Unable to boot to some PXE ROM
-
There have been some issues with some PXE ROMs that are unable to boot to the PXE screen. Every time the system is boot the system just locks at the boot screen and reboot. There are not BIOS updates for those systems as the support site does not offered any updates. Is there another type of PXE that needs to be loaded into FOG to be able to boot those systems?
[url=“/_imported_xf_attachments/1/1577_IMAG0309.jpg?:”]IMAG0309.jpg[/url][url=“/_imported_xf_attachments/1/1578_IMAG0310.jpg?:”]IMAG0310.jpg[/url][url=“/_imported_xf_attachments/1/1579_IMAG0311.jpg?:”]IMAG0311.jpg[/url][url=“/_imported_xf_attachments/1/1580_IMAG0312.jpg?:”]IMAG0312.jpg[/url]
-
Is this issue isolated to specific models of computers? And if they support (U)EFI booting, is Secure Boot off and legacy boot enabled?
-
The issue is related to the specific boot ROM. As there are different model of computers that have it. Some are UEFI some are just plain old BIOS. In the UEFI the legacy is enable. I have the same issue with some Lenovo too. Except that the Lenovo give IPv4 Error and then try to boot to IPv6 where it is not enabled. But I would like to focus on the first issue at hand. Sorry for the tangent.
-
I’d probably start by updating your undionly.kpxe file to the newest one Tom has on his website ([url]https://mastacontrola.com/ipxe/latest/[/url]) or trying another iPXE boot file and see if you can get one of the machines you’re having issues with to boot. Instructions can be found here: [url]http://www.fogproject.org/wiki/index.php/Building_undionly.kpxe#Use_other_undionly_or_ipxe_files[/url]. By default you’re probably using undionly.kpxe, but there are other files located in /tftpboot on the FOG server that you could try to see if that resolves the issue with other ROMs. I take the symbolic link approach for testing to avoid having to deal with the DHCP options. For example, to try booting undionly.kpxe.INTEL, you would:
[code]
cd /tftpboot
mv undionly.kpxe{,.ORIG}
ln -s undionly.kpxe.INTEL undionly.kpxe
[/code]to go back to the original undionly.kpxe if none of the other files work, just do the following to revert the changes
[code]
rm undionly.kpxe
mv undionly.kpxe.ORIG undionly.kpxe
[/code]You may find that there’s no common boot file that all machines will boot from, so things get complicated from there. It’s about experimenting to find out what works
-
To add on, the ones that are specific to attempting to load IPv4 and IPv6 sounds to me like they’re attempting to boot via (U)EFI boot systems which would only work with the extension ending in .efi
Further to add on, I’ve updated the undionly/snp/ipxe files on my website so they really do reflect the “latest”. They are embedded with attempting to use the PXE Handed DHCP address rather than trying to re-initialize the devices which I’ve found can cause some issues.
I would recommend starting with the undionly.kpxe, then try the undionly.kkpxe for the best over all support. For systems that may still have problems with this (beyond systems that need the efi files) try the ipxe.pxe file and see if all your systems work with these.
I don’t know what else to say to try and help out either. You all must remember I don’t have all of the devices I’m attempting to get operational. So all of it is just a guessing game from my side.
-
Thank you for all you assistance guys.
All files under /tftpboot were updated with the new files from the link. One point to mention is that the DHCP servers is a DD-WRT device with a redirection to the pxe file. Also it seems that all my system are looking for a default.ipxe file which it is resident under /tftpboot. It seems the code only directs to the default.ipxe and not to any other file in under /tftpboot folder. The default.ipxe has a script that seems to look for the appropriate PXE file but I am not sure. BTW: FOG ver 1.2.0 if that makes any difference.
-
Update:
I tried to boot from one of my systems that I know for a fact it boots from PXE and it is freezing when trying to communicate to FOG. It seems it cannot find the local storage. I got a better idea, this is getting really messy. I would like to start from a clean slate, does any one know how I can move my images to a new system? Can any one point me in the right direction?
-
Update2:
Well a major malfunction has been discovered which it is driving me to think it is not the PXE boot. The systems can download but can’t upload images. Any advice on what it might be?
-
Update3:
Find the issue. In the so many changes the permission were changed. Changing the permission back fix the issue of writing images. However some of the systems still don’t boot using the Realtek B02 D00. Some of them have the RTL 8111B/8111C V2.15 (080314) version. Most of them in the same family are the one giving me issues booting to PXE.
-
Interesting question: Does fog 1.2.0 support Windows 8/8.1?
After booting and trying to upload the images to the server any of the working systems is getting locked when trying to upload the image to the server. If enter is depressed then the following messages are displayed:
*Saving partition tables…Invalid partition data
Done
*Task Completed!
The system is going for down NOW!
Sent SIGTERM to all processes
Sent SIGKILL to all processes
Requesting system poweroffEven knowingly the image hasn’t been upload to the server the task is completed, is this a bug? It seems that the conditions to meet Windows 8/8.1 are not properly define. Is there a FOGprep like there is for Windows 7? Doing just SYSPREP is not working. Any advice?
[url=“/_imported_xf_attachments/1/1588_FOG_Img_Error.png?:”]FOG_Img_Error.png[/url][url=“/_imported_xf_attachments/1/1589_FOG_Img_Mangnt.png?:”]FOG_Img_Mangnt.png[/url]
-
If you’re doing Sysprep why aren’t you using a nonresizable image type? Further to that, it seems that this issues happens when your disk believes it’s mbr but contains gpt data. Why? Because the image upload detects if there’s gpt data on the disk. Many time this is unexpected but it does happen. During the deploy process it guesses that the disk is gpt or mbr based on the filesizeof the d1.mbr file. If it’s neither 32256 or 512 it guesses it was supposed to be gpt. When a gdisk writes the data back to disk the partitions aren’t written properly and leaves the disk with “invalid partition data”. There is not a simple means to check for this and fail gracefully that I’ve been able to find yet so it just goes on to the next item in the script. Hopefully this helps you out a little. If recommend trying SVN if you can and see if it affords you a better time. Also try reuloading the image of you can but make sure the disk has no gpt data if it’s supposed to be using mbr style partition tables.
-
Hi Tom.
Thank you for your response. Base on the option given on the drop down menu, it has to go with single disk multiple partitions. Windows 7 and Windows 8 multipartition the HDD by default, that said, the best option would be Multiple Partition Image - Single Disk which in FOG it is not resizable. I haven’t have any issues with Windows 7 only with Windows 8. Also 4096 KB sectors are native to Windows 8, keep in mind that my aim is to create a distributable image of Windows 8. I will look and see what else it can be done to upload images. I am not sure how to remove the GPT partition as I believe it is define by Windows (under investigation LOL).
-
Update:
As a test I moved Windows 8 to a regular BIOS with MBR and it works just great. However, I am getting systems that only support UEFI. Meaning GPT will be used as a file system in stead of MBR. The BIOS do not offer a way to disable GPT. Will there be a fix to support GPT in the near future? As we all know MBR will eventually dispensary and GPT will be the dominant beast on the field.Thanks in advance for your assistance.
-
Can you list the devices that are “only UEFI”?
-
[quote=“Wolfbane8653, post: 40652, member: 3362”]Can you list the devices that are “only UEFI”?[/quote]
The systems are all ACER computers. I have been up and down in the BIOS and I don’t see anything to disable UEFI. I am also getting some Lenovos and it seems that I will be running on the same issue. On the Lenovo I was able to upload the original image but when I download it back to the system well, the system crashed and a repair had to be done. On the last one, the hard drive was wiped and Windows 8.1 was installed manually (less bloated) to customize the system, in this instance the image just locked again.
One point to clarify is that it never fails, it just sit in the very last screen before actually doing the upload. I have to either press enter or shutdown the system to force the system out of the induced coma. Does that help?
I am still trying to figure out if there are any logs that can be pulled to see what is going on. I know it could be an issue with GPT discovery in the code as mention previously. Is it possible to fix the issue with GPT? The only way to create the image was using raw but it will take too long.
Thank you in advance for any assistance.
-
[quote=“DarkEnergy, post: 40703, member: 26785”]The systems are all ACER computers.[/quote]
Can we get some make and models?
Maybe if someone else has the same device or close they would jump in here and help out. I have an Acer desktop here on a side work order but it works well with fog. -
[quote=“Wolfbane8653, post: 40709, member: 3362”]Can we get some make and models?
Maybe if someone else has the same device or close they would jump in here and help out. I have an Acer desktop here on a side work order but it works well with fog.[/quote]You are focusing in the make an model when you should be focusing in the type of BIOS the system is running. Does your ACER is UEFI or just plain old BIOS? Many systems independent of the manufacturer use the same mobo with the same BIOS/UEFI in the board. For instance Many systems comes with AMI or Phoenix BIOS. It would be more beneficial to pain attention at the type of BIOS and file systems. For instance in my case I have to use GPT as part of the UEFI as I cannot disable UEFI. The Acer is a Veriton M275 series. I will provide the UEFI as soon as the system finish creating the image.
Thank you for all your assistance.
-
[quote=“DarkEnergy, post: 40719, member: 26785”]You are focusing in the make an model when you should be focusing in the type of BIOS the system is running. Does your ACER is UEFI or just plain old BIOS? Many systems independent of the manufacturer use the same mobo with the same BIOS/UEFI in the board. For instance Many systems comes with AMI or Phoenix BIOS. It would be more beneficial to pain attention at the type of BIOS and file systems. For instance in my case I have to use GPT as part of the UEFI as I cannot disable UEFI. The Acer is a Veriton M275 series. I will provide the UEFI as soon as the system finish creating the image.
Thank you for all your assistance.[/quote]
DarkEnergy,
We’re trying to get the make and model because we want to see if we have any body with the same make and model so we can replicate the results. Focusing on the make and model is the only way to ensure your system will work and we have a duplicate system. The BIOS is not relevant as a manufacturer (in many cases) can customize the BIOS. This is why some systems use Secureboot and others might use Windows Mode or some description of some sort.
All we’re trying to do is help. I assure you, giving us your exact make or model (even if we don’t have it within ourselves) will go leaps and bounds as we can possibly find an emulator to work off if (hopefully).
-
Hi Tom:
Yes, I understand where you coming from but base on my experience I have seeing many systems using the same mobo with the same BIOS. While, as you mention, companies can customized the BIOS there are minor modifications. The idea was focus on other possible FOG’s users withe different make and model (feel like I am talking about a car…LOL) but they have the same modo with the same BIOS.
BTW:
The BIOS/UEFI is: Version 2.15.1227 AcerThank you.
-
Right, but the if we have the make and model, cause most don’t keep track of the bios and maker of the mobo, they keep track of the make and model of the system.
It’s easier to ask for:
Who has Acer Aspire D255 v. 3.0?
VS.
Who has Motherboard ASUS H81M-A Firmware Revision 1.3.0a1753?
