Kubernetes launches a working group to define conformance for workloads
The Kubernetes project has set up a working group to write a shared standard for how workloads behave at runtime, along with tests and conformance tooling. Cluster operators and maintainers of applications deployed on Kubernetes are directly affected and are being asked for input.
What to do now
On 27 October 2026 the Kubernetes community announced a new Workload Conformance Working Group. Its job is to produce a common standard describing how applications should behave when running on a Kubernetes cluster.
Existing programmes focus on the platform rather than what runs on it. Certified Kubernetes checks the cluster itself, and Kubernetes AI Conformance checks whether a cluster can run AI and machine learning workloads. There is no equivalent for the workloads, so operators still review each application manually before routine maintenance or an upgrade.
The project points to problems most operations teams will recognise: an application that cannot handle disruption and only reveals it during an ordinary node drain, an upgrade that breaks a service still relying on a removed API version, and applications that request privileges they do not need or reserve resources they never consume. Kubernetes documentation already covers much of this, yet every operator repeats the same assessment and application maintainers field the same questions from each environment they support.
The group's charter lists three deliverables: a specification of expected workload behaviour for each Kubernetes minor version, a method for testing each requirement against a running workload, and upstream conformance tooling that implements those tests.
The planned scope covers security, resilience during disruptions, resource usage, networking, storage, API usage and observability. The group has not yet agreed on specific requirements or on how to test them, but one rule is fixed: every requirement must have an objective test. The source gives no timeline for when the specification or tooling will be available.
The project is explicitly asking both the people who build workloads and the people who run clusters to take part, so the standard reflects constraints on both sides.
What to do now
- Read the announcement and charter of the Workload Conformance Working Group on the Kubernetes blog and decide whether your team wants to contribute to the requirements.
- Until a specification exists, check as general good practice that your applications survive a node drain, for example by setting appropriate PodDisruptionBudgets.
- Before your next Kubernetes upgrade, inventory manifests that still depend on removed or deprecated API versions.
- Review the privileges and resource reservations requested by your workloads and remove those that are not needed.