Description

The requirement is from CNV-59488

The primary goal of this RHEL Epic is to ensure robust and consistent persistence of vTPM device state within qemu-kvm. This includes developing and implementing mechanisms for consistently flushing, saving, and restoring vTPM state, particularly during VM snapshot operations that do not involve a full savevm or migration to file.
Currently, VM snapshots taken by platforms like OpenShift Virtualization may save only the disk state and associated files, such as the vTPM state file. However, the on-disk vTPM state might be outdated or inconsistent during these operations. This Epic aims to address this by providing a reliable way to save a consistent snapshot of the vTPM state. This foundational work in RHEL is crucial for enabling dependent platforms to achieve scalable and reliable VMState persistence
 

User Stories

As a kubevirt developer, I need the underlying RHEL hypervisor to ensure vTPM state is consistently flushed to disk during "disk-only" snapshots, so that I can reliably capture and restore a complete and accurate VM state without compromising data integrity or requiring excessive resources. 

As a kubevirt developer I want to enable VMState by default for all VMs so that I have storage available to implement changed block tracking and migration state synchronization as first class features.

As a RHEL virtualization user, I expect my "disk-only" snapshots to consistently capture the true state of my virtual TPM devices, ensuring that security features are always functional and data protected as expected upon restoration.

What SSTs and Layered Product teams should review this?

OCPV