Goal

Provide a containerized test suite verifying that a given cluster is real-time-worthy.

User Stories

• As a cluster administrator,
  I want run RT VMs,
  so that I can run workload that requires predictable/low latency.
• As a cluster administrator,
  I want to run a checkup,
  in order to verify that my real-time configured nodes are ready to serve real-time virtualized workloads.
• As a namespace administrator,
  I want to verify that the cluster is configured correctly for real-time virtualized workloads.
• As a guest OS vendor (e.g RHEL Virt QE),
  I would like to prepare a guest image, place it on a PVC and test it with the checkup.
• As a cluster administrator, I want to ensure minimal performance degradation in RT VMs under high network load, to maintain consistent real-time capabilities.
• As a system integrator, I want to validate that RT VMs can handle rapid context switches without significant latency spikes, ensuring reliability for time-sensitive applications.

Non-Requirements

• We don't have to automate all the possible test cases in order to release the support. We should focus on those that cover the most ground and requirements of existing customers. Our documentation then should explain what has been tested.
• We don't have to run virt-launcher's housekeeping on reserved CPUs to release the support. While it would be nice to have that optimization implemented, we can also release with KubeVirt always wasting one isolated CPU per VM.
• We don't need to set runtime class per-VM. Setting it globally is good enough.

Notes

• KVM RT introduction
• OCP RT documenation
• RT checkup design
• RT on CNV test plan