kodekloud-engineer/kubernetes/cks/exam-by-claude/solutions.md

# CKS Mock — Solutions (k8s v1.35)

Reference answers. There's usually more than one valid path; these are the exam-fast ones. Control-plane file edits assume `docker exec -it cks-control-plane bash` (or `cks-worker` for node tasks).

> **Before ANY apiserver edit:** `cp /etc/kubernetes/manifests/kube-apiserver.yaml ~/kube-apiserver.yaml.bak` inside the node. If the pod won't come back, `cp` it back.

---

## Task 1 — NetworkPolicy

```yaml
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata: {name: default-deny, namespace: prod}
spec:
  podSelector: {}
  policyTypes: [Ingress]
---
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata: {name: allow-frontend, namespace: prod}
spec:
  podSelector: {matchLabels: {app: db}}
  policyTypes: [Ingress]
  ingress:
    - from:
        - podSelector: {matchLabels: {app: frontend}}
      ports:
        - {protocol: TCP, port: 5432}
```

Verify:
```bash
kubectl -n prod exec deploy/frontend -- curl -s --max-time 3 db:5432   # -> db-ok
kubectl -n prod exec deploy/attacker -- curl -s --max-time 3 db:5432   # -> timeout
```
Gotcha: this cluster runs Cilium (not kindnetd), which enforces NetworkPolicy deterministically — no "if your build doesn't enforce it" caveat. If the attacker curl doesn't actually time out, don't blame the CNI; debug it:
```bash
cilium status                                              # agents healthy on all 3 nodes?
kubectl -n kube-system exec ds/cilium -- cilium-dbg endpoint list   # policy enforcement column for the db/attacker/frontend endpoints
```

---

## Task 2 — Ingress TLS

```bash
openssl req -x509 -nodes -days 365 -newkey rsa:2048 \
  -keyout tls.key -out tls.crt -subj "/CN=hello.cks.local/O=cks"
kubectl -n web create secret tls hello-tls --cert=tls.crt --key=tls.key
```
```yaml
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: hello
  namespace: web
spec:
  ingressClassName: nginx
  tls:
    - hosts: [hello.cks.local]
      secretName: hello-tls
  rules:
    - host: hello.cks.local
      http:
        paths:
          - path: /
            pathType: Prefix
            backend: {service: {name: hello, port: {number: 80}}}
```
Verify:
```bash
curl -k --resolve hello.cks.local:443:127.0.0.1 https://hello.cks.local/
```

---

## Task 3 — kube-bench remediation

Run it (Job is easiest inside kind):
```bash
kubectl apply -f https://raw.githubusercontent.com/aquasecurity/kube-bench/main/job.yaml
kubectl logs -f job/kube-bench | less
```
Or download the binary and `kube-bench run --targets master,node` on the node.

Kubelet — edit on the node `/var/lib/kubelet/config.yaml`:
```yaml
authentication:
  anonymous:
    enabled: false
  webhook:
    enabled: true
authorization:
  mode: Webhook
readOnlyPort: 0
```
```bash
systemctl restart kubelet
```
API server — `/etc/kubernetes/manifests/kube-apiserver.yaml`, add to `command`:
```
- --profiling=false
```
Re-run kube-bench; those checks flip to PASS.

---

## Task 4 — RBAC least privilege

Replace the Role (binding already points at `ci-runner-role`):
```yaml
apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata: {name: ci-runner-role, namespace: dev}
rules:
  - apiGroups: [""]
    resources: ["pods", "configmaps"]
    verbs: ["get", "list", "watch"]
```
```bash
kubectl -n dev replace -f role.yaml   # or edit
kubectl auth can-i get pods    --as=system:serviceaccount:dev:ci-runner -n dev   # yes
kubectl auth can-i delete pods --as=system:serviceaccount:dev:ci-runner -n dev   # no
kubectl auth can-i '*' secrets --as=system:serviceaccount:dev:ci-runner -n dev   # no
```

---

## Task 5 — API server hardening

`/etc/kubernetes/manifests/kube-apiserver.yaml`, in `command`:
```
- --anonymous-auth=false
- --profiling=false
- --enable-admission-plugins=NodeRestriction   # merge with existing list, comma-separated
```
If `--enable-admission-plugins` already exists, append `,NodeRestriction` to it — don't add a second flag.
Wait for restart:
```bash
kubectl -n kube-system get pod -l component=kube-apiserver -w
```
Note: with `--anonymous-auth=false`, `/healthz` and `/livez` still work for the kubelet because it authenticates; unauthenticated curl gets 401.

---

## Task 6 — automountServiceAccountToken

```bash
kubectl -n apps patch sa web-sa -p '{"automountServiceAccountToken": false}'
kubectl -n apps rollout restart deploy/web
kubectl -n apps exec deploy/web -- ls /var/run/secrets/kubernetes.io/serviceaccount 2>&1  # No such file
```
(Alternatively set `automountServiceAccountToken: false` in the pod template — SA-level is cleaner here and is what the task asks.)

---

## Task 7 — AppArmor (GA securityContext API)

Load profile on the worker:
```bash
docker exec cks-worker apparmor_parser -q /root/deny-write
docker exec cks-worker aa-status | grep k8s-deny-write   # enforce
```
Pod — **use the appArmorProfile field, not the beta annotation**:
```yaml
apiVersion: v1
kind: Pod
metadata: {name: locked, namespace: sysh}
spec:
  nodeName: cks-worker
  securityContext:
    appArmorProfile:
      type: Localhost
      localhostProfile: k8s-deny-write
  containers:
    - name: c
      image: busybox:1.36
      command: ["sh","-c","sleep 1h"]
```
```bash
kubectl -n sysh exec locked -- sh -c 'echo x > /tmp/x'   # Permission denied
```
Gotcha: the profile only exists on `cks-worker`, so you must pin the pod there or the load must be on every schedulable node.

---

## Task 8 — seccomp

```yaml
apiVersion: v1
kind: Pod
metadata: {name: traced, namespace: sysh}
spec:
  nodeName: cks-worker
  securityContext:
    seccompProfile:
      type: Localhost
      localhostProfile: profiles/audit.json
  containers:
    - {name: c, image: busybox:1.36, command: ["sh","-c","sleep 1h"]}
---
apiVersion: v1
kind: Pod
metadata: {name: defaulted, namespace: sysh}
spec:
  nodeName: cks-worker
  securityContext:
    seccompProfile: {type: RuntimeDefault}
  containers:
    - {name: c, image: busybox:1.36, command: ["sh","-c","sleep 1h"]}
```
`localhostProfile` is relative to the kubelet seccomp root (`/var/lib/kubelet/seccomp`), so `profiles/audit.json` resolves to `/var/lib/kubelet/seccomp/profiles/audit.json`.

---

## Task 9 — Pod Security Admission (restricted)

Label the namespace:
```bash
kubectl label ns restricted-ns \
  pod-security.kubernetes.io/enforce=restricted \
  pod-security.kubernetes.io/enforce-version=latest \
  pod-security.kubernetes.io/warn=restricted \
  pod-security.kubernetes.io/audit=restricted --overwrite
```
Fix the Deployment pod template to satisfy `restricted`:
```yaml
    spec:
      securityContext:
        runAsNonRoot: true
        seccompProfile: {type: RuntimeDefault}
      containers:
        - name: c
          image: nginx:1.27
          securityContext:
            allowPrivilegeEscalation: false
            privileged: false
            runAsNonRoot: true
            capabilities: {drop: ["ALL"]}
```
nginx:1.27 wants to bind :80 — under `runAsNonRoot` it'll fail unless you use an unprivileged image/port. For the exam-grade "complies + rolls out," use `nginxinc/nginx-unprivileged:1.27` (listens on 8080) or set `runAsUser: 101`. The graded bit is restricted-compliance; the unprivileged image makes it actually run.

---

## Task 10 — Encryption at rest

Key + config on the control-plane:
```bash
mkdir -p /etc/kubernetes/enc
head -c 32 /dev/urandom | base64   # copy the value
cat > /etc/kubernetes/enc/enc.yaml <<EOF
apiVersion: apiserver.config.k8s.io/v1
kind: EncryptionConfiguration
resources:
  - resources: ["secrets"]
    providers:
      - aescbc:
          keys:
            - name: key1
              secret: <PASTE_BASE64_32B_KEY>
      - identity: {}
EOF
```
apiserver manifest — add flag + volume + mount:
```yaml
# command:
- --encryption-provider-config=/etc/kubernetes/enc/enc.yaml
# volumes:
- name: enc
  hostPath: {path: /etc/kubernetes/enc, type: DirectoryOrCreate}
# volumeMounts:
- name: enc
  mountPath: /etc/kubernetes/enc
  readOnly: true
```
After apiserver is back, rewrite all secrets:
```bash
kubectl get secrets -A -o json | kubectl replace -f -
```
Verify with etcdctl inside the control-plane:
```bash
ETCDCTL_API=3 etcdctl \
  --cacert=/etc/kubernetes/pki/etcd/ca.crt \
  --cert=/etc/kubernetes/pki/etcd/server.crt \
  --key=/etc/kubernetes/pki/etcd/server.key \
  get /registry/secrets/vault-ns/pre-existing | hexdump -C | head
# expect: k8s:enc:aescbc:v1:key1:...  and NO plaintext SUPERSECRET
```

---

## Task 11 — RuntimeClass

```yaml
apiVersion: node.k8s.io/v1
kind: RuntimeClass
metadata: {name: gvisor}
handler: runsc
---
apiVersion: v1
kind: Pod
metadata: {name: sandboxed, namespace: runtime}
spec:
  runtimeClassName: gvisor
  containers:
    - {name: c, image: nginx:1.27}
```
Stays Pending in kind (no `runsc` handler in containerd) — expected. Config is the deliverable.

---

## Task 12 — ValidatingAdmissionPolicy

```yaml
apiVersion: admissionregistration.k8s.io/v1
kind: ValidatingAdmissionPolicy
metadata: {name: pod-hardening}
spec:
  failurePolicy: Fail
  matchConstraints:
    resourceRules:
      - apiGroups: [""]
        apiVersions: ["v1"]
        operations: ["CREATE","UPDATE"]
        resources: ["pods"]
  validations:
    - expression: >-
        !object.spec.containers.exists(c,
          has(c.securityContext) && has(c.securityContext.privileged)
          && c.securityContext.privileged == true)
      message: "privileged containers are not allowed"
    - expression: >-
        !has(object.spec.volumes) ||
        !object.spec.volumes.exists(v, has(v.hostPath))
      message: "hostPath volumes are not allowed"
---
apiVersion: admissionregistration.k8s.io/v1
kind: ValidatingAdmissionPolicyBinding
metadata: {name: pod-hardening-binding}
spec:
  policyName: pod-hardening
  validationActions: ["Deny"]
  matchResources: {}   # all namespaces
```
Also check `initContainers`/`ephemeralContainers` if you want it airtight; for the task, `containers` is what's graded. Test:
```bash
kubectl run bad --image=nginx:1.27 --privileged           # denied
kubectl run ok  --image=nginx:1.27                        # admitted
```

---

## Task 13 — Trivy scan & evict

```bash
for img in nginx:1.19.0 debian:10 nginx:1.27; do
  echo "== $img =="
  trivy image --severity HIGH,CRITICAL --quiet --scanners vuln "$img" | tail -5
done
```
`nginx:1.19.0` and `debian:10` will show HIGH/CRITICAL; delete those pods:
```bash
kubectl -n images delete pod legacy-app old-debian
kubectl -n images get pods   # clean-app remains
```
Exam tip: map image→pod first (`kubectl -n images get pods -o custom-columns=POD:.metadata.name,IMG:.spec.containers[*].image`), scan each, delete by HIGH/CRITICAL count > 0.

---

## Task 14 — Kyverno policies

```yaml
apiVersion: kyverno.io/v1
kind: ClusterPolicy
metadata: {name: disallow-latest-tag}
spec:
  validationFailureAction: Enforce
  background: false
  rules:
    - name: require-explicit-tag
      match: {any: [{resources: {kinds: [Pod]}}]}
      validate:
        message: "images must not use :latest or an empty tag"
        pattern:
          spec:
            containers:
              - image: "!*:latest"
    - name: require-tag-present
      match: {any: [{resources: {kinds: [Pod]}}]}
      validate:
        message: "image tag is required"
        pattern:
          spec:
            containers:
              - image: "*:*"
---
apiVersion: kyverno.io/v1
kind: ClusterPolicy
metadata: {name: allowed-registries}
spec:
  validationFailureAction: Enforce
  background: false
  rules:
    - name: only-cks-registry
      match: {any: [{resources: {kinds: [Pod]}}]}
      validate:
        message: "images must come from registry.cks.local"
        pattern:
          spec:
            containers:
              - image: "registry.cks.local/*"
```
Test:
```bash
kubectl -n team-a run bad  --image=nginx:latest                    # denied (latest)
kubectl -n team-a run bad2 --image=docker.io/nginx:1.27            # denied (registry)
kubectl -n team-a run good --image=registry.cks.local/nginx:1.27   # admitted
```
Note: recent Kyverno also exposes the newer `validate.foreach`/CEL syntax; the pattern form above is the fastest to write under time.

---

## Task 15 — ImagePolicyWebhook

`/etc/kubernetes/admission/admission-config.yaml`:
```yaml
apiVersion: apiserver.config.k8s.io/v1
kind: AdmissionConfiguration
plugins:
  - name: ImagePolicyWebhook
    configuration:
      imagePolicy:
        kubeConfigFile: /etc/kubernetes/admission/imagepolicy-kubeconfig.yaml
        allowTTL: 50
        denyTTL: 50
        retryBackoff: 500
        defaultAllow: false
```
`/etc/kubernetes/admission/imagepolicy-kubeconfig.yaml`:
```yaml
apiVersion: v1
kind: Config
clusters:
  - name: image-checker
    cluster:
      server: https://image-checker.local/check   # no live backend needed for the drill
users:
  - name: apiserver
contexts:
  - name: default
    context: {cluster: image-checker, user: apiserver}
current-context: default
```
apiserver manifest:
```yaml
# command:
- --enable-admission-plugins=NodeRestriction,ImagePolicyWebhook   # merge with existing
- --admission-control-config-file=/etc/kubernetes/admission/admission-config.yaml
# volume + mount for /etc/kubernetes/admission (readOnly)
```
With `defaultAllow: false` and no reachable backend, `kubectl run t --image=nginx:1.27` → rejected by ImagePolicyWebhook. Flip to `defaultAllow: true` afterwards if you want the cluster usable.

---

## Task 16 — Falco custom rule

`/etc/falco/falco_rules.local.yaml`:
```yaml
- rule: Shell in container
  desc: Detect a shell spawned inside a container
  condition: >
    spawned_process and container
    and proc.name in (sh, bash)
  output: >
    Shell in container (container_id=%container.id container_name=%container.name
    proc=%proc.name user=%user.name)
  priority: WARNING
  tags: [container, shell, mitre_execution]
```
```bash
systemctl restart falco     # or: falco -r /etc/falco/falco_rules.yaml -r /etc/falco/falco_rules.local.yaml
# trigger:
kubectl run trigger --image=busybox:1.36 -- sh -c 'sleep 1h'
kubectl exec -it trigger -- sh
# capture:
journalctl -u falco | grep "Shell in container" | tail -1 > /opt/course/falco-hits.txt
```
Gotchas: `container` macro excludes host processes; `spawned_process` = execve. If running Falco via systemd, alerts land in `journalctl -u falco` (and/or `/var/log/syslog`) — grab from whichever your output channel is. Check the rule loaded: `falco --list | grep -i "Shell in container"` or watch startup logs for parse errors.

---

## Task 17 — Audit logging

`/etc/kubernetes/audit/policy.yaml`:
```yaml
apiVersion: audit.k8s.io/v1
kind: Policy
omitStages: ["RequestReceived"]
rules:
  - level: RequestResponse
    resources:
      - group: ""
        resources: ["pods"]
    verbs: ["create","update","patch","delete"]
  - level: Metadata
    resources:
      - group: ""
        resources: ["secrets","configmaps"]
  - level: None
```
apiserver manifest — flags:
```yaml
- --audit-policy-file=/etc/kubernetes/audit/policy.yaml
- --audit-log-path=/var/log/kubernetes/audit/audit.log
- --audit-log-maxage=7
- --audit-log-maxbackup=2
- --audit-log-maxsize=50
```
volumes + mounts:
```yaml
# volumes:
- name: audit-policy
  hostPath: {path: /etc/kubernetes/audit, type: DirectoryOrCreate}
- name: audit-logs
  hostPath: {path: /var/log/kubernetes/audit, type: DirectoryOrCreate}
# volumeMounts:
- name: audit-policy
  mountPath: /etc/kubernetes/audit
  readOnly: true
- name: audit-logs
  mountPath: /var/log/kubernetes/audit
  readOnly: false
```
Verify inside the control-plane:
```bash
kubectl -n vault-ns get secret pre-existing
tail -f /var/log/kubernetes/audit/audit.log | grep pre-existing   # Metadata level, no body
```
Big gotcha: the audit-logs mount must be `readOnly: false` or the apiserver crashloops silently. And the log dir hostPath is *inside the control-plane container* — that's fine here.

---

### Reset between attempts
```bash
kind delete cluster --name cks
./bootstrap.sh && ./seed.sh
```