I'm running into a reproducible issue trying to install Proxmox VE 9.x on an
HPE ProLiant DL380 Gen12.
The server was previously running
Proxmox VE 8.4 without any issues. The system uses a
ZFS RAID1 root pool on two ~1.8 TB drives, with separate storage for the VMs.
I initially attempted an upgrade from PVE 8.4 to PVE 9.x. The package upgrade itself appeared to complete and the installed pve-manager version was eventually confirmed as
9.2.10, but afterwards the system had serious boot problems.
The ZFS root pool itself was healthy and could be imported manually. Both members of the rpool mirror were ONLINE, so this did not appear to be a disk or ZFS data-integrity issue.
I spent quite some time trying to repair the EFI/boot environment, including proxmox-boot-tool, checking/reinitializing the EFI System Partitions, removing obsolete boot entries, etc., but I could not get the PVE 9 installation to boot reliably.
At that point I decided not to troubleshoot the upgraded installation any further and instead tried a
completely clean installation of Proxmox VE 9.2.
The clean installation was configured as:
- HPE ProLiant DL380 Gen12
- UEFI boot
- ZFS RAID1
- Two ~1.8 TB disks selected exclusively for the new rpool
- Existing VM/data disks were not selected by the installer
The clean PVE 9.2 installer consistently reached:
99% – make system bootable
and then remained there indefinitely. I left it for a considerable amount of time, but it never completed.
This was particularly interesting because I subsequently booted the
Proxmox VE 8.4 installer on exactly the same machine, selected the same two disks and the same ZFS RAID1 configuration, and the installation completed
very quickly and without a single issue.
The newly installed PVE 8.4 system boots normally and is currently operational again.
For additional context, after reinstalling 8.4:
- The existing large ZFS VM pool could be imported successfully.
- zpool status reports all members ONLINE.
- The existing VM configurations became available again after rejoining the node to the cluster.
- The existing VMs can be started normally.
- The Mellanox 25 GbE interface also works normally under PVE 8.4.
So at this point I have reproduced two separate PVE 9.x-related problems on this machine:
- Upgrade from PVE 8.4 → 9.x: upgrade completes, but the resulting system has major EFI/boot issues.
- Clean PVE 9.2 installation: installer hangs indefinitely at 99% make system bootable.
Reinstalling
PVE 8.4 with the same hardware, same two system disks, same UEFI configuration and same ZFS RAID1 layout works perfectly.
This makes me suspect there may be an incompatibility/regression in the PVE 9.x installer/boot tooling on this particular HPE Gen12 platform rather than a problem with the ZFS pool or physical disks.
Has anyone seen similar behaviour with
HPE ProLiant Gen12 hardware and PVE 9.x, particularly the installer hanging at make system bootable?
I'm happy to collect logs or hardware/firmware information from the currently working PVE 8.4 installation if that would help identify the cause. I would also be interested to know which installer/debug logs would be most useful to capture during another PVE 9.2 installation attempt.
Thanks!
