Posts made by rodluz

@george1421 Okay so I found it in /proc/devices and I was able to get the HP program working!!
I had to add the program files to the init.xz filesystem and modify parts of the program to not use the modprobe & rmmod commands.
I also had to create the directories needed for dynamically loaded modules and add the compiled module into it so that the program can look for it there.

I now have to do the third part which is custom postinit scripts to use the system MAC address instead of the dongle address. This should be pretty easy.

Once I have that working, I am going to do a full write-up of it for reference for myself in the future and anyone else that comes across this.

Hello, it has been a while since I have tried anything for this.
I now have more time and more equipment to actually work on this.

I have taken some time to learn a bit more about the Linux kernel and adding a module to it (still so much more to learn).
@george1421 I used the reference link, excluding the arm specific parts of the Kconfig, and compiled the kernel. I can see that the driver was added during the compiling process because I see this output CC drivers/hpuefi/hpuefi.o, but I can’t figure out if I did it correctly.
When I pxe boot to Debug Mode I can’t run lsmod or modprobe to find if it was installed.

These are the steps I did to compile the kernel. Please let me know if this is correct.

1 . Clone FOS repo from GitHub
2. Run ./build.sh --kernel-only --arch x64
3. When the output “We are ready to build. Would you like to edit the config file [y|n]?” comes up, I opened another terminal and then did this:

a. Made directory hpuefi in /path/to/fos/kernelsourcex64/drivers/
b. Copied the hpuefi module files hpuefi.c & hpuefi.h to that directory.
c. Made Makefile in /path/to/fos/kernelsourcex64/drivers/hpuefi with content:

obj-$(CONFIG_HPUEFI)    += hpuefi.o

d. Made Kconfig in /path/to/fos/kernelsourcex64/drivers/hpuefi with content:

config HPUEFI
tristate "HP UEFI Kernel Module"

e. Modified /path/to/fos/kernelsourcex64/drivers/Makefile adding:

obj-$(CONFIG_HPUEFI)    += hpuefi/

f. Modified /path/to/fos/kernelsourcex64/drivers/Kconfig adding before endmenu:

source "drivers/hpuefi/Kconfig"

g. Modified /path/to/fos/kernelsourcex64/arch/x86/configs/x86_64_defconfig adding:

CONFIG_HPUEFI=y

4 . Returned to the original terminal with the prompt still open and entered n.
5. Entered y for the prompt: HP UEFI Kernel Module (HPUEFI) [N/y/?] (NEW)
6. Pressed Enter for everything else.

After it finished, I copied the bzImage to my testing FOG server.
I can boot into Debug Mode but don’t know if the module was added successfully.

Sorry for the long read but hope this is enough information to figure out if/what I did wrong.

@george1421 Thank you for the explanation. I am trying to do this not as a fix for the passthrough to work, instead as a patch/workaround. @Sebastian-Roth I don’t think this should be added to FOS for everyone since it doesn’t really fix anything, again it’s just a workaround for me. I have read many threads of people bringing up this issue and maybe this can be a sort of solution for now but it would only work on HP machines.
In my case, iPXE is getting the correct MAC (passthrough from BIOS) but FOS is not. My thought was to run this utility when registering the computer to FOG so that the MAC address attached to it is the passthrough one instead of the adapter.

I tried this utility on a laptop running Ubuntu and I know that I can retrieve the system MAC address for sure, and that it is correct. I am limited in my Linux knowledge to know how to add the module to FOS and then compile it. I honestly got lost with the reference that @george1421 provided.

Hello I need some guidance in modifying FOS to include a new package/software.

I have some new HP laptops that don’t have an ethernet port, this means that I have to use a USB-C to Ethernet adapter. When I register those laptops to FOG, the MAC address it adds is from the adapter instead of the passthrough system MAC address. It seems to me like the issue is with HP’s UEFI/BIOS programming, because it also doesn’t passthrough when in Windows/Linux.

HP has a software called “HP Flash and Replicated Setup Utilities for Linux” which I can use to retrieve the system’s MAC address from Linux.

My idea is to modify FOS and add that software to it, then modify the FOG registration script to check if the UEFI/BIOS has a system MAC address that differs to the one FOS can see. (Side note: I know that this would only work with HP computers and it is fine because we only use HP)

I know that I can modify the registration script using this https://wiki.fogproject.org/wiki/index.php/Modifying_the_Init_Image, but I am limited in my Linux knowledge in modifying FOS with the software.

Here are the install commands from “user guide”:

To install the kernel module:
    • tar xzf hpuefi-mod-x.x.x.tgz
    • cd hpuefi-mod-x.x.x/
    • make
    • sudo make install
To install the utility applications:
    • tar xzf hp-flash-<version><architecture>.tgz
    • cd hp-flash-<version><architecture>
    • sudo ./install.sh

Since FOS is stripped down a lot, it doesn’t have the make command or the Linux headers to be able to run the make command, also it would be horrible to have to install the software every time FOS loads.

I would greatly appreciate any help possible.
Thank you in advanced!

@sebastian-roth I will have to wait until late next month to update. I have to image around 200 computers in the coming weeks so I’ll leave those laptops as the last devices to image in case something breaks with the update.
Thanks for all your help, I will reply to the thread after I update the server.

@sebastian-roth Okay so this is the whole output

>>>Start PXE over IPv4.
  Station IP address is 10.28.68.52

  Server IP address is 10.27.70.100
  NBP filename is ipxe.efi
  NBP filesize is 1060352 Bytes
Downloading NBP file...

  Succeed to download NBP file.
iPXE initializing devices...ok


iPXE 1.21.1+ (gc64d) -- Open source Network Boot Firmware -- http://ipxe.org
Features: DNS FTP HTTP HTTPS iSCSI NFS TFTP SRP VLAN AoE EFI Menu
Configuring (net0 30:24:A9:...)..... ok
Received DHCP answer on interface net0
tftp://10.27.70.100/default.ipxe... ok
http://10.27.70.100/fog/service/ipxe/boot.php... ok
Booting... (Press ESC to access the menu)

The iPXE version is newer, because it used to be 1.20.1+ before the change.

I was the one that set up the DHCP setting for my colleague’s and my VLANs so they are both ipxe.efi.
I checked to make sure that nothing is broken between VLANs and switches and that everything is Gigabit.
I also tried disabling the ADVANCE MENU LOGIN & PXE MENU HIDDEN just to check and still same issue. I went into the FOG Configuration and updated the MAC addresses too but no luck.

@sebastian-roth Okay so I did that and it’s still acting the same way.

I guess the next step would be to update to the dev-branch?

@sebastian-roth Thanks for the help. I prefer to swap out the iPXE binaries first and see if that helps.

Do I just copy those linked files into /tftpboot then change ownership to fogproject:root and chmod 655 the files, or is there something else I need to do?

@sebastian-roth Hello and thank you for the reply. I tried multiple times on the other FOG server just to make sure with different “bad” laptops and they all worked.

The other FOG server is on 1.5.9.94.

I took a “bad” laptop to the server room and plugged it in to the same switch that my FOG server is connected to and I am still having the same issue.

I originally thought it could be the NIC on those “bad” laptops so ran speed tests using Windows and Linux(Ubuntu) and they are fine. Maybe it could be something with the MAC on the laptops and FOG not accepting it.
I tried using a USB-C to Ethernet adapter to test and it works perfectly with that.

I could just use multiple of those adapters to image the computers but they hold the host data after I register the laptops.

EDIT:
I originally had the kernel version using 4.19.64. Updated it last week to 5.10.34. Just updated it to 5.10.50 today and still having the same issue.

FOG Version: 1.5.9
OS: Ubuntu 20.04 - up to date

Hello, I am having a very strange issue. I have 30 new HP EliteBook 850 G6 laptops with sequential serial numbers. Some of the laptops are working perfectly when trying to register them as hosts. Others, are not booting to the host registration screen because the bzImage part takes around a minute to go up 1%.
When going into pxe boot, the laptop take a while to get to the iPXE menu items.

http://10.27.70.100/fog/service/ipxe/boot.php... ok  #This happens instantly
http://10.27.70.100/fog/service/ipxe/boot.php................ ok  # This takes around 10-15 seconds
http://10.27.70.100/fog/service/ipxe/boot.php......... n%  # This takes about 5-10 seconds

Once I press enter on the Perform Full Host Registration and Inventory menu option, I get:

bzImage................................ n%  # Takes around a minute to go up 1%

At first I thought I may have a batch of laptops that are bad, so I looked through the serial numbers and noticed its all over the place.

I took a laptop that I know doesn’t work to a coworker’s area and used her FOG to see what would happen and there was no issue. This makes me think I screwed up on something with my FOG server.

Only changes I have done to the server recently are post download scripts and add new images/snapins.

Any ideas?

Thanks in advanced!

@george1421 @Sebastian-Roth So I modified my postdownload scripts to manually mount the correct drive if the code that was originally there didn’t do it. It is working for me so I am adding my post download scripts below in case anyone comes across this thread and finds it helpful for them. This works for me in my environment, but may not work for everyone.

I took inspiration from the getHardDisk() function in https://github.com/FOGProject/fos/blob/master/Buildroot/board/FOG/FOS/rootfs_overlay/usr/share/fog/lib/funcs.sh

The scripts are a bit of a mess but working

fog.postdownload

#!/bin/bash
## This file serves as a starting point to call your custom postimaging scripts.
## <SCRIPTNAME> should be changed to the script you're planning to use.
## Syntax of post download scripts are
#. ${postdownpath}<SCRIPTNAME>

clearScreen;

dots "Running post download scripts."

mkdir /ntfs &>/dev/null
ntfs-3g -o force,rw $part /ntfs

echo "Mounting Device";

if [ "$?" = 0 ]; then
        if [ ! -d "/ntfs/Windows" ]; then
                . ${postdownpath}mountLargestPartition
        fi

        echo "Done"
        debugPause

        . ${postdownpath}getMachineModel
        . ${postdownpath}getPlatformVersion
        . ${postdownpath}fog.drivers
        . ${postdownpath}fog.unattend

        if [ -z "$auditMode" ]; then
                . ${postdownpath}fog.bcu
        fi

        echo "Unmounting Device";
        umount /ntfs;
        debugPause
else
        echo "Failed to mount device";
        sleep 30;
        debugPause
fi

---

mountLargestPartition

#!/bin/bash

dots "Running the manual mount script"

drive=$(lsblk -dpno KNAME -I 3,8,9,179,202,253,259 | uniq | sort -V)
driveCleaned=${drive//'/dev/'/} #Remove `/dev/` from variable

partition=""
partSize=0

# Get the largest partition
while IFS= read -r line; do
        tempPartSize=$(blockdev --getsize64 /dev/$line)

        if [ $tempPartSize -gt $partSize ]; then
                partSize=$tempPartSize
                partition="/dev/$line"
        fi
done < <(ls /dev | grep -E "${driveCleaned}[a-zA-Z0-9]+")

umount /ntfs &> /dev/null
ntfs-3g -o force,rw $partition /ntfs

---

getMachineModel

#!/bin/bash

dots "Retrieving Machine Model"

machine=$(dmidecode -s system-product-name) # Gets machine model
machine="${machine// /_}"  # Replace spaces with underscores

echo "${machine}"
debugPause

---

getPlatformVersion

#!/bin/bash

dots "Retrieving Platform Version from Registry"

# Create a registry file to be able to search the ReleaseId from
# reged -x gives out an abort error message but from all my testing it still works.
reged -x /ntfs/Windows/System32/config/SOFTWARE HKEY_LOCAL_MACHINE\\SOFTWARE \\ /out.reg &> /dev/null || true

if [ -f /out.reg ]
then
        PlatformVersion=$(cat /out.reg | grep 'ReleaseId')
        PlatformVersion=${PlatformVersion:13:4} # Grab only actual the ReleaseId. Full text looks like "ReleaseId"="1909"
        echo "${PlatformVersion}"
else
        echo "No registry file found."
        debugPause

        dots "Retrieving Platform Version from PlatformVersion.rek file"

        if [ -f "/ntfs/Windows/PlatformVersion.rek" ]
        then
                PlatformVersion=$(<"/ntfs/Windows/PlatformVersion.rek")
                echo "${PlatformVersion}"
                debugPause
        else
                echo "No platform version file found."
                debugPause
        fi
fi

---

fog.drivers

#!/bin/bash

dots "Preparing Drivers"

mkdir /ntfs/Drivers &>/dev/null;
echo "In Progress"

dots "Adding drivers to driver path"
cp -ar "/images/drivers/${machine}/${PlatformVersion}/" /ntfs/Drivers

if [ "$?" = 0 ]; then
        echo "Done"
else
        echo "Failed!"
fi

debugPause

regfile="/ntfs/Windows/System32/config/SOFTWARE"
echo "regfile ${regfile}"

key="\Microsoft\Windows\CurrentVersion\DevicePath"
echo "key ${key}"

devpath="%SystemRoot%\inf;%SystemDrive%\Drivers"
echo "devpath ${devpath}"

reged -e “$regfile” &>/dev/null <<EOFREG
ed $key
$devpath
q
y
EOFREG

echo “Drivers done.”

---

fog.unattend

#!/bin/bash

dots "Searching for Platform Version"

# If $PlatformVersion is not null/empty
if [ -n "$PlatformVersion" ]; then
        echo "Found"
        debugPause

        dots "Removing old unattend file"
        rm -f /ntfs/Windows/Panther/*nattend.xml
        if [ "$?" = 0 ]; then
                echo "Done"
        else
                echo "Failed to remove file!"
                echo "Aborting fog.unattend"
                debugPause
                sleep 60
                return
        fi
        debugPause


        dots "Checking image mode"
        if [ -n "$auditMode" ] && [ "$auditMode" -eq 1 ]; then
                echo "AUDIT"
                debugPause

                dots "Copying audit mode unattend file"
                cp "/images/unattends/UnattendAuditMode.xml" "/ntfs/Windows/Panther/Unattend.xml"
                if [ "$?" = 0 ]; then
                        echo "Done"
                else
                        echo "Failed!"
                fi
                debugPause
        else
                echo "OOBE"
                debugPause

                dots "Copying unattend file with driver path"
                cp "/images/unattends/UnattendWithDriverPath.xml" "/ntfs/Windows/Panther/Unattend.xml"
                if [ "$?" = 0 ]; then
                        echo "Done"
                else
                        echo "Failed!"
                fi
                debugPause
        fi
else
        echo "No valid Platform Version."
        echo "Skipping unattend."
        debugPause
fi

---

fog.bcu

#!/bin/bash

dots "Copying BIOS change script"

if [ -f "/images/drivers/CMSL-BIOS_Change.ps1" ]
then
        cp "/images/drivers/CMSL-BIOS_Change.ps1" "/ntfs/Drivers"
        echo "Done"
        debugPause
else
        echo " Failed! Could not find bios change script."
        debugPause
        sleep 60
fi

echo "fog.bcu done."
debugPause

I did a few more tests and realized that the $part variable is also not set on the image that is working, so probably not that. The other thing I realized, my working image has the NTFS partition on /dev/sda4 but the “bad” image has it on /dev/sda3. Not sure why the new image is different.

A bit more context, in case it could be the issue:

• The “good” image is Windows 10 Version 1909
• The “bad” image is Windows 10 Version 20H2

Hello, I am having an issue when deploying one of my images. After Partclone is done, my post download scripts start, however it seems that the $part variable is not set correctly so the /ntfs directory only has Recovery and System Volume Information and is not mounting my ntfs partition.

I haven’t changed anything of the mounting part from fog.postdownload.

dots "Running post download scripts."

mkdir /ntfs &>/dev/null
ntfs-3g -o force,rw $part /ntfs

echo "Mounting Device";

if [ "$?" = 0 ]; then
        echo "Done"
        debugPause
        . ${postdownpath}getMachineModel
...
...

I’m pretty sure it’s something with the specific image I am using, since I tried a different image and I don’t have that problem. I don’t know how my new image could be affecting this, but I didn’t do anything different for the “bad” image.

I’m not sure what I should try to figure out what the issue is.
I was able to mount the /dev/sda3 partition manually using ntfs-3g -o force,rw /dev/sda3 /ntfs

@george1421 Thanks so much! So what fixed it for me was just removing all the other iPXE commands. So now my parameters are:

login
kernel bzImage loglevel=4 initrd=init.xz root=/dev/ram0 rw ramdisk_size=275000 web=http://10..../fog/ consoleblank=0 nvme_core.default_ps_max_latency_us=0 rootfstype=ext4 mac=${net0/mac} ftp=10.... storage=10....:/images/ storageip=10.... irqpoll chkdsk=0 capone=1 type=down img=1909GoldenImage imgType=n imgPartitionType=all imgid=43 osid=9 imgFormat=0 auditMode=1
imgfetch init.xz
boot || menu

This works perfectly for what I need.

@george1421 I did make a selection but maybe I made I mistake on the kernel command?
Also where would I put the sanboot option?

@george1421 Thanks for the response. I am not super familiar with all the ipxe menu stuff so I am doing something wrong.

My custom menu parameters are:

login
params
param mac0 ${net0/mac}
param arch ${arch}
param username ${username}
param password ${password}
isset ${net1/mac} && param mac1 ${net1/mac} || goto bootme
isset ${net2/mac} && param mac2 ${net2/mac} || goto bootme
kernel bzImage loglevel=4 initrd=init.xz root=/dev/ram0 rw ramdisk_size=275000 web=http://10..../fog/ consoleblank=0 nvme_core.default_ps_max_latency_us=0 rootfstype=ext4 mac= ftp=10.... storage=10....:/images/ storageip=10.... irqpoll chkdsk=0 capone=1 type=down img=1909GoldenImage imgType=n imgPartitionType=all imgid=43 osid=9 imgFormat=0 auditMode=1 isdebug=yes
imgfetch init.xz
boot

I have the Menu Show with option set to All Hosts

When I try to run it, I get this error:

http://10..../fog/service/ipxe/boot.php... ok
Booting from SAN device 0x80
Boot from SAN device 0x80 failed: No such device (http://ipxe.org/2c222087)
Could not boot: No such device (http://ipxe.org/2c222087)
Could not boot: No such device (http://ipxe.org/2c222087)
Chainloading failed, hit 's' for the iPXE shell: reboot in 10 seconds

Hello, I am wondering if there is any way to pass a variable from a custom ipxe menu item that can then be read and used by FOS?
I duplicated the fog.deployimage menu item and would want to pass a variable that would be used in my post download scripts.

What I would want to do, is have my post download scripts only run some scripts depending if that variable is true/1 or false/0. So if I select that new menu item it would have the variable.

@Sebastian-Roth I know this thread is a bit old, but my question is exactly this thread.

I am having the same issue as @jtappen. In my case, all of my hosts had the enforce checkbox selected on my prior FOG server. After I moved everything to a new server for some reason the checkboxes got unchecked.
I tested with a few computers by clicking the checkbox and it works perfectly fine. My problem is that I have over 200 hosts so I would like to “checking ALL the checkboxes” by updating the database.

Would anything break if I run UPDATE hosts SET hostEnforce='1'; to update all hosts in one blow?

@george1421 Ahh okay thank you!