated=1530656545.38

And you'll see that the pods have an additional "updated=..." label.

Tear down

Log out of the bastion host and run the following to destroy the environment (type yes at the prompt to confirm)

make teardown

...snip... Destruction complete after 26s Destroying... (ID: kube-net) Still destroying... (ID: kube-net, 10s elapsed) Still destroying... (ID: kube-net, 20s elapsed) Destruction complete after 25s

Destroy complete! Resources: 7 destroyed.


The install script fails with a Permission denied when running Terraform

The credentials that Terraform is using do not provide the necessary permissions to create resources in the selected projects. Ensure that the account listed in gcloud config list has necessary permissions to create resources. If it does, regenerate the application default credentials using gcloud auth application-default login.

Error during scp

Occasionally, the gke-tutorial-bastion host will not be ready for the subsequent scp command, in this case there will be an error that looks like: terraform scp error If this happens, simply re-run make create.

Invalid fingerprint error during Terraform operations

Terraform occasionally complains about an invalid fingerprint, when updating certain resources. If you see the error below, simply re-run the command. terraform fingerprint error

Relevant Material

This is not an officially supported Google product

You can’t perform that action at this time.
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session.