Proliant RHEL8 Reboot at install for AMD CPU

Problem

When trying to install RHEL 8.0 GA (May 2019) on a Proliant the machine will restart / shutdown after the installer (anaconda) has started, and just before the graphics mode starts. No useful logging is available or is lost in the restart. This occurs on both bare metal, and with a virtual machine. This has been observed on AMD Opteron (eg DL585 G5) and Turion (eg MicroServer G7).

Diagnostic

  • For KVM instances verify without a kickstart file;
  • For KVM instances verify with a simplified virt-inst configuration – For example:
 virt-install \
     --connect qemu:///system \
     --name testing-rhel8 \
     --ram 1024 \
     --vcpus 4 \
     --virt-type kvm \
     --cdrom rhel8-dvd.iso \
     --arch x86_64 \
     --os-variant rhel7 \
     --disk /var/lib/libvirt/images/RHEL8.img --force \
     --wait 0  \
     --hvm
  • For KVM instances increase /etc/libvirt/libvirtd.conf debug logging level to at least 3 if not higher to 1;
  • The machine either reboots (if bare metal) or shuts down (if virtualized);
  • Check /proc/cpuinfo for the CPU the host is using – note model and version;
  • Note during boot up if the following is displayed very early on in the boot process: “Warning: AMD Opteron – this hardware has not undergone testing by Red Hat and might not be certified. Please consult https://hardware.redhat.com for certified hardware.”

Solution

None currently.

Not all hardware configurations appear supported. So it may be worth noting the vintage of the processor and the hardware model.

Certified AMD CPU’s can be found here for those with access: https://access.redhat.com/support/policy/amd
Certified Platforms from Partners can be found here: https://hardware.redhat.com

Otherwise – if you don’t – Opteron 83xx series – up to RHEL6.x, and Turion not certified at all, and the Proliant Microservers are G8 and G10 on RHEL6,7 and 7 respectively.

Root Cause

This hardware has not been certified to work with RHEL8.0 – in the case of the Opteron and Turion, not since RHEL 6.x

As a lack of support for CPU side channel attacks become more common from the manufacturer, more and more platforms that once would have been run on to sweat the asset’s value – will need to obsoleted as they are no longer patched or supported by their manufacturer.

History

Two issues, one within a personal configuration had been observed. They both presented identically – and it took sleep to link the two. Proliant – more specifically AMD processors.
Checking certified (very different from supported) hardware for RHEL8 it was apparent that the certification (ie tested on) for older AMD has been dropped quite some time ago.
Curiously this appears to run without issue with current Ubuntu Desktop/Server.

https://bugzilla.redhat.com/show_bug.cgi?id=1711928

Leave a Reply

Your email address will not be published. Required fields are marked *

%d bloggers like this:
Skip to toolbar