implementing-pod-security-admission-controller

Implementing Pod Security Admission Controller

Overview

Pod Security Admission (PSA) is a built-in Kubernetes admission controller (stable since v1.25) that enforces Pod Security Standards at the namespace level. It replaces the deprecated PodSecurityPolicy (PSP) and provides three security profiles: Privileged, Baseline, and Restricted, with three enforcement modes: enforce, audit, and warn.

When to Use

• Wiring PSA up on a cluster for the first time
• Setting cluster-wide default enforcement via AdmissionConfiguration
• Exempting system namespaces, service accounts, or runtime classes from enforcement
• Debugging why a pod was rejected, or why enforcement is silently not firing
• Staging a safe rollout: warn and audit first, enforce once violations reach zero
• Pulling PSA violations out of the kube-apiserver audit log

Not this skill: deciding which profile a workload should run under, or what

securityContext changes Restricted demands. Use

implementing-kubernetes-pod-security-standards.

Prerequisites

• Kubernetes v1.25+ (PSA is stable/GA)
• kubectl with cluster-admin access
• No dependency on external tools - PSA is built into kube-apiserver

Pod Security Standards

Privileged Profile

• Unrestricted - No restrictions applied
• Use case: System-level pods (kube-system, monitoring)

Baseline Profile

• Minimally restrictive - Prevents known privilege escalation
• Blocks: privileged containers, hostPID, hostIPC, hostNetwork, hostPorts, certain volume types, adding capabilities beyond runtime defaults

Restricted Profile

• Heavily restricted - Follows security best practices
• Requires: non-root, drop ALL capabilities, seccomp RuntimeDefault, read-only root filesystem considerations
• Blocks: Everything in Baseline plus running as root, privilege escalation, non-approved volume types

Enforcement Modes

| Mode | Behavior | Use Case |

|------|----------|----------|

| enforce | Reject pods violating policy | Production enforcement |

| audit | Log violations to audit log | Pre-enforcement assessment |

| warn | Show warnings to user | Developer feedback |

Implementation

Apply to Namespace via Labels

# Restricted enforcement with audit and warn
apiVersion: v1
kind: Namespace
metadata:
  name: production
  labels:
    pod-security.kubernetes.io/enforce: restricted
    pod-security.kubernetes.io/enforce-version: v1.28
    pod-security.kubernetes.io/audit: restricted
    pod-security.kubernetes.io/audit-version: v1.28
    pod-security.kubernetes.io/warn: restricted
    pod-security.kubernetes.io/warn-version: v1.28

# Baseline enforcement for staging
apiVersion: v1
kind: Namespace
metadata:
  name: staging
  labels:
    pod-security.kubernetes.io/enforce: baseline
    pod-security.kubernetes.io/enforce-version: v1.28
    pod-security.kubernetes.io/audit: restricted
    pod-security.kubernetes.io/audit-version: v1.28
    pod-security.kubernetes.io/warn: restricted
    pod-security.kubernetes.io/warn-version: v1.28

# Privileged for system namespaces
apiVersion: v1
kind: Namespace
metadata:
  name: kube-system
  labels:
    pod-security.kubernetes.io/enforce: privileged

Apply Labels with kubectl

# Set restricted enforcement
kubectl label namespace production \
  pod-security.kubernetes.io/enforce=restricted \
  pod-security.kubernetes.io/enforce-version=v1.28 \
  pod-security.kubernetes.io/audit=restricted \
  pod-security.kubernetes.io/warn=restricted

# Set baseline enforcement
kubectl label namespace staging \
  pod-security.kubernetes.io/enforce=baseline \
  pod-security.kubernetes.io/audit=restricted \
  pod-security.kubernetes.io/warn=restricted

# Check current labels
kubectl get namespace production -o jsonpath='{.metadata.labels}' | jq .

Dry-Run Testing

# Test what would happen with restricted policy on a namespace
kubectl label --dry-run=server --overwrite namespace staging \
  pod-security.kubernetes.io/enforce=restricted

# Output shows existing pods that would violate the policy
# Warning: existing pods in namespace "staging" violate the new PodSecurity enforce level "restricted:latest"

Cluster-Wide Defaults (AdmissionConfiguration)

# /etc/kubernetes/psa-config.yaml
apiVersion: apiserver.config.k8s.io/v1
kind: AdmissionConfiguration
plugins:
  - name: PodSecurity
    configuration:
      apiVersion: pod-security.admission.config.k8s.io/v1
      kind: PodSecurityConfiguration
      defaults:
        enforce: baseline
        enforce-version: latest
        audit: restricted
        audit-version: latest
        warn: restricted
        warn-version: latest
      exemptions:
        usernames: []
        runtimeClasses: []
        namespaces:
          - kube-system
          - kube-public
          - kube-node-lease
          - calico-system
          - gatekeeper-system
          - monitoring
          - falco

Apply to API Server

# Add to kube-apiserver manifests
# /etc/kubernetes/manifests/kube-apiserver.yaml
spec:
  containers:
  - command:
    - kube-apiserver
    - --admission-control-config-file=/etc/kubernetes/psa-config.yaml
    volumeMounts:
    - name: psa-config
      mountPath: /etc/kubernetes/psa-config.yaml
      readOnly: true
  volumes:
  - name: psa-config
    hostPath:
      path: /etc/kubernetes/psa-config.yaml
      type: File

Compliant Pod Examples

Restricted-Compliant Pod

apiVersion: v1
kind: Pod
metadata:
  name: restricted-pod
  namespace: production
spec:
  securityContext:
    runAsNonRoot: true
    runAsUser: 1000
    runAsGroup: 3000
    fsGroup: 2000
    seccompProfile:
      type: RuntimeDefault
  automountServiceAccountToken: false
  containers:
    - name: app
      image: myregistry/myapp:v1.0.0
      securityContext:
        allowPrivilegeEscalation: false
        readOnlyRootFilesystem: true
        capabilities:
          drop:
            - ALL
      resources:
        limits:
          cpu: 500m
          memory: 256Mi
        requests:
          cpu: 100m
          memory: 128Mi
      volumeMounts:
        - name: tmp
          mountPath: /tmp
  volumes:
    - name: tmp
      emptyDir: {}

Baseline-Compliant Pod

apiVersion: v1
kind: Pod
metadata:
  name: baseline-pod
  namespace: staging
spec:
  containers:
    - name: app
      image: myregistry/myapp:v1.0.0
      securityContext:
        allowPrivilegeEscalation: false
      resources:
        limits:
          cpu: 500m
          memory: 256Mi

Migration from PodSecurityPolicy

Step 1: Audit Current State

# Check existing PSPs
kubectl get psp

# Check which service accounts use which PSP
kubectl get clusterrolebinding -o json | \
  jq '.items[] | select(.roleRef.name | startswith("psp-")) | {name: .metadata.name, subjects: .subjects}'

Step 2: Map PSP to PSA Profiles

# For each namespace, determine required PSA level
for ns in $(kubectl get ns -o jsonpath='{.items[*].metadata.name}'); do
  echo "Namespace: $ns"
  kubectl label --dry-run=server namespace $ns \
    pod-security.kubernetes.io/enforce=restricted 2>&1 | head -5
done

Step 3: Apply PSA Labels (Audit First)

# Start with audit mode
kubectl label namespace production \
  pod-security.kubernetes.io/audit=restricted \
  pod-security.kubernetes.io/warn=restricted

Step 4: Review and Fix Violations

# Check audit logs for violations
kubectl get events --field-selector reason=FailedCreate -A

Step 5: Enable Enforcement

kubectl label namespace production \
  pod-security.kubernetes.io/enforce=restricted

Monitoring

# Check PSA violations in events
kubectl get events --all-namespaces --field-selector reason=FailedCreate

# Check audit logs
kubectl logs -n kube-system kube-apiserver-* | grep "pod-security.kubernetes.io"

# List namespace PSA labels
kubectl get namespaces -L pod-security.kubernetes.io/enforce

Best Practices

1. Start with audit+warn before enforce to assess impact
2. Use dry-run to test enforcement before applying
3. Exempt system namespaces (kube-system, monitoring) in cluster defaults
4. Pin version (enforce-version) for predictable behavior across upgrades
5. Set cluster-wide baseline as default, then restrict specific namespaces
6. Combine with Gatekeeper for additional custom policies beyond PSA
7. Use restricted profile for all production workloads
8. Document exemptions with clear justification