pve 9.2 fails to install

F

fiskarn

New Member
Jul 7, 2026
2
0
1
Hi everyone,

I'm hitting a persistent installation failure when trying to install Proxmox VE 9.2-1 on an older Intel platform. The graphical installer reaches 100% data extraction but consistently crashes during the bootloader/initramfs generation phase with the error: "bootloader setup errors: - unable to install initramfs".

The exact same hardware installs Proxmox VE 8.4 flawlessly without any issues, which points towards a kernel/driver regression in the newer 9.x kernel regarding the onboard Intel Z97 SATA controller.

### Hardware Environment:
* **Motherboard:** ASUS Z97-AR (Intel Z97 Express Chipset)
* **BIOS Settings tested:** AHCI mode, Aggressive LPM Disabled, Dynamic Storage Accelerator Disabled, SATA Express disabled (forced M.2 mode).
* **Storage tested:** Brand new Samsung SSD & older Intel SSD (Tested standalone on isolated Intel SATA ports 1 and 2, with different SATA and power cables).
* **Installation Media:** Brand new ISO downloads (SHA256 verified), flashed via terminal using `dd bs=4M oflag=sync`.

### Kernel Error Logs during PVE 9.2 Failure:
Behind the installer GUI, the Linux kernel throws persistent NCQ/FPDMA queue errors and completely drops the link to the drive under heavy write loads at the end of the installation sequence:

```text
[ 197.534276] ata5.00: exception Emask 0x60 SAct 0x3000000 SErr 0x40800 action 0x6 frozen
[ 197.534402] ata5.00: irq_stat 0x20000000, host bus error
[ 197.534523] ata5.00: SError: { HostInt CommWake }
[ 197.534642] ata5.00: failed command: WRITE FPDMA QUEUED
[ 197.534759] ata5.00: cmd 61/08:c0:a8:df:27/11:00:03:00/40 tag 24 ncq dma 2232320 ou
[ 197.534759] res 40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x60 (host bus error)
[ 197.534995] ata5.00: status: { DRDY }
[...]
[ 197.866264] Buffer I/O error on dev dm-1, logical block 6857386, lost sync page write
[ 197.866271] EXT4-fs (dm-1): failed to convert unwritten extents to written extents -- potential data loss! (inode 921131, error -5)
[ 197.866406] Aborting journal on device dm-1-8.
[ 197.867301] EXT4-fs error (device dm-1): ext4_journal_check_start:86: comm kworker/u32:5: Detected aborted journal
[ 197.867317] EXT4-fs error (device dm-1): ext4_journal_check_start:86: comm dpkg: Detected aborted journal

```
When switching ports to ata2 (SATA6G_2), the exact same behavior occurs:
```text
ata2.00: exception Emask 0x60 SAct 0x40000003 SErr 0x800 action 0x6 frozen
ata2.00: irq_stat 0x20000000, host bus error
ata2.00: SError: { HostInt }
ata2.00: failed command: WRITE FPDMA QUEUED
ata2.00: cmd 61/f8:e0:90:ec:80/01:00:09:00/40 tag 0 ncq dma 4194304 ou
res 40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x60 (host bus error)
ata2.00: status: { DRDY }
I/O error, dev sda, sector 159444112 op 0x1:(WRITE) flags 0x0 phys_seg 7 prio class 0

```
### Steps taken that did NOT resolve the issue on PVE 9.2:
1. Swapping out the SSD entirely (from old Intel to brand new Samsung).
2. Changing both SATA data cables and power delivery lines from the PSU.
3. Shifting drives away from the shared SATA Express controller ports over to native Intel ports (SATA6G_1 and _2).
4. Disabling Aggressive LPM and Dynamic Storage Accelerator features in the UEFI.
5. Attempting both the regular Graphical Installer and the Terminal UI Installer.
Since Proxmox VE 8.4 (using an older kernel) installs flawlessly on the exact same target drives and layout, it looks like the newer kernel driver/libata in PVE 9.x is mishandling NCQ/commands on this specific Z97 chipset under sustained I/O load.
Is there a known kernel parameter workaround (libata.force=...) I can pass to the PVE 9.2 boot arguments to bypass this host bus error, or has anyone else experienced this regression on 9-series Intel chipsets?
Thanks in advance!
 
Try to append this options in the grub:

Code: 
pcie_aspm=off intel_iommu=off intel_idle.max_cstate=0 acpi_power_meter=off nomodeset

This options has save me for a lot of issue.

pcie_aspm=off
  • What it does: Disables Active State Power Management for PCI Express (ASPM).
  • What it is for: Prevents PCIe links from entering low-power states.
  • Common use: Fixes system freezes, Wi-Fi/Ethernet drops, or NVMe SSD instability caused by poor power state transitions.

intel_iommu=off
  • What it does: Disables Intel Input-Output Memory Management Unit (IOMMU).
  • What it is for: Turns off the hardware mapping that isolates direct memory access (DMA) for peripherals.
  • Common use: Resolves compatibility bugs with older or problematic motherboards and GPUs. Note: Disabling this prevents hardware passthrough (like assigning a physical GPU directly to a virtual machine).

intel_idle.max_cstate=0
  • What it does: Completely disables deep power-saving states (C-states) on Intel processors.
  • What it is for: Forces the CPU to constantly run in the C0 state (maximum performance, no sleep modes).
  • Common use: Eliminates micro-stuttering, audio latency, and random system freezes caused by faulty CPU voltage shifts. It increases power consumption.

acpi_power_meter=off
  • What it does: Disables the Linux ACPI power meter driver.
  • What it is for: Stops the system from trying to read power consumption sensors from the motherboard or power supply.
  • Common use: Prevents a flood of error messages in the terminal logs (such as ACPI Error: AE_NOT_FOUND) on servers or desktops with buggy BIOS implementations.

nomodeset
  • What it does: Prevents the Linux kernel from loading video drivers before the graphical interface starts.
  • What it is for: Forces the system to use basic video resolution (usually VESA) during the boot process.
  • Common use: Used to bypass black screen issues caused by faulty graphics drivers (NVIDIA or AMD), allowing you to boot into the system and install the correct drivers.

 
Last edited:
fiskarn said:
Hi everyone,

I'm hitting a persistent installation failure when trying to install Proxmox VE 9.2-1 on an older Intel platform. The graphical installer reaches 100% data extraction but consistently crashes during the bootloader/initramfs generation phase with the error: "bootloader setup errors: - unable to install initramfs".

The exact same hardware installs Proxmox VE 8.4 flawlessly without any issues, which points towards a kernel/driver regression in the newer 9.x kernel regarding the onboard Intel Z97 SATA controller.

### Hardware Environment:
* **Motherboard:** ASUS Z97-AR (Intel Z97 Express Chipset)
* **BIOS Settings tested:** AHCI mode, Aggressive LPM Disabled, Dynamic Storage Accelerator Disabled, SATA Express disabled (forced M.2 mode).
* **Storage tested:** Brand new Samsung SSD & older Intel SSD (Tested standalone on isolated Intel SATA ports 1 and 2, with different SATA and power cables).
* **Installation Media:** Brand new ISO downloads (SHA256 verified), flashed via terminal using `dd bs=4M oflag=sync`.

### Kernel Error Logs during PVE 9.2 Failure:
Behind the installer GUI, the Linux kernel throws persistent NCQ/FPDMA queue errors and completely drops the link to the drive under heavy write loads at the end of the installation sequence:

```text
[ 197.534276] ata5.00: exception Emask 0x60 SAct 0x3000000 SErr 0x40800 action 0x6 frozen
[ 197.534402] ata5.00: irq_stat 0x20000000, host bus error
[ 197.534523] ata5.00: SError: { HostInt CommWake }
[ 197.534642] ata5.00: failed command: WRITE FPDMA QUEUED
[ 197.534759] ata5.00: cmd 61/08:c0:a8:df:27/11:00:03:00/40 tag 24 ncq dma 2232320 ou
[ 197.534759] res 40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x60 (host bus error)
[ 197.534995] ata5.00: status: { DRDY }
[...]
[ 197.866264] Buffer I/O error on dev dm-1, logical block 6857386, lost sync page write
[ 197.866271] EXT4-fs (dm-1): failed to convert unwritten extents to written extents -- potential data loss! (inode 921131, error -5)
[ 197.866406] Aborting journal on device dm-1-8.
[ 197.867301] EXT4-fs error (device dm-1): ext4_journal_check_start:86: comm kworker/u32:5: Detected aborted journal
[ 197.867317] EXT4-fs error (device dm-1): ext4_journal_check_start:86: comm dpkg: Detected aborted journal

```
When switching ports to ata2 (SATA6G_2), the exact same behavior occurs:
```text
ata2.00: exception Emask 0x60 SAct 0x40000003 SErr 0x800 action 0x6 frozen
ata2.00: irq_stat 0x20000000, host bus error
ata2.00: SError: { HostInt }
ata2.00: failed command: WRITE FPDMA QUEUED
ata2.00: cmd 61/f8:e0:90:ec:80/01:00:09:00/40 tag 0 ncq dma 4194304 ou
res 40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x60 (host bus error)
ata2.00: status: { DRDY }
I/O error, dev sda, sector 159444112 op 0x1:(WRITE) flags 0x0 phys_seg 7 prio class 0

```
### Steps taken that did NOT resolve the issue on PVE 9.2:
1. Swapping out the SSD entirely (from old Intel to brand new Samsung).
2. Changing both SATA data cables and power delivery lines from the PSU.
3. Shifting drives away from the shared SATA Express controller ports over to native Intel ports (SATA6G_1 and _2).
4. Disabling Aggressive LPM and Dynamic Storage Accelerator features in the UEFI.
5. Attempting both the regular Graphical Installer and the Terminal UI Installer.
Since Proxmox VE 8.4 (using an older kernel) installs flawlessly on the exact same target drives and layout, it looks like the newer kernel driver/libata in PVE 9.x is mishandling NCQ/commands on this specific Z97 chipset under sustained I/O load.
Is there a known kernel parameter workaround (libata.force=...) I can pass to the PVE 9.2 boot arguments to bypass this host bus error, or has anyone else experienced this regression on 9-series Intel chipsets?
Thanks in advance!
Click to expand...
Have you tried this?

Fix: Kernel Boot Parameters
You can bypass this issue by forcing the kernel to throttle or disable NCQ for that specific drive or controller during installation.

Step 1: Apply the installer boot argument
  1. Boot up your Proxmox VE 9.2 installer USB.
  2. At the GRUB boot menu, highlight your preferred installer option (Graphical or Console) and press e to edit the boot arguments.
  3. Move your cursor to the line starting with linux /boot/vmlinuz....
  4. Append one of the following arguments to the very end of that line (separated by a space):
Option A (Recommended - Disable NCQ entirely):
libata.force=noncq

Option B (Alternative - Force a lower NCQ queue depth if Option A fails):
libata.force=32:ncqdepth=1

Option C (If the issue persists, disable both NCQ and LPM completely):
libata.force=noncq,nolpm


Press F10 or Ctrl+X to boot the installer with these modified flags.

Builder of Proximo - The Proxmox MCP you can hand the keys — VE + Backup Server + Mail Gateway + Datacenter Manager on one audited trust core (plan · prove · undo · diagnose). MCP + A2A
 
You must log in or register to reply here.
Top Bottom
Back