What were you trying to do that didn't work?

We attempted to configure RHEL (RPM-based, GRUB2-based systems) to automatically fall back to a previously known-good kernel whenever a newly installed kernel fails to boot or panics early in the boot process. The goal was to implement Greenboot-like functionality on standard RHEL installations, where the system would only "promote" a new kernel after passing health checks.

This capability is not available. GRUB2 does not support health-based kernel promotion or rollback, Greenboot does not work on RPM-based systems, and systemd-bless-boot requires systemd-boot, which RHEL does not use.

What is the impact of this issue to you?

Kernel updates become risky on systems that rely on custom hardware drivers. If a new kernel is incompatible with a vendor driver, the host panics during modprobe and becomes unbootable. Since there is no automatic rollback, the system remains down until manual intervention.
This increases downtime, reduces reliability, and makes unattended updates unsafe, especially in production or appliance-style deployments.

Please provide the package NVR for which the bug is seen:

This is not a bug in a specific package. It is a feature gap across the kernel update and boot infrastructure on RHEL, involving:

• kernel (various RHEL 10.0 builds)

• grub2

• dracut

• greenboot (incompatible in package mode)

Expected results

The system should be able to:

• Boot the newly installed kernel in a "trial" state

• Detect that the system did not reach a healthy boot target

• Automatically fall back to the last known-good kernel

• Avoid leaving the system in a non-bootable or panic-loop state

This is comparable to Greenboot's behavior on image-based systems.