Boot loop after imaging UEFI

[jemerson93]

Running into an issue with newer Dell systems.

Here is what goes through at the moment. Complete image, laptop reboots, boots to hard drive and boot loops with REFIND. Try removing PXE boot from the equation and then no boot device found. These laptops are UEFI only and are unable to boot Legacy. I have our DHCP server configured to allow UEFI imaging and booting but I am at a loss at how to resolve this issue.

Currently the BIOS is set as follows. Secure Boot - disabled. Legacy ROMS Enabled and Legacy Boot Enabled.

Any ideas?

For UEFI server options, 066 is server address and 067 is ipxe.efi

[Tom Elliott]

@jemerson93 you’re contradicting yourself. You said they are unable to boot via legacy mode but that your bios is configured to do so.

[Wayne Workman]

@jemerson93 Tom’s right. So if you have Legacy ROMs enabled and Legacy Boot enabled, the systems are infact operating in BIOS/Legacy mode or some kind of bios/uefi intercompatibility mode - not UEFI. So for that, you’ll need to use a boot file that is built for BIOS/Legacy, so you might try undionly.kkpxe first. Also, refer to this: https://wiki.fogproject.org/wiki/index.php?title=BIOS_and_UEFI_Co-Existence

[jemerson93]

@wayne-workman @Tom-Elliott My apologies, I see that I mistyped a few things.

These laptops have the following boot modes. UEFI and Legacy External Devices. If I disable secure boot and image the laptop, after imaging, I get stuck in a missing CSM firmware REFIND boot loop.

The issue I am having is I cannot get the UEFI image and boot to work. It will image, but get stuck in the REFIND boot loop.

[Wayne Workman]

@jemerson93 said in Boot loop after imaging UEFI:

It will image, but get stuck in the REFIND boot loop.

Try different exit types for that particular host. You register your hosts, right?

[Tom Elliott]

Also what version of fog?

[jemerson93]

@Wayne-Workman @Tom-Elliott As an update, I have completed the steps on the BIOS & UEFI coexistence in the link previously.

I complete a full host registration prior to imaging.

Current version of Fog is 1.5.2.

I could try using a different exit type for that host but I’m running into the REFIND issue with quite a few Dell laptops we received today.

[jemerson93]

@Wayne-Workman @Tom-Elliott

Couple other bits of information is that after imaging the hard drive, the BIOS shows NIC IPV4 and NIC IPV6 but no other boot devices.

Disabling those as the boot devices leaves No boot device found.

Most recent kernels are installed as well.

[Wayne Workman]

@jemerson93 said in Boot loop after imaging UEFI:

I could try using a different exit type for that host but I’m running into the REFIND issue with quite a few Dell laptops we received today.

You can set your global default inside of web gui -> FOG Configuration -> FOG Settings use your browser’s search feature with ctrl+f and look for the word exit. Also, you can also set exit types for particular hosts via groups in-mass. So there are options there that allow you to do less repetitive work. So my suggestion is still to try different exit types.

[jemerson93]

@wayne-workman @Tom-Elliott As an update, I’ve tried multiple exit modes and have ran into chainloading failed with those.

Another instance I noticed is after the imaging is completed, if I enter the BIOS, I have the usual NIC IPV4 and IPV6 but no other boot devices such as Windows Boot Manager or hard drive.

Kind of at a loss on what I am missing for UEFI to work. I’ve gone through and verified that I’ve done everything located below.
https://wiki.fogproject.org/wiki/index.php?title=BIOS_and_UEFI_Co-Existence

[Wayne Workman]

@jemerson93 It sounds like your DHCP is working correctly - your problem is with the exit modes. What exact models of systems are you having these problems on?

[jemerson93]

@wayne-workman Currently the issue is residing on Dell Latitude 5590s. Received a few new ones to deploy and all are experiencing this issue.

[jemerson93]

@wayne-workman Update: Also happening on Dell Latitude 7480s. Only support UEFI. Below is how this one was imaged

UEFI Network Stack enabled
Secure Boot disabled
SATA changed to ACHI

Stuck in REFIND loop. If exit type is changed, receiving a chainloading fail.

Earlier I tested an old workstation set to Legacy and imaged fine. Still thinking it’s an issue with UEFI or my image.

Currently I have 1 master image being used (split off into multiple smaller images). Sysprep it and upload to Fog. I don’t have separate images for Legacy/UEFI. Should I create a specific image a certain way for only UEFI machines?

[Wayne Workman]

@jemerson93 said in Boot loop after imaging UEFI:

I don’t have separate images for Legacy/UEFI. Should I create a specific image a certain way for only UEFI machines?

You absolutely need an image for Legacy and an image for UEFI. One does not work on the other.
Have you tried every exit type individually? Are you fully turning off the system between each try? A simple ctrl+alt+delete does not clear the firmware memory properly and will botch a retry of a different exit type.

[jemerson93]

@wayne-workman That could be a big issue though. I am creating my images via a VM. Anything specific I should look at for creating a UEFI based image?

I did do CTRL ALT DELETE to test each exit so I can power off and go from there.

[Wayne Workman]

@jemerson93 said in Boot loop after imaging UEFI:

Anything specific I should look at for creating a UEFI based image?

Not really, if FOG is capturing it successfully then you’re good to go for that part.

@george1421 any ideas with the refind loop?

[jemerson93]

@wayne-workman As another update, I’ve tested all exit types and all have failed to boot.

Exit gives me a chainloading fail
The rest just keep doing the boot to hard drive in Fog

I edited the /var/www/html/fog/service/ipxe/refind.conf to scanfor internal,external,hdbios,biosexternal and timeout to 0 per another forum post and still no luck.

Again, even after imaging, BIOS shows both NIC boots but no other boot device (such as Windows Boot Manager) or a hard drive. Believe all laptops being affected are M2 SSD’s. Unsure if they are nvme.

[Wayne Workman]

@jemerson93 said in Boot loop after imaging UEFI:

Believe all laptops being affected are M2 SSD’s. Unsure if they are nvme.

Then it may be that there is something wrong with the imaging process - it may be that the exit types are working fine but the local disk simply isn’t bootable. Are there any errors during image capture or deploy?

[jemerson93]

@wayne-workman I do not notice any errors during capture or deploy.

[jemerson93]

@wayne-workman curiosity question as I was doing some more research. My current image I’m testing is formatted as a MBR partition. I read that UEFI needs to be formatted as GPT to be bootable. I just created a VM (using Proxmox) as UEFI bios and converted it to GPT. Going to upload it to Fog in the am. Think that could deal with the REFIND loop?