PXE boot from bridged adaptor

[jeff.sadowski]

I have Fedora27 as my host OS.
Running virtualbox 5.2.6
I bridged my virtualbox machine to my outside interface and my local dhcp server has pxe boot available.
There is nothing currently installed on my vm I wanted to try and use a pxe boot installer I had used to install the machine it is running on.

I disabled the firewall. I disabled selinux.
Not sure what the issue is or what other information would be helpful to troubleshoot this.

The guest never gets an ip address from the dhcp server.

[andyp73]

Not sure what the issue is or what other information would be helpful to troubleshoot this.

The VBox.log file from your guest (zipped) would be a good place to start, along with any relevant information show on the VM guest screen when it fails to boot (72dpi screen shots of the relevant bit work best).

-Andy.

[jeff.sadowski]

Attached Screenshot and zipped log file. There really doesn't seem to be anything useful in them. What other information should I try to include?
Should I try network sniffing with wireshark and see what it picks up on the bridged adaptor?

[jeff.sadowski]

Interesting I installed wireshark and did a filter to and from my vm's mac address
my filter looks like so

Code: Select all

eth.addr == 08:00:27:8c:f2:ba

I see 8 DHCP Discover packets,

All of them say [Malformed Packet]

I attached the filtered packets in typical wireshark gzipped format.

I installed wireshark on my windows dhcp server.

I applied the same filter and nothing shows up while booting the vm.

something is lost on the way.

[andyp73]

Not sure which of us has the non-standard configuration (or whether they are just different). I have the extpack installed and have a remote boot VM that uses the Intel PXE ROM rather than the iPXE one you have. Is that an option for you?

-Andy.

[jeff.sadowski]

It appears that my interface card only allows outbound traffic with it's mac address maybe there is information on the driver for it.

> lspci |grep Ether
00:19.0 Ethernet controller: Intel Corporation Ethernet Connection I217-LM (rev 04)

I thought this was a problem with Ubuntu 16.04 and since I had never had a problem with my home machine (which I have fedora on) I thought I'd switch and see if I could get rid of the problem.

oddly enough I still have the problem and it appears to be related to the network card and or its drivers.

It seems to be using the e1000e driver so I will look into it.

[jeff.sadowski]

Not sure which of us has the non-standard configuration (or whether they are just different). I have the extpack installed and have a remote boot VM that uses the Intel PXE ROM rather than the iPXE one you have. Is that an option for you?

-Andy.

I didn't even see an option. I just picked all the standard stuff for a fedora 64 bit machine. I am convinced that the issue is related to my ethernet card and or its drivers. Otherwise the packets I see on my local machine would have also been seen by the windows dhcp server and they are not.

[Martin]

Did you verify that both iptables and firewalld services are really stopped?

[jeff.sadowski]

Did you verify that both iptables and firewalld services are really stopped?

Thank you for the suggestion

iptables -L was showing stuff but the policies were ACCEPT so I ran the following just incase

iptables -I INPUT -j ACCEPT
iptables -I OUTPUT -j ACCEPT
iptables -I FORWARD -j ACCEPT
iptables -F

systemctl status firewalld shows inactive dead

then I tried again and I still get the same results. but at least I know now my iptables is completely open now.

[Martin]

The reason for my question was that Fedora is now seems to be using the new 'firewalld" firewall by default and not the old 'iptables'.

[jeff.sadowski]

The reason for my question was that Fedora is now seems to be using the new 'firewalld" firewall by default and not the old 'iptables'.

Yeah I forgot to show that I checked systemctl on firewalld. and sestatus shows disabled.
It has to be the e1000e driver not sending packets and I didn't see an option that looked helpful here
https://gist.github.com/pklaus/319367

I tried adding a file /etc/modprobe.d/e1000e.conf

and added

options e1000e CrcStripping=0

to see if that helps when I reboot it still performs the same but from what I was reading here
https://askubuntu.com/questions/51226/h ... parameters

I should see CrcStripping in /sys/module/e1000e/parameters/ but I don't
so it seems that option isn't being applied.

updated:
that page is incomplete. I checked with "dmesg" and my option is getting applied so that option didn't help

[jeff.sadowski]

Not sure which of us has the non-standard configuration (or whether they are just different). I have the extpack installed and have a remote boot VM that uses the Intel PXE ROM rather than the iPXE one you have. Is that an option for you?

-Andy.

Other troubleshooting I had done. If I put it on a nat (with the dhcp from virtualbox) network it gets an ip address. I can use that to install if I setup the nextserver on it somehow. but I would like to be on the same ip scope as my host machine.

[jeff.sadowski]

Ugh it looks like I would need to build a machine as a dhcp server to give a nextserver option as I don't see any way to add options to the dhcp server provided with virtualbox.
Better option might be to use a libvirt network. If I can't set it up in there I can at least setup a dhcp server on my host and have it send it over the libvirt interfaces.

Hmm seems like someone else wanted to use those options

https://www.virtualbox.org/ticket/2759

I can't quite interpret what I need to do to point my pxe boot machine at my windows tftp server(also my dhcp server) and the pxelinux.0 file on it.

When I put it from bridged to nat the following shows in my log it is looking for a file
~/.config/VirtualBox/TFTP/<Machinename>.pxe
But I'm not sure what the contents of that file should be.

[jeff.sadowski]

I tried creating ~/.config/VirtualBox/TFTP/<machinename>.pxe

It's contents are

Code: Select all

chain tftp://<tftpserver ip>/pxelinux.0

the machine log file show

TFTP: ~/.config/VirtualBox/TFTP/<machinename>.pxe -> VINF_SUCCESS

however the vm is stuck at

tftp://<ip from nat dhcpserver>/<machinename>.pxe... ok

and just sits there.

[jeff.sadowski]

More odd behavior while trying to use libvirt.

When I set the vm's network back to bridge and select virt0

then try and start it It has the same behavior as when bridged with the outside interface.

however looking at libvirtd status I see
that libvirt's dhcp server is offering an ip.
So libvirt is trying to communicate with the vm
but the vm doesn't recognize it
output of systemctl status libvirtd

Code: Select all

Redirecting to /bin/systemctl status libvirtd.service
● libvirtd.service - Virtualization daemon
   Loaded: loaded (/usr/lib/systemd/system/libvirtd.service; enabled; vendor preset: enabled)
   Active: active (running) since Wed 2018-02-21 10:48:27 MST; 2h 33min ago
     Docs: man:libvirtd(8)
           http://libvirt.org
 Main PID: 1059 (libvirtd)
    Tasks: 18 (limit: 32768)
   CGroup: /system.slice/libvirtd.service
           ├─1059 /usr/sbin/libvirtd
           ├─1513 /usr/sbin/dnsmasq --conf-file=/var/lib/libvirt/dnsmasq/default.conf --leasefile-ro --dhcp-script=/usr/libexec/libvirt_leaseshelper
           └─1514 /usr/sbin/dnsmasq --conf-file=/var/lib/libvirt/dnsmasq/default.conf --leasefile-ro --dhcp-script=/usr/libexec/libvirt_leaseshelper

Feb 21 13:10:35 daddles dnsmasq-dhcp[1513]: DHCPDISCOVER(virbr0) 08:00:27:f7:54:98
Feb 21 13:10:35 daddles dnsmasq-dhcp[1513]: DHCPOFFER(virbr0) 192.168.122.107 08:00:27:f7:54:98
Feb 21 13:21:40 daddles dnsmasq-dhcp[1513]: DHCPDISCOVER(virbr0) 08:00:27:f7:54:98
Feb 21 13:21:40 daddles dnsmasq-dhcp[1513]: DHCPOFFER(virbr0) 192.168.122.107 08:00:27:f7:54:98
Feb 21 13:21:40 daddles dnsmasq-dhcp[1513]: DHCPDISCOVER(virbr0) 08:00:27:f7:54:98
Feb 21 13:21:40 daddles dnsmasq-dhcp[1513]: DHCPOFFER(virbr0) 192.168.122.107 08:00:27:f7:54:98
Feb 21 13:21:40 daddles dnsmasq-dhcp[1513]: DHCPDISCOVER(virbr0) 08:00:27:f7:54:98
Feb 21 13:21:40 daddles dnsmasq-dhcp[1513]: DHCPOFFER(virbr0) 192.168.122.107 08:00:27:f7:54:98
Feb 21 13:21:43 daddles dnsmasq-dhcp[1513]: DHCPDISCOVER(virbr0) 08:00:27:f7:54:98
Feb 21 13:21:43 daddles dnsmasq-dhcp[1513]: DHCPOFFER(virbr0) 192.168.122.107 08:00:27:f7:54:98

08:00:27:f7:54:98 is the mac on my vm. I rebuilt it a few times seeing if there where other options but didn't find any,