PXE Not booting, what is the default bios filename to boot?

[mannp]

Thanks for sharing your config :)

I wondered what the default boot image menu is for the pxe server you are using, as I cannot get the machines to see the default menu / boot image for pxe.

Thanks

[khuedoan]

I'm not sure on the question, I use the second option in the docs, can you provide some more logs and/or screenshots please?

For example:

• Command used
• PXE server logs
• BIOS config screenshot(s)
• Boot menu screenshot(s)

[mannp]

For my working netboot.xyz server I have to set IP of the pxe server, EFI 32bit filename, EFI 64bit filename and default bios filename, on opnsense, so that the machines on that vlan know where the pxe server is.

I guess the question is, how do the node machines know where the temp PXE server is on my desktop?

At the moment they find the PXE server, but they error that there is no image / menu to boot and repeat.

./scripts/pxe-logs gives me the error 'unknown flag: --project-directory', so I guess I am missing something obvious :)

[khuedoan]

The DHCP config will point the nodes to the temp PXE server (which is the same PC you're running Ansible on):

homelab/metal/roles/pxe_server/templates/dhcpd.conf.j2

Lines 12 to 14 in 8b31c9e

	class "pxeclients" {
	match if substring (option vendor-class-identifier, 0, 9) = "PXEClient";
	next-server {{ ansible_default_ipv4.address }};

You can check the rendered DHCP config in this path:

homelab/metal/roles/pxe_server/tasks/main.yml

Line 16 in 8b31c9e

dest: "{{ role_path }}/files/data/pxe-config/dhcpd.conf"

[khuedoan]

./scripts/pxe-logs gives me the error 'unknown flag: --project-directory', so I guess I am missing something obvious :)

Oh that's a bug due to this commit: 40523e6

The Ansible role use Docker Compose v1 (see ansible-collections/community.docker#216), but the pxe-logs script use Docker Compose v2 and I forgot about that.

[khuedoan]

Done, please pull latest master and try again

[mannp]

I guess the critical missing info, is that I am trying to use four vm's on proxmox for the 4 physical machines and maybe confusing the issue here.

I will have a read on ansible pxe boot :)

[khuedoan]

Ah if you're using VMs just use a Rocky Linux 8 image and ignore the boot.yml playbook and go directly to the cluster.yml playbook, you don't need PXE boot at all:

homelab/metal/Makefile

Lines 15 to 18 in 7e21741

	cluster:
	ansible-playbook \
	--inventory inventories/${env}.yml \
	cluster.yml

[khuedoan]

Some configuration to note on the base image:

homelab/metal/roles/pxe_server/templates/kickstart.ks.j2

Lines 37 to 49 in 7e21741

	# Disable SELinux
	selinux --disabled
	# Disable firewall
	firewall --disabled
	%packages
	@^minimal-environment
	iscsi-initiator-utils
	%end
	# Enable iSCSI for Kubernetes storage
	services --enable=iscsid

[mannp]

I was hoping your setup could create and configure the nodes for me via pxe with minimal config of the vm.

What would the local pxe server be presenting to the nodes eg tftp://192.168.1.94/

[khuedoan]

My rendered DHCP config looks like this:

subnet 192.168.1.0 netmask 255.255.255.0 {
    option routers 192.168.1.1;
    range 192.168.1.2 192.168.1.254;

    class "pxeclients" {
        match if substring (option vendor-class-identifier, 0, 9) = "PXEClient";
        next-server 192.168.1.9;

        if option architecture-type = 00:07 {
            filename "grubx64.efi";
        }
    }
}

The packages will be downloaded via the HTTP server (http://192.168.1.9)

[mannp]

It looks like this setup is creating a dhcp server itself to configure the physical machines, is that correct?

I am getting an 'Access denied' when my vm's try to access the pxe server, but they are looking at my opnsense dhcp server for the details.