The OPA Gatekeeper team has just released version 4.0, and the most significant breaking change is the immediate cessation of support for Kubernetes versions older than 1.26. This is a substantial jump from Gatekeeper 3.13, which supported clusters back to version 1.21. For organizations running legacy clusters or those with slower upgrade cycles, this will force a critical decision path.
The primary driver for this change is the migration of Gatekeeper's webhook from the deprecated `admissionregistration.k8s.io/v1beta1` API to `admissionregistration.k8s.io/v1`. The v1 API became generally available in Kubernetes 1.22, but remained in beta and compatible for several versions prior. By anchoring support at 1.26, the Gatekeeper maintainers are effectively mandating that users operate clusters where the v1 API has been stable and default for a significant period. This reduces their test matrix and allows them to leverage newer features and security constructs of the API.
If you attempt to install Gatekeeper 4.0 on a pre-1.26 cluster, the installation will fail because the `ValidatingWebhookConfiguration` and `MutatingWebhookConfiguration` resources will reference an API version your API server does not recognize. You'll see an error similar to:
```yaml
# Example of the breaking manifest snippet from the new Helm chart:
apiVersion: admissionregistration.k8s.io/v1
kind: ValidatingWebhookConfiguration
metadata:
name: gatekeeper-validating-webhook-configuration
...
```
For those unable to upgrade their Kubernetes control plane in the near term, you have two main options:
* **Pin to Gatekeeper 3.13.x:** This will be your last feature-updated version. You will only receive critical security patches, if any, and must plan an eventual cluster upgrade.
* **Implement a Policy Shimming Layer:** If Gatekeeper is non-negotiable for policy enforcement, you could run its pods on a modern, manageable "policy hub" cluster and use a syncing mechanism (like `gator` CLI or a custom controller) to evaluate resources from your older, workload clusters. This adds significant complexity.
From a security and compliance standpoint, this move highlights the operational necessity of maintaining a reasonably current Kubernetes version. Running clusters on versions older than 1.26 (released in December 2022) poses inherent risks, as you fall outside the window of upstream security patches and critical bug fixes. While Gatekeeper's change is a tooling constraint, it aligns with broader zero-trust principles that assume a compromised or outdated infrastructure component.
My recommendation is to treat this as a catalyst for modernization. If you are blocked by a downstream platform dependency (e.g., a managed service from a cloud provider that lags on Kubernetes versions), this is a strong data point to escalate with your vendor. For in-house clusters, begin planning the upgrade to at least Kubernetes 1.26 immediately, and consider targeting 1.28 or 1.29 to build a buffer against future tooling shifts.
This is a more significant break than it might appear on the surface. While you're correct about the API version being the stated reason, the jump to 1.26 as the floor likely ties directly to the removal of the `PodSecurityPolicy` (PSP) API in 1.25. Many Gatekeeper policies were used as PSP replacements. By requiring 1.26+, the maintainers can assume that disruptive, cluster-wide PSP removal is already a resolved issue for the user base, eliminating a major source of legacy policy migration complexity from their support burden.
The forced upgrade path will create a hard bifurcation. Organizations with mature platform teams on recent K8s versions will benefit from a cleaner codebase and potentially new features. Those on legacy versions, often in heavily regulated or slow-moving sectors, now face a real risk of forking Gatekeeper 3.13 or seeking alternative policy engines like Kyverno, which still lists support for 1.23+. This could fragment the community around policy-as-code.
No free lunch in cloud.