- Introduction
- Architecture
- Prerequisites
- Deployment
- Validation
- Set up Nginx
- Set up AppArmor-loader
- Set up Pod-labeler
- Tear down
- Troubleshooting
- Relevant Material
- Key Takeaways
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:
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.
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.
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.
Click the button below to run the demo in a Google 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
- Terraform >= 0.12.3
- Google Cloud SDK version >= 253.0.0
- kubectl matching the latest GKE version
- bash or bash compatible shell
- GNU Make 3.x or later
- A Google Cloud Platform project where you have permission to create networks
The Google Cloud SDK is used to interact with your GCP resources. Installation instructions for multiple platforms are available online.
The kubectl CLI is used to interteract with both Kubernetes Engine and kubernetes in general. Installation instructions for multiple platforms are available online.
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.
Prior to running this demo, ensure you have authenticated your gcloud client by running the following command:
gcloud auth application-default login
This project requires the following Google Cloud Service APIs to be enabled:
compute.googleapis.com
container.googleapis.com
cloudbuild.googleapis.com
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
project="YOUR_PROJECT"
region="YOUR_REGION"
zone="YOUR_ZONE"
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.
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.
...snip...
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
Again on the bastion host, execute:
kubectl apply -f manifests/nginx.yaml
deployment.apps/nginx created
service/nginx-lb created
networkpolicy.networking.k8s.io/nginx-from-external 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:
...
metadata:
annotations:
container.apparmor.security.beta.kubernetes.io/nginx: localhost/k8s-nginx
...
We can confirm this by executing the following command showing the Message
field:
kubectl describe pod -l app=nginx
...snip...
Labels: app=nginx
pod-template-hash=3314336736
Annotations: container.apparmor.security.beta.kubernetes.io/nginx=localhost/k8s-nginx
kubernetes.io/limit-ranger=LimitRanger plugin set: cpu request for container nginx
Status: Pending
Reason: AppArmor
Message: Cannot enforce AppArmor: profile "k8s-nginx" is not loaded
IP:
Controlled By: ReplicaSet/nginx-775877bc7b
Containers:
nginx:
Container ID:
...snip...
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
https://kubernetes.io/docs/concepts/workloads/controllers/daemonset/#what-is-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
networkpolicy.networking.k8s.io/deny-apparmor-communication 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 10.137.0.1 <none> 443/TCP 7m
nginx-lb LoadBalancer 10.137.15.122 [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:
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
role.rbac.authorization.k8s.io/pod-labeler created
serviceaccount/pod-labeler created
rolebinding.rbac.authorization.k8s.io/pod-labeler 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.
Log out of the bastion host and run the following to destroy the environment (type yes
at the prompt to confirm)
make teardown
...snip...
module.network.google_compute_subnetwork.cluster-subnet: Destruction complete after 26s
module.network.google_compute_network.gke-network: Destroying... (ID: kube-net)
module.network.google_compute_network.gke-network: Still destroying... (ID: kube-net, 10s elapsed)
module.network.google_compute_network.gke-network: Still destroying... (ID: kube-net, 20s elapsed)
module.network.google_compute_network.gke-network: Destruction complete after 25s
Destroy complete! Resources: 7 destroyed.
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
.
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: If this happens, simply re-run make create
.
Terraform occasionally complains about an invalid fingerprint, when updating certain resources. If you see the error below, simply re-run the command.
- AppArmor
- Terraform - Kubernetes Engine
- Bastion Host
- RBAC
- Securing Kubernetes Engine
- Security Contexts
This is not an officially supported Google product