Skip to content

Latest commit

 

History

History
119 lines (94 loc) · 3.68 KB

README.md

File metadata and controls

119 lines (94 loc) · 3.68 KB
Raw

KIKA netboot

Services used are: dnsmasq, nginx, syslinux (for pxelinux, menu, vesamenu, etc..), and nfs for ubuntu live. Supports BIOS or UEFI net boot.

Boot options:

  • Archlinux net install
  • Ubuntu live boot
  • Fedora live boot
  • Debian net install
  • CentOS net install
  • Memtest86+
  • SystemRescueCD
  • Darik's boot and nuke

Everything lives in /home/netboot:

  • /home/netboot/iso - distro .iso image files are downloaded here
  • /home/netboot/tftproot - the pxelinux config files
  • /home/netboot/nfs - .iso image files are loop mounted here

Dnsmasq (dhcp + tftp)

The dnsmasq config (in /etc/dnsmasq.d/pxe.conf) is:

enable-tftp
tftp-root=/home/netboot/tftproot

dhcp-match=x86PC, option:client-arch,0
dhcp-match=BC_EFI, option:client-arch,7
dhcp-match=BC_EFI, option:client-arch,9
dhcp-boot=tag:x86PC,/syslinux/bios/lpxelinux.0
dhcp-boot=tag:BC_EFI,/syslinux/efi64/syslinux.efi
dhcp-option-force=tag:x86PC,209,::/pxelinux.cfg/default
dhcp-option-force=tag:BC_EFI,209,::/pxelinux.cfg/default

This config supports both BIOS and UEFI machines, giving them the coresponding pxelinux bootloader.

To start and enable dnsmasq run:

sudo systemctl enable --now dnsmasq.service

PXELinux

In our case /home/netboot/tftproot/syslinux is a copy of /usr/lib/syslinux/ from Archlinux - cause it was much newer than the syslinux from Debian jessie. But one can also use the syslinux official binary releasess or a symlink to /usr/lib/syslinux/ - mind the different structure though.

On BIOS systems we use bios/lpxelinux.0 - has the most of features, on UEFI we use efi64/syslinux.efi.

Check the other files in this directory, starting from default to see the syslinux menu configuration used.

NFS

To setup a read-only export of the whole /home/netboot/nfs directory, put the following line in /etc/exports:

/home/netboot/nfs *(ro,async,no_subtree_check,no_root_squash,insecure,crossmnt)

and then just enable and start the nfs-server service:

sudo systemctl enable --now nfs-server.service

nginx

need to expose these two with http:

  location /tftproot {
      alias /home/netboot/tftproot/;
      autoindex off;
      allow all;
  }
  location /nfs {
      alias /home/netboot/nfs/;
      autoindex off;
      allow all;
  }

Don't forget to enable and start nginx too:

sudo systemctl enable --now nginx.service

Distros

For ubuntu and other casper based distros (Mint etc), Fedora and Arch you only need the .iso image file loop mounted. For example (the fstab entry):

/home/netboot/iso/ubuntu-16.04.2-desktop-amd64.iso /home/netboot/nfs/ubuntu-16.04.2-desktop-amd64 auto loop,nofail,ro 0 0
/home/netboot/iso/Fedora-Workstation-Live-x86_64-26-1.5.iso /home/netboot/nfs/fedora auto loop,nofail,ro 0 0

If running a lot of distros, make sure to enlarge the maximum allowed loop devices:

#/etc/modprobe.d/local.conf
options loop max_loop=32

Booting explained for dummies

How booting works, step by step:

  1. Computer is turned on
  2. BIOS is loaded and run
  3. We select network boot
  4. PXE part of BIOS is loaded and run
  5. PXE BIOS requests a DHCP address
  6. dnsmasq replies with an address and boot server and filename (pxelinux.0)
  7. PXE BIOS uses tftp to download pxelinux.0 and runs it
  8. pxelinux (inherits the IP setup from the BIOS) it uses tftp to download a config file (typically pxelinux.cfg/default)
  9. the config file is a typical syslinux configuration specifying a kernel and initial ramfs (initrd) an archive unpacked in a ram filesystem
  10. pxelinux.0 tftp downloads the choosen kernel and initrd in RAM
  11. the kernel runs, initializes itself, and finds the ram filesystem
  12. /init in the ramfs is run as process 1 (ussually a script)