Problem loading /bzimage32...

[Ph3noM]

Hi guys!
I have a fresh install ( fog 1.2.0 on debian 8 ), and i’m stuck when i try to perform a full or quick registration from a client pc, the system load /bzimage32… but it stops at ~20%…
I have never been able to go further in the process. (But i can boot on the HDD or start a memtest…)

Anyone have an idea? (I’m noob and it’s a study project )

Thanks you!!

[Tom Elliott]

Please try using a different ipxe file. I’m guessing you’re using undionly.kpxe right now. Maybe try the undionly.kkpxe (anything that works fine with the undionly.kpxe will typically work just as fine with using the undionly.kkpxe)

If that still doesn’t work, you may need to try (for this particular system) the ipxe.pxe file.

[Ph3noM]

I guess you are talking about this:

i’ll try to change it.

Thanks for your prompt reply!

[george1421]

@Tom-Elliott While this is slightly off point, why do we have so many pxe boot kernels? Is there a chart or table some place which tells us the differences or why to use kernel X over kernel Y?

[Wayne Workman]

@george1421 said:

@Tom-Elliott While this is slightly off point, why do we have so many pxe boot kernels? Is there a chart or table some place which tells us the differences or why to use kernel X over kernel Y?

Yes. https://wiki.fogproject.org/wiki/index.php?title=Filename_Information

[george1421]

@Ph3noM No that is the path to where the kernels live. What Tom is referring to is in your dhcp server option 67 needs to be changes to that new kernel that Tom mentioned.

[george1421]

@Wayne-Workman That is close to what I was asking about. There are a few variants that have different proper names two including suffix combinations. Having clear use cases would be helpful. The issue I see is you may have to do a crazy dhcp setup to get boot kernel Y for only the dell model xx and boot kernel Z for Hp model 1 and kernel T for HP model 4. Its all a bit confusion if you don’t have a road map somewhere.

[Wayne Workman]

@george1421 I’ve been thinking about just exactly that for a while. There are a few ways to approach it but it’s all ultimately up to how the hardware identifies itself inside it’s Vendor Class Identifier.

If a client identifies as JUST PXEClient:Arch:00000 then there’s no way of knowing if it needs undionly.kkpxe or undionly.kpxe

However, if some Dell XYZ model identifies as PXEClient:Arch:00000:XYZ then that’s easy to solve, you simply create a new class in ISC-DHCP (or windows server) to hand out the correct file for that Vendor Class Identifier.

For those that don’t identify uniquely, the only way to solve this is to create DHCP reservations that specify what file they should use.

I’m working on a DHCP plugin for FOG to do just this. The idea is to add a field to host general info and group general info to hold a spot for “Boot file”. Once a boot file is set for a host, then a DHCP reservation is simply created in the ISC-DHCP configuration for that specific MAC address(s) and that file name. Easy, right? In addition, I plan to create some basic tools for setting up vendor classes and subnet options, ranges, and so on - only in the scope of FOG usage. Plus a lease viewer.

[Ph3noM]

@george1421
Oh thank you, that’s why I could not find these files ahah
I’ll try this!

Edit:
It still doesn’t works with theses files on my DHCP server option 67:

-undionly.kpxe
-undionly.kkpxe
-ipxe.pxe (with ipxe.pxe the load increases up to ~80% but it still freezes ).

Now i don’t know what to do

[Ph3noM]

I specify that my bench is virtualised with Virtualbox…

[Tom Elliott]

@Ph3noM Is the nic Natted or Bridged?

[Ph3noM]

@Tom-Elliott “Internal network”

[Tom Elliott]

@Ph3noM Can you make it a bridged network, preferrably using E1000

[Ph3noM]

@Tom-Elliott I used this to isolate my network-test from my company’s network. But i’ll try this tomorrow.
(Pardon my english )

Anyway thanks guys for your replys!

[Ph3noM]

Hi!
I used E1000 on my VMs aaaand it works! (But still in “internal network” mode)
Unfortunately, I am faced with a new problem, i can upload an image (from a XP client), but I can’t deploy it on the same Client (for testing)

I changed type of HDD controler, nothing new…

Do you have some advices?
(I feel that I could never finish this properly ahah)

ty

Edit: The client’s system is about ~5Go but the image’s size on the server is only 1,34Go, is it normal??

[Tom Elliott]

@Ph3noM It’s very normal that the size on disk is much smaller than the required disk to put the image on. This is because we compress the file as it’s being uploaded.

The problem you’re seeing with 1.2.0 and XP is well known about and was fixed in SVN version 2096 or 2097, I can’t remember which exactly, and the version of SVN 1.2.0 was released at was 2094, so you could fix this by installing trunk, specifically version 2097. But if you’re willing to upgrade to 2097, why not just go all in and use latest trunk? There have been MANY improvements and adjustments I’m sure you would find you actually like/prefer/enjoy.

[Ph3noM]

No it’s ok now, i just made a mistake.
I am very happy!
Thanks all for the help!

[Wayne Workman]

@Ph3noM said:

No it’s ok now, i just made a mistake.
I am very happy!
Thanks all for the help!

What was the mistake?