kube-bench VS Kyverno

However, no matter how well our applications are secured, the security of our entire IT environment ultimately depends on the security of our infrastructure. Therefore, in the lab to follow, we will shift our focus away from Kubernetes workloads and instead explore how we can evaluate and improve upon the security of our Kubernetes clusters with kube-bench, the industry-leading Kubernetes benchmarking solution developed by Aqua.

The official repository can be found here with detailed installation instructions.

curl -L [https://github.com/aquasecurity/kube-bench/releases/download/v0.6.10/kube-bench_0.6.10_linux_amd64.deb](https://github.com/aquasecurity/kube-bench/releases/download/v0.6.2/kube-bench_0.6.2_linux_amd64.deb) -o kube-bench_0.6.10_linux_amd64.deb

I haven't used Kubernetes for a while, but shouldn't kube-bench (1) be enough? Do you have to check anything manually?

(1) https://github.com/aquasecurity/kube-bench

kube-bench checks whether Kubernetes is deployed securely by running the checks documented in the CIS Kubernetes Benchmark. We can deploy kube-bench as a Job that runs daily and consume its report in CI/CD to pass or fail the pipeline based on the level of severity.

kube-bench can be executed as a simple command on the host, as a container on the host using Docker command, or as a job inside Kubernetes Cluster. In case it is run inside a container/pod, it will need access to the PID namespace of the host system. The methods to run kube-bench in AKS, EKS, GKE, On-prem cluster, Openshift and ACK (Alibaba Cloud Container Service For Kubernetes) are different but well documented.

Check this out: kube-bench - CIS Benchmark

Kube-bench, written as a Go application, is deployable as a container. Ready-made job.yaml files make it easy to run Kube-bench inside a Kubernetes cluster or on a managed Kubernetes service, such as Azure Kubernetes Service (AKS), Amazon Elastic Kubernetes Service (EKS), Google Kubernetes Engine (GKE), or OpenShift. Here's a link to Kube-Bench on Github

kube-bench - Checks whether Kubernetes is deployed securely by running CIS Kubernetes Benchmark

Anyway, I haven’t checked for sure as I’m away from laptop but it should be possible to use something like Kyverno to block that operation. We had to do similar in the past to hotfix a bug in our CLI tool. I wrote a blog post about it that might give you an idea: https://www.giantswarm.io/blog/restricting-cluster-admin-permissions

Cosign is used for signing containers through a variety of different methods. It has strong integration with other open source tools, such as Kyverno.

cosign: https://docs.sigstore.dev/cosign/overview/ kyverno: https://kyverno.io/

Kyverno - Kubernetes Native Policy Management

You could use a policy tool like kyverno or OPA.

Kyverno is present in the management cluster;

You do still have to create a policy for every namespace, but don't have to worry about labeling individual pods. We're starting to move to Helm/kustomize for our namespaces to deploy default things like network policies to each one, and we're also starting to use kyverno more, which I think is a little more purpose built for this type of thing than metacontroller is.

I knew it was unsupported so about 6 months ago I had started an effort to switch to Kyverno, which is far better and actually supported. The version of Kyverno I was using had a v1beta1 AdmissionController. Fortunately that was in a helm chart so easily caught by pluto before my upgrade.

Kyverno Kyverno is a policy engine designed for Kubernetes, Kyverno policies can validate, mutate, and generate Kubernetes resources plus ensure OCI image supply chain security.