Arch Linux

mark.clegg

Member

From: Lancashire

Registered: 2017-11-07

Posts: 2

iPXE netboot failing with HTTP 5xx Server Error

A few weeks ago I installed the netboot PXE images on my OpenWRT router, with the aim of being able to PXE boot the Arch installation on new devices instead of having to download a USB key every time. (Seemed a nice idea at the time).

OpenWRT uses dnsmasq, so I enabled tftp, dropped the images in the tftproot directory, and added the following to the dnsmasq.conf:

# PXE Configuration for ARCH NetBoot
dhcp-vendorclass=BIOS,PXEClient:Arch:00000
dhcp-vendorclass=UEFI32,PXEClient:Arch:00006
dhcp-vendorclass=UEFI,PXEClient:Arch:00007
dhcp-vendorclass=UEFI64,PXEClient:Arch:00009

dhcp-boot=net:BIOS,ipxe.31e77c156b6d.pxe
dhcp-boot=net:UEFI32,ipxe.176077d7fccb.efi
dhcp-boot=net:UEFI,ipxe.176077d7fccb.efi
dhcp-boot=net:UEFI64,ipxe.176077d7fccb.efi

This was working fine for both BIOS and UEFI PXE boot - until yesterday.

I now get an error after either of the images has loaded: (ipxe.31e77c156b6d.pxe or ipxe.176077d7fccb.efi, tried on a few systems):

iPXE 1.0.0+ (ea832) -- Open Source Network Boot Firmware -- http://ipxe.org
Features: DNS HTTP HTTPS iSCSI TFTP SRP AoE ELF MBOOT PXE bzImage Menu PXEXT
Waiting for link-up on net0.... ok
Configurting (net0 xx:xx:xx:xx:xx:xx)................... ok
https://ipxe.archlinux.org/releng/netboot/archlinux.ipxe... HTTP 5xx Server Error (http://ipxe.org/1d0c6539)
iPXE>

I'm pretty sure my dnsmasq config is ok as it was working previously, I haven't changed anyhting, and the iPXE image is being chained form the NIC's PXE firmware.  It seems to be a problem with downloading the arch iPXE script (which downloads fine from a browser (and the SSL cert looks valid).
Occasionally the error is preceded with "[OSCSP Let's Encrypt]"

If I "chain" the errored URL at the iPXE prompt - I just get the same error.

Any ideas? - Thanks in advance.

mark.clegg

Member

From: Lancashire

Registered: 2017-11-07

Posts: 2

Re: iPXE netboot failing with HTTP 5xx Server Error

Seems a bug has now been posted: https://bugs.archlinux.org/task/66407