Performing Kubernetes Cis Benchmark With Kube Bench

Performing Kubernetes CIS Benchmark with kube-bench

Overview

kube-bench is an open-source Go tool by Aqua Security that runs the CIS Kubernetes Benchmark checks. It verifies control plane, etcd, worker node, and policy configurations against security best practices, producing actionable pass/fail/warn reports.

When to Use

• When conducting security assessments that involve performing kubernetes cis benchmark with kube bench
• When following incident response procedures for related security events
• When performing scheduled security testing or auditing activities
• When validating security controls through hands-on testing

Coverage Gaps & Validation

kube-bench only audits the node it runs on, so a "clean" run is usually an incomplete one:

• Node-only vs control-plane: on managed clusters (EKS/GKE/AKS) the control plane and etcd are not on your nodes, so --targets master,etcd checks are skipped or N/A - you are seeing worker (node) results only. Run the matching managed benchmark (--benchmark eks-1.2.0, gke-1.4.0, aks-1.0) so the tool doesn't apply self-managed checks that can't pass.
• WARN = manual, not pass: many 5.x policy checks (RBAC wildcards, Pod Security, NetworkPolicy per namespace) emit [WARN] requiring manual verification; counting only [FAIL] undercounts risk.
• Auto-detection misses: the file-permission checks (1.1.x, 4.1.x) only fire if kube-bench finds the manifests/kubeconfig at expected paths - non-standard or kubeadm-vs-binary layouts silently skip them.
• Config-file, not runtime: kube-bench reads flags/files; it does not confirm the API server is actually running with those flags if config and process diverge.

Validate: run every relevant target (--targets master,etcd,node,policies on self-managed) and confirm the summary check count matches the benchmark's expected total - a low total means sections were skipped. Cross-check flag findings against the live process (ps aux | grep kube-apiserver) and re-run after remediation to confirm FAIL→PASS rather than FAIL→skipped.

Prerequisites

• Kubernetes cluster (v1.24+)
• kubectl with cluster-admin access
• Node access for direct runs or privileged pod access

Installation

# Binary installation
curl -L https://github.com/aquasecurity/kube-bench/releases/download/v0.7.3/kube-bench_0.7.3_linux_amd64.tar.gz | tar xz
sudo mv kube-bench /usr/local/bin/

# Run as Kubernetes Job
kubectl apply -f https://raw.githubusercontent.com/aquasecurity/kube-bench/main/job.yaml
kubectl logs job/kube-bench

# Run as a pod with host access
kubectl apply -f https://raw.githubusercontent.com/aquasecurity/kube-bench/main/job-master.yaml
kubectl apply -f https://raw.githubusercontent.com/aquasecurity/kube-bench/main/job-node.yaml

Running Benchmarks

Full Benchmark

# Run all checks (auto-detects node type)
kube-bench run

# Run with JSON output
kube-bench run --json > kube-bench-results.json

# Run with JUnit output for CI
kube-bench run --junit > kube-bench-results.xml

Component-Specific Checks

# Control plane (master) checks
kube-bench run --targets master

# Worker node checks
kube-bench run --targets node

# etcd checks
kube-bench run --targets etcd

# Policies checks
kube-bench run --targets policies

# Control plane + etcd
kube-bench run --targets master,etcd

Managed Kubernetes

# Amazon EKS
kube-bench run --benchmark eks-1.2.0

# Google GKE
kube-bench run --benchmark gke-1.4.0

# Azure AKS
kube-bench run --benchmark aks-1.0

# Red Hat OpenShift
kube-bench run --benchmark rh-1.0

Filtering Results

# Show only failures
kube-bench run --targets master | grep "\[FAIL\]"

# Run specific check
kube-bench run --check 1.2.1

# Run check group
kube-bench run --group 1.2

CIS Benchmark Sections

Output Example

[INFO] 1 Control Plane Security Configuration
[INFO] 1.1 Control Plane Node Configuration Files
[PASS] 1.1.1 Ensure that the API server pod specification file permissions are set to 600
[PASS] 1.1.2 Ensure that the API server pod specification file ownership is set to root:root
[FAIL] 1.1.3 Ensure that the controller manager pod specification file permissions are set to 600
[WARN] 1.1.4 Ensure that the scheduler pod specification file permissions are set to 600

== Summary ==
45 checks PASS
12 checks FAIL
8 checks WARN
0 checks INFO

CI/CD Integration

GitHub Actions

name: CIS Benchmark
on:
  schedule:
    - cron: '0 6 * * 1'

jobs:
  kube-bench:
    runs-on: ubuntu-latest
    steps:
      - name: Configure kubectl
        uses: azure/setup-kubectl@v3

      - name: Run kube-bench
        run: |
          kubectl apply -f https://raw.githubusercontent.com/aquasecurity/kube-bench/main/job.yaml
          kubectl wait --for=condition=complete job/kube-bench --timeout=120s
          kubectl logs job/kube-bench > kube-bench-report.txt

      - name: Check for failures
        run: |
          FAILS=$(grep -c "\[FAIL\]" kube-bench-report.txt || true)
          echo "Failed checks: $FAILS"
          if [ "$FAILS" -gt 0 ]; then
            echo "::warning::$FAILS CIS benchmark checks failed"
          fi

      - name: Upload report
        uses: actions/upload-artifact@v4
        with:
          name: kube-bench-report
          path: kube-bench-report.txt

Remediation Examples

1.2.1 - Ensure --anonymous-auth is set to false

# /etc/kubernetes/manifests/kube-apiserver.yaml
spec:
  containers:
  - command:
    - kube-apiserver
    - --anonymous-auth=false

4.2.1 - Ensure --anonymous-auth is set to false on kubelet

# /var/lib/kubelet/config.yaml
authentication:
  anonymous:
    enabled: false
  webhook:
    enabled: true

5.2.1 - Minimize wildcard RBAC

# Find roles with wildcard permissions
kubectl get clusterroles -o json | jq '.items[] | select(.rules[].resources[] == "*") | .metadata.name'

Best Practices

1. Run kube-bench before and after cluster provisioning
2. Schedule weekly scans via CronJob for drift detection
3. Export JSON for SIEM/compliance reporting
4. Fix FAIL items first, then address WARN items
5. Use benchmark profiles matching your Kubernetes distribution
6. Track score over time to measure security posture improvement
7. Combine with admission controllers to prevent drift