@Tom-Elliott

Thank you for getting back to me so fast.

Definitely an issue with external firewalls or selinux. I had the install script bug on things mainly with any of the “remi” repos so is why I installed manually. any of these packages were installed with “appstream” and not “remi”

in any event I am not using this server on the internet so have disabled the firewall service.

For example the PHP I installed was 8.0.30 on the “appstream” repo and the remi repo has a 8.4.13. I think the version of PHP was giving me this problem. In order to get the default member FOG version button to work I had to comment out the line FOGCore::checkAuthAndCSRF in /var/www/html/fog/status/kernelvers.php - If I put the uncomment it out, the button then does not show the versions. So I can recreate it with my set up. when it doesn’t work, I checked pressing F12 in my browser in developer mode and can see that the “/var/www/html/fog/status/kernelvers.php” has a 403 forbidden error.

I’ve attached a picture here:

Maybe a security setting (selinux and firewalld are both disabled) but since this is rocky possibly I need to do commands to permit these?

Before I did this temp fix with the commenting out, I tested a new “/var/www/html/fog/status/kernelvers.php” file (after making a backup of the original to see if this could be fixed) and that broke the FOG Version information script (I get a “Failed to get latest info” - which as the server is not on the internet is fine, other than I would get in bold red letters “This is not the latest version of FOG before” - which is the only bug I seemed to have. Since I did this it is not a “FOG” issue per se, but could the version of php have something to do with it?

I saved the file created for “/var/www/html/fog/status/kernelvers.php” (I saved it as a text file) which broke this function (i revered to the original as this was a test)

kernelvers.txt

This is the original file (saved as a text file) with the comment out the line FOGCore::checkAuthAndCSRF:

kernelvers_orginal.txt

As to my pxe problem. I removed the change in /etc/systemd/system/tftp.socket.d/override.conf and pxe is working fine now which is strange but I did do a lot of other steps in troubleshooting which I forget now which probably is what actually fixed it.

I can provide the logs but in a nutshell I think the issues I have is that I installed different versions of the software than the installer does or selinux is still somehow blocking this.

Matt

@ rodluz @Tom-Elliott

After some effort on the install script I got FOG working on my rocky 9.6 set up with a few caveats.

1. I had to install many of the packages manually, like php, mysql, tftp due to selinux and firewalls (including hardware firewalls) so ended up with PHP 8.0.30-3 on this system. This version of PHP does not have a seperate JSON package.

2. I installed 1.5.10.1698 from the stable branch. I plan on trying from the dev branch later to see if that will resolve my issues.

3. I had to do some work to get tftp and pxe to work correctly. I’ll lay this out here.

• created a symlink from /var/lib/tftpboot > /tftpboot

• enabled and started the tftp service manually - systemctl enable tftp & systemctl start tftp

• then created a file in /etc/systemd/system/tftp.socket.d/override.conf (after making a directory “/tftp.socket.d” and giving that directory proper ownership & permissions) with the following code:

[Socket]
ListenDatagram=69

[Service]
ExecStart=
ExecStart=/usr/sbn/in.tftpf --foreground --secure /tftpboot

• reloading the systemd service daemon - “systemctl daemon-reload” & “systemctl restart tftp.socket”
  The server would then pxeboot at least BIOS at this point.

Now what doesn’t work is that:

The Web UI currently seems to have php conflicts. I installed php-8.0.30-3 and the modules connected to this when installing the server. The web UI has a quirk on the FOG configuration page when you have a button for DefaultMember FOG Version: () it is not printing the version of FOG or the versions of bzimage or init.xz files and in checking using curl http://ip-address-of-server/fog/service/getversion.php - i ony see - 1.5.10.1698

here is a picture of the error

This isn’t a big deal for me at the moment a I can capture and deploy an image. but I think there should be a way to fix it. I wanted to check if the php version is the issue or if I should use a dev branch (I kind of don’t want to)

@ rodluz

Okay - So I got FOG working on my rocky 9.6 set up with a few caveats.

1. I had to do some work to get tftp and pxe to work correctly.

• created a symlink from /var/lib/tftpboot > /tftpboot

• enabled and started the tftp service manually - systemctl enable tftp & systemctl start tftp

• then created a file in /etc/systemd/system/tftp.socket.d/override.conf (after making a director /tftp.socket.d and giving that file permissions) with the following code:

[Socket]
ListenDatagram=69

[Service]
ExecStart=
ExecStart=/usr/sbn/in.tftpf --foreground --secure /tftpboot

• reloading the systemd service daemon - “systemctl daemon-reload” & “systemctl restart tftp.socket”

The server would then pxeboot at least BIOS at this point.

The Web UI currently seems to have php conflicts. I installed php-8.0.30-3 and the modules connected to this when installing the server. The web UI has a quirk on the FOG configuration page when you have a button for DefaultMember FOG Version: () it is not printing the version of FOG or the versions of bzimage or init.xz files and in checking using curl http://ip-address-of-server/fog/service/getversion.php - i ony see - 1.5.10.1698

here is a picture of the error

error.png

@ rodluz,

Sorry I missed your newer post.

Yeah I checked and it is enabled . I will disable it once I have this in production and off of the web as yes selinux will block aspects of FOG

Just as a datum, I found FOG to work very well with redhat distros. I am not a fan of debian based distros due to package management and stablility - debian I can get working, ubuntu though…

You can set this as resolved.

Matt

@rodluz,

Thank you for your time and answer.

This is a fresh install of rocky on a bare metal server - I did have to install it first on a SATA drive, then copy everything and then install with chroot to manually add a raid controller driver for the ancient 3ware 9750 RAID controller on it. In any event I am up to the point where I downloaded everything now and just need to let the FOG installer set it all up. Unfortunately I don’t have a photo - but I can tell you the error was in the log file I sent you in my last post, which I am adding here.

foginstall2.log

There is an post I did about a year ago on downloading the FOS binaries. I tried using curl and would time out. I ended up downloading more current FOS binaries (bzimage, init.xz, etc) from github that I have running on a current production server as this works for the newish computers we have. I figured this is not the issue this time but I also solved that by running the installed with a --nofail Exit flag which would keep going if it failed to download it with the installer (as I manually put them in the location anyway)

The server is on a connection that is firewalled and some things work and some things not, which remi is one of them although I can browse the site, I think the caching on that repo is the issue -maybe they should add a few mirrors. I can download the rpms manually from that repo - but that is a very painful and troublesome way to install things on linux.

On the apptream question, this is installed off of the web, not the iso - i ran "sudo dnf php --disable-repo=remi --disable-repo=remi-safe --disable-repo=remi-modular and it would install php and the repo shown was “appstream” - maybe from centOS stream sources? IDK but php and the various modular files and so one installed in this fashion as did tftp-server, mysql and mysql-server, and several others I forgot now.

yeah I have used FOG for about 10 years now and we love this project here as once we have it working it’s stable as a rock. The first install I did was on a fedora box back wheN it was version 20, then CentOS 7, recently about a year ago we set it up with rocky linux 8.9 (as CentOS died) to include newer features and bzimage files in order to use the newer model computers we need to image.

Once I have this base install working, which at this point i

Matt

mrowand
5 minutes ago

@rodluz

Thank you for your return on my post. I don’t have a photo of this when it happened as I did a work around and manually installed these packages since. The logs from that attempt don’t seem to be in the /fogproject/bin directory anymore. When looking at ti, It looks like mainly an issue with the “remi-modular” repo. Which is probably my end actually. even when I had to install these manually with dnf, I had to disable that repo. so TFTP-SERVER, PHP and MYSQL and are from appstream repos from base rocky repos.

I am giving the logs of my past two attempts. I deleted IP addresses from it for security.
fog_error_1.5.10.1698.log foginstall2.log

Anyway, since it skips it as not needed and still works I can ignore it. which is what I thought anyway. I did end up modifying a different script and deleted out “php-json” which you can see with the foglinstall2log. The first log also seems truncated sorry.

I am getting a bug with the init.zx, bzimage files downloading, which happened to me the last time I installed fog. the last time I manually downloaded these files and placed them where they go and re-ran the script I think with a --skip-broken flag, or something like this… Not sure why that’s happening - I think it is firewall related.

I usually update these anyway for newer computers.

I am able to do work-arounds and get it to work, I just thought it odd.

Sorry if not helpful.

Regards, Matt

Dear Developers,

So I am installing FOG 1.5.10 on a server with Rocky Linux 9, from stable using git. This uses php8 by default, with a caveat that the package php-json is now included as a module in php-common. I checked and there is a module named json listed in php, just not a package. The installer is checking for this package and it seems to fail.

I can’t seem to find any rpm for “php-json” in any repo or rpm search and the core installer called “common” has a module for json. what is the correct work around for this? I can find it for php7 (which I used the last time I installed FOG, on a rocky linux 8 system), or php5 (on an old centOS 7 system).

in any event - the remi repo also seems to have cache issues at the moment going through firewalls so I can manually go to a rpmfind and download the rpms. I don’t see one for php-json.

Anyway. I also had to manually install mysql, tftp, tftp-server, lftp, gcc, genisoimage and about 7 other packages “not found” by the installer, which are available in the bae repos… anyway. am I using a untested OS?

Regards,

Matt

Dear Fannarino,

First, what version of FOG are you using? I assume you are using this FOG server for DHCP?

with your VM is the networking set to NAT for the VM or Bridged?

First - Did the install script finish without any errors? I would check that as the script handles all the database stuff FOG needs. Check the Docs on installing FOG to ensure you did all the steps on the database.

Since you are using a different network than that one you use for the internet, then you need to ensure all IP addresses for the interface you are using match. If you set these up with a script using a network for the internet and then changed it, then you will need to ensure all IP addresses match.

Second - On your Fog Settings in the Web UI, check your various IP Address settings particularly
Fog Configuration: TFTP Server
Fog Configuration: Web Server
Storage Management > DefaultMember

Next, check in your Fedora VM, check the following.
/etc/dhcp/dhcpd.conf
/tftpboot/default.ipxe

Then check your FOS, which is the bzimage files FOG uses to load the OS your computer is pxebooting. this might effect proper booting or execution of any task in FOG. You can check this in Fog Configuration, then click on the blue button that says "DefaultMember FOG Version: ( ) it will list two files.
bzimage Version
bzimage32 Version

If this is too old you can replace these, which are located in redhat OSes (I use rocky linux myself) under /var/www/html/fog/service/ipxe. You would need to replace about 4 files which you can download from github, under fog project.

These are the points I can think of.

I am having trouble with no seeming changes on our FOG Server, I have several images that created as a 0 byte file and not a folder with .img files Don’t know what it is doing. The task when it is running seems to capture files, so not sure where these are being stored

I have the Storage set as defaultmember and default group.

The images I have are in" /images" folder on the server each in a directory by iimage and I do have 500 GB of space in the volume

I am running fog 1.5.9 and bzimage 6.12.35

this is only in the last couple of days
…

So looking at it futher it seems to create a 0b marker file with the image name and then puts the files in /images/dev/mac-address-of-host/ and the image files when capturing images. Then when deploying images it cannot find it as it is looking for /images/image_name.

So this is an odd change since we changed to bzimage 6.12.35

I can manually recreate the /images/image_name directory and then manually move the files to it and it will deploy.

Possible my bzimage is re-directing it to the “/dev/mac_address” directory?

We prefer the /images/Image_name method as it is easier to troubleshoot when we get errors,

Should we update FOG itself to 1.5.10 or whateever is newer and stable (i don’t like beta or nightly builds as this is a production machine) to get that functionality back?

Dear Tom,

Thanks, and this is what I figured. At least I can pxe boot it and capture the image now, if I switch it to “AHCI/NVMe” mode, and switch back to “RAID” mode to boot the single NVMe drive.

I made this known so we can reach out to Dell and possibly have this changed as it actually doesn’t need this “RAID” mode.

Also we still have to disable secure boot on any computer either HP or Dell to get the pxe boot to work.

We did manage to use FOG to capture a RAID from an HPE DL380 Proliant Gen10 RAID controller and use that to load Windows Server 2022 now, so it does seem the kernel is the issue here on this intel motherboard RAID.

Regards,

Matt

I am using Fog 1.5.9 and a bzImage files with kernel 6.1.32 as this what we needed for our HP Z G9 workstations. I later changed it the FOS of 2025-07-15

We are testing a new computer, a DellProSlim computer, model qcs1250, which are new demo model. using an Intel Rapid Management Driver.

These have 3 options in the UFEI/BIOS under storage for Windows to boot from
RAID Mode
AHCI/NVMe Mode
Disabled.

Dell Had the “RAID Mode” as default and AHCI/NVMe will not let us boot the drive. The RAID mode is the default and is the mode that Windows 11 boots from.

We attempted to even just register the computer in fog, but the FOS would halt and not do anything. I updated the bz image from release 2025-07-15 and it would at least get to a point where it had an error of not detecting the NVMe drive.

I did a work around to switch it to “AHCI/NVMe mode” and it worked and I could register and capture the image and the FOS kernel would detect it. I had to swithc it back to “RAID Mode” to boot Windows 11 again for a single drive.

I can try the latest FOS kernel of 31 July, but I have a feeling this is a long standing issue with the linux kernel used.

Is there a BZ image we can try that has this Intel RAID stuff in it to try? Otherwise I can get this set up with Windows 11 from scratch or some registry hacks so we can leave it on AHCI mode.

If anyone else has these gen of Dell computers would be good to know.

@Tom-Elliott

Thanks Tom. Understood.

An Idea I had was programming this into the BIOS or requesting HP to make a BIOS/UFEI which has this feature of “channels”

Or using a boot file, such as with secure boot, but that would require HP or Dell or whichever PC manufacturer to make boot files for secure boot.

but this would be out of the scope of the project.

I do appreciate your input though.

Regards,

Matt

@Tom-Elliott

Thanks Tom.

I since went with 1.5.9 and was able to re-install the server with that, I did have to run the installer a few times to load all the missing packages.

I ended up putting the kernel, init files there manually. Then with the installer running and adding the --ignore fail switch installed the files that way.

Then I updated to the latest kernel in the FOG GUI so I have the latest kernel running as I need this for current HP Z2 G9 mini and small form factor computers.

This is running so I considered this solved.

Matt

Dear FOG developers.

Have been using FOG for years and sometimes we’ve had to re-image or newly purchased PCs that have multiple hard-drives installed (either SATA Hard-drives, SSDs or NVMe drives).

This has caused at times when imaging that a different drive than the one intended is imaged. As a work around we currently remove the drives we don’t need to image, image the PC and reinstall the drives.

Would like to see if there is a way to specify TFTP and PXE to image the correct drive for the OS as a feature.

We are currently using FOS Kernel 6.1.63, FOG 1.5.9 and is installed on Rocky Linux 8.9

Thanks,

Matt

@ FOG Project

I am new to the forums but not to using FOG.

Basically I had to re-install our master FOG server from scratch as the OS volume failed. The original master was on CentOS 7.7 I am now using Rocky Linux 8.9

I luckiliy have a backup of most images and the database.

Anyway, I first tried installing 1.5.10 first using git to download the install files and then ran the shell script. I chose the “master branch” as I usually don’t do the dev stuff as this will be a production server ( I do have a dev branch on a VM for testing

I have a proxy set up here which I did set up the server to go through the proxy. I later tried a line without any proxy and it made no difference.

The script first had an issue with the REMI repo so it would not download certain packages but the installer somehow did not fail and it go to the end.

I re-ran the script after moving the remi specific repos and all packages installed however the installer would hang on downloading the binaries (kernel, init and services files)

I tried what I found in the forums to use curl to download these manually and I would get a 3KB file for the zip and 3kb file for checksum which were both empty zip files that were corrupt.

the curl commands I tried:

curl -kOL http://fogproject.org/binaries1.5.10.zip
curl -kOL http://fogproject.org/binaries1.5.10.zip.sha256

Then I opened firefox and manually entered the URL and I got a webpage that had a nginx error in fedora on it.

So I then deleted the 1.5.10 foilder and downloaded the 1.5.9 tarball and ran the install script into the same problem with the binaries downloading.

I tried the curl commands this time and these worked.

curl -kOL http://fogproject.org/binaries1.5.9.zip
curl -kOL http://fogproject.org/binaries1.5.9.zip.sha256

and I was able to run the checksum and manually put the url to download the files manually using firefox both from the linux server and a windows server so the url doesn’t work.

The files download seemed to hang for a very long tim (like two hours) and nothing would happen.

So on the line with the proxy I ran the installer with the -X option added and running the export http:_proxy commands and it would finally error on the binaries failing to download. The installer finished and I was able to download the binaries manually otherwise.

Do I put these files in “/var/www/html/fog/service/ipxe” manually? or is there some other location needed for the server to work.

The error I ran into is new to 1.5.10 as I have used fog with a proxy connected or not for 10 years and never ran into the binaries fail;ing to download before

it seems the binaries.zip file for 1.5.10 is corrupt

Regards,

Matt

Dear Fog Project,

We have been using for for about 10 years and it is a great program. We recently had to rebuild a FOG server.

I have the same problem as the OP.

I am installing FOG on rocky linux version 8.9 as CentOS is now a rolling release distro.

The server in question has all packages (curl, bc, dhcp, mysql, php, etc) in the long list installed I tried on a line with no proxy and with some filtering and the results are the same.

I also tried version 1.5.9 and the FOG installer hangs on the “downloading binaries, init and client libraries” There is no output that shows any error and it doesn;t fail - it almost as if it is hanging on something at the other end.

On the same server I tried the following commands in the /root/fogproject/tmp directory (unless there is another location these should be installed in) when I was trying to install 1.5.10.

curl -kOL https://fogproject.org/binaries1.5.10.zip
curl -kOL https://fogproject.org/binaries1.5.10.zip.sha256

Both of these files are a 3 kb file which has nothing in it. I tried to download the files using that url in firefox and I get a nginix error.

so then I tried 1.5.9 which I suscessfuly installed 6 months ago on this same internet line. I get the same basic error on the installer but I can manually download the files and use the url in firefox.

curl -kOL https://fogproject.org/binaries1.5.9.zip
curl -kOL https://fogproject.org/binaries1.5.9.zip.sha256

These files are 64 MB.

I usually figure out these issues and can install FOG and get it working but this one has me stumped.

Best Regards,

Matt