How to Install CentOS 7 on Intel NUC DN2820FYKH

I’ve confirmed this problem was fixed with CentOS 7.1 + BIOS Update 0048.

I tried to install CentOS 7 on Intel NUC DN2820FYKH which I bought to introduce the new OS, but the machine was hung up just after boot with “[0.000000] tsc : Fast TSC calibration failed” error. That was reported in CentOS Forums: CentOS 7.0 on Intel NUC, but no solution there. I looked for one using a holiday and figured it out.

You might think this is a TSC (TimeStamp Counter) problem due to the above error message. But removing “quiet”, you see the machine is hung up after “Loaded X.509 cert ‘CentOS Linux Driver update signing key”. I guess it seems to be a kernel problem because “Loaded X.509 cert ‘CentOS Linux kernel signing key” follows this message.

Anyway, the below is a step-by-step for my machine. Take care various traps.

  1. Boot from a USB flash drive instead of an external DVD drive. And you shouldn’t connect a mouse and a LAN cable. The machine doesn’t often boot up when mouse or LAN cable is connected (Unknown reproducible condition)
  2. Create the USB flash drive with the below command. Do not use Fedora LiveUSB Creator.  A USB flash drive created by Fedora LiveUSB Creator can start an installer (Anaconda), but you can’t choose any packages by “This media is not good to install from” error when you verify “Auto-detected installation media” from “Installation Source”. And the network doesn’t work either. That means you can’t move forward in the installation process. (This problem is reported in CentOS Forums: CentOS7, stuck in installation, critical errors)
  3. Update BIOS to the latest version, 0038
  4. Uncheck UEFI Boot at BIOS. The other default configs are OK
    UEFI Boot checkbox at Intel Visual BIOS
  5. Boot up the machine with the USB flash drive by F10 key
  6. Choose “Install CentOS 7” and then press Tab key. Delete “quiet” at the end. The installer doesn’t start if the “quiet” is available
  7. “[0.000000] tsc : Fast TSC calibration failed” still appears, but I believe the disk access indicator is blinking. Wait 10 seconds, and then the installer would start up.
  8. You can connect a LAN cable now and it would be recognized. But you have to focus by Tab key because a mouse isn’t recognized. Note that I only tried Minimum Install. I’m not sure if you install GNOME
  9. After the installation, reboot the machine
  10. Choose “CentOS Linux, with Linux 3.10.0-123.el7.x86_64” and press E key. Scroll by arrow keys (not vi) and delete “quiet” from “linux16 /vmlinuz-3.10.0-123.el7.x86_4 …” row. The OS doesn’t start if the “quiet” is available
  11. If the machine is still hung up before Login prompt, stop it by power button, and then start it again. I’m not sure but Login prompt might appear next time (Unknown reproducible condition)
  12. Delete “quiet” from /etc/default/grub
  13. Apply it to Grub 2 config file by the below command

That’s all. Enjoy the new OS. Note that the latest kernel version, kernel-3.10.0-123.6.3.el7.x86_64 causes these problem, too.

4 thoughts on “How to Install CentOS 7 on Intel NUC DN2820FYKH”

  1. For the error: “[0.000000] tsc : Fast TSC calibration failed”
    Try the Bootparameter in Grub:
    clocksource=acpi_pm

  2. I have downloaded the latest Centos 7 ISO CentOS-7-x86_64-DVD-1503-01.iso. The target is “Intel BOXNUC5CPYH Celeron NUC Cel-N3050 (1.6GHz) Mini PC Barebone Desktop Kit, 2.5in SATA, HDMI, VGA, GB LAN, 4xUSB3.0, Audio out, WiFi-AC+BT, VESA”

    I tried installing from a DVD on a USB DVD drive but it was not successful. I then tried it from a USB flash drive which did work. I did not upgrade the BIOS.

    With the USB drive I got the “[0.000000] tsc : Fast TSC calibration failed” message but it booted anyway – after a delay.

    The install screen did warn the hardware had not been tested for this install.

    I used a USB keyboard and mouse combo and the mouse worked fine at all stages. At no stage did I edit the boot parameters to ‘quiet’.

    The install went fine and I was able to reboot and it worked with no errors in the boot process asides from the final GUI (see later)

    I updated to the latest package versions (November 2015) and rebooted again.

    The main part of the machine booted fine and I can connect via LAN using ssh.

    The problem is the graphics display which says “Oh no! Something has gone wrong …” I assume it’s a gnome or x-windows message.

    As I said, the main machine is working fine. It’s just a graphics GUI problem now. For my purpose of a small remotely managed server it’s not an issue but I’d prefer to nail the problem now in case I need graphics later.

    1. Thank you for your report about the latest NUC.

      > Oh no! Something has gone wrong …

      Awful message… No info…

コメントを残す