PXE boot fails due to network issues

General support questions
Post Reply
mt190
Posts: 1
Joined: 2013/11/26 18:09:07

PXE boot fails due to network issues

Post by mt190 » 2019/02/19 21:45:04

Hi Guys

I'm trying to setup a PXE server which will install CentOS 7.
I have a DHCP on windows server 2008, which points at a PXE server.
The PXE is setup to do UEFI installations, and with secure boot enabled, so it sends down shimx64.efi, which then loads grub2.

I can get a server connected to that network load up dracut, and start setting everything up, but as soon as it tries to fetch my kickstart file, from a local webserver, or the installation files that also reside on the same server, I run into all sorts of network issues, which I cannot figure out.

The error I get is network in unreachable/can't resolve hostname, sounds simple enough, something went wrong and no IP was obtained from the DHCP, but when I check the IP in the dracut rescue shell which I eventually get dumped to, there is one there, I can even ping the host where the files are and I get a response, yet for some reason, when I try to automate the install, it never works.

I have a feeling that this is a timing issue, and either the NIC is slow to come up, or the router does not provide an IP address fast enough, so I started looking for some type of nicdelay or sleep options, but I do not see those anymore since everything was passed to dracut.
I don't know of anyway to verify this, I never debugged dracut to see what is happening in there.

I'm hoping maybe someone has a clue what to do here, because I've spent several days on this now and I can not get it to work.

here is my setup:

Cent OS 7.6 1810 x86_64 on the PXE as well as what I'm trying to install.
The server is lenovo SR500

Grub.cfg:

Code: Select all

set default="0"
function load_video {
         insmod efi_gop
         insmod efi_uga
         insmod video_bochs
         insmod video_cirrus
         insmod all_video
}
load_video
set gfxpayload=keep
insmod net
insmod efinet
insmod tftp
insmod gzio
insmod part_gpt
insmod ext2
set timeout=60
search --no-floppy --set=root -l 'CentOS 7.6 x86_64'
menuentry 'RHEL 7' --class fedora --class gnu-linux --class gnu --class os {
  linuxefi images/centos/7.6.1810/vmlinuz ipv6.disable ip=dhcp rd.neednet=1 inst.ks=http://10.135.199.88/kickstart/ansible/centos7_minimal.cfg inst.repo=http://10.135.199.88/centos/7.6.1810/os/x86_64
  initrdefi images/centos/7.6.1810/initrd.img
}
I can do a legacy PXE installation, which has similar issues, but if I specify an IP for the kickstart host, it works, not in UEFI, and I'm not sure why.

Appreciate any help.

mt190

Post Reply