Skip to content

Platform Compatibility

CloudTaser runs on any Kubernetes cluster with Linux nodes. However, the protection score varies by platform due to kernel configuration differences.

Protection Score by Platform

Platform Node OS memfd_secret kprobe enforcement Max Score
GKE Ubuntu ✅ ✅ 100/115
GKE COS ✅ ❌ 85/115
AKS Ubuntu 22.04 ❌ ✅ 85/115
AKS Azure Linux ✅ ❌ 85/115
EKS Ubuntu ✅ ✅ ~100/115
EKS Amazon Linux ❓ ❓ TBD
EKS Bottlerocket ❓ ❌ TBD
K3s/kubeadm Ubuntu 22.04+ Depends on kernel ✅ Up to 100/115

For the highest protection score (100/115), use GKE with Ubuntu node images:

gcloud container node-pools create cloudtaser-pool \
  --cluster=my-cluster \
  --image-type=UBUNTU_CONTAINERD \
  --machine-type=e2-standard-4

This is the only managed Kubernetes option that provides both memfd_secret (kernel-invisible memory) and CONFIG_BPF_KPROBE_OVERRIDE (synchronous syscall blocking).

Why scores differ

memfd_secret (15 points)

memfd_secret() creates memory regions invisible to the kernel itself. Requires CONFIG_SECRETMEM=y in the kernel.

  • Available on: GKE (COS + Ubuntu), AKS Azure Linux, kernels 5.14+ with CONFIG_SECRETMEM
  • Missing on: AKS Ubuntu 22.04 (Azure kernel disables CONFIG_SECRETMEM despite 5.14+ kernel)
  • Fallback: memfd_create + mlock — secrets still protected from swap/core dumps but visible to kernel modules

kprobe enforcement (15 points)

bpf_override_return() allows the eBPF agent to synchronously block syscalls (e.g., prevent /proc/environ reads). Requires CONFIG_BPF_KPROBE_OVERRIDE=y.

  • Available on: Ubuntu kernels (Canonical enables this by default)
  • Missing on: Cloud-native kernels (Azure Linux, GKE COS, Bottlerocket)
  • Fallback: Reactive kill — the agent detects the violation and terminates the process after the fact

Why cloud kernels disable kprobe override

Cloud providers disable CONFIG_BPF_KPROBE_OVERRIDE because:

  1. It was designed for kernel developer testing, not production use
  2. It can be used offensively (block kill signals, bypass LSM hooks)
  3. Minimal attack surface is a design goal for cloud-native kernels

See eBPF issue #95 for detailed research.

Remaining 15 points

The remaining points require infrastructure changes not available on standard managed Kubernetes:

  • cpu_mitigations (5 pts): Kernel boot parameters for CPU vulnerability mitigations. Not configurable on managed K8s.
  • confidential_vm (10 pts): AMD SEV-SNP hardware memory encryption. Requires specific VM types (GKE: --enable-confidential-nodes, AKS: DCasv5/ECasv5).