another init.xz issue

george1421

@bmaster001 said in another init.xz issue:

No idea if it’s using a raid controller. There’s no option like that in the bios anyway…
I tried “has_usb_nic=1 mdraid=true” in the kernel arguments field for the host, but it keeps hanging on “init.xz…ok” when I create a capture-task for it.

Just for clarity, if you are booing with the FOS USB stick, you need to update the image args in the /boot/grub/grub.cfg file. The FOG server is not part of the booting process at this level.

bmaster001

@Quazz said in another init.xz issue:

I actually took the time to look through the spec sheet this time, seems like the Ethernet port is located on the dock you plug the device in.

I don’t think that makes it a USB NIC, not sure how much support there is for such docks to be honest.

Of course that doesn’t make it a USB NIC, but if the readme file that comes with the drivers say that it’s a USB NIC, then I tend to believe that The device itself doesn’t have any connections, so I assume that the dock is a “close extension” of the device, and we can see it as one.

bmaster001

@george1421 said in another init.xz issue:

Just for clarity, if you are booing with the FOS USB stick, you need to update the image args in the /boot/grub/grub.cfg file. The FOG server is not part of the booting process at this level.

Just for clarity too: I put “has_usb_nic=1 mdraid=true” at the end of the line "linux $myimage loglevel=7 … " near the bottom of the file?

bmaster001

Two updates:

Switching between AHCI and IDE in the bios doesn’t make any difference when I try the capture task.
This device is starting to annoy me. The usb connector on top doesn’t seem to work anymore now (usb keyboard nor usb stick are recognised). The only other usb connection still works but then I have to choose between USB stick and USB keyboard This thing also has a built-in ups so removing the power to really reset it, is not so easy: I have to remove a couple of screws to reach a reset-button. Ugh. I think I’m gonna leave it disconnected from the power during the night, and retry tomorrow morning.

Thanks again for the help guys, I’ll get back tomorrow with the results of the FOS usb stick… I hope.

george1421

@bmaster001 said in another init.xz issue:

@george1421 said in another init.xz issue:

Just for clarity, if you are booing with the FOS USB stick, you need to update the image args in the /boot/grub/grub.cfg file. The FOG server is not part of the booting process at this level.

Just for clarity too: I put “has_usb_nic=1 mdraid=true” at the end of the line "linux $myimage loglevel=7 … " near the bottom of the file?

Sorry I should have been a bit more descriptive, I knew exactly what I was talking about.

When you boot off the USB stick using the image I sent you. The booting kernel only looks at that usb drive for settings. So adding things into the fog console like mdraid=true will not make into the booting image. You need to update the /boot/grub/grub.cfg file and add those parameters into the capture deploy line.

Such as in this example.

menuentry "1. FOG Image Deploy/Capture" {
 echo loading the kernel
 linux  $myimage loglevel=$myloglevel initrd=init.xz root=/dev/ram0 rw ramdisk_size=127000 keymap= web=$myfogip/fog/ boottype=usb consoleblank=0 rootfstype=ext4 has_usb_nic=1 mdraid=true 
 echo loading the virtual hard drive
 initrd $myinits
 echo booting kernel...
}

Sebastian Roth

@bmaster001 Thanks for posting the picture. The messages about raid6 might seem strange on first sight but are actually quite normal. See here or various other dmesg outputs on pastebin…

Hanging right after some ACPI messages is what I find interesting. Maybe try kernel parameters to turn of ACPI altogether to see if it makes any difference. See here for a list of all the different kernel parameters. I’d start by trying acpi=off…

george1421

@bmaster001 said

Just for clarity too: I put “has_usb_nic=1 mdraid=true” at the end of the line "linux $myimage loglevel=7 … " near the bottom of the file?

Now that I look at this a second time, the answer is yes. The last menu entry is debug, so add these in that line. But I would try Sebastian’s suggestion of acpi=off first since that is where the process seems to stop. Once you can get into debug mode then you can add the settings that work to the menu entry 1 for capture / deploy.

bmaster001

Finally, some progress!
When I disable UEFI Boot in the bios, and add ACPI=OFF to the kernel parameters, it boots from the FOS stick!

So what’s next now? I use this method to capture/deploy this type of device? Or are there other steps we can take to make if network-boot?

Update: When I choose the “FOG Image Deploy/Captyre” GRUB entry (to which I added the ACPI=OFF parameter too), then it halts with error “fatal error: unknown request type :: Null” (with or without a tastk scheduled for this host). The “Quick Registration” entry as well. “Client System Information” seems to work fine.

george1421

@bmaster001 Hey that’s great. I didn’t have high expectations because of the class of computer it is. Yes you need to schedule a capture task on the fog server then boot the FOS client. We need to clean up the FOS client a bit and of someone forget to schedule a capture/deploy before booting FOS to have FOS wait. But that’s another issue.

Now to the other issue. If you are using FOG r8050 or newer you should not get the type-null error as long as you schedule the task first in FOG and then boot the FOS client. The capture deploy step is the only action that requires a job to be scheduled on the fog server first. Your OP stated you are on 8099. Is that still accurate?

Tom Elliott

@bmaster001 I don’t know what the value looks like that you’re selecting (as you’re using the FOGS-L USB System), but from the sounds of things you’re in the “debug” console? If that is the case, when you type fog the init you’re host is currently booted into has no understanding of the parameters the normal PXE boot would hand out. Do make it recognize things, you need to (still) schedule a tasking on the FOG GUI for that host.

All you should need to do is schedule the tasking on the GUI and then type the command fog as you had before. The script will then attempt to make a request to the fog server to setup the parameters needed for that tasking.

george1421

@Tom-Elliott FWIW, on the grub menu there is the capture/deploy that should not give him debug mode. The very last menu entry in the grub menu IS debug. From what the OP describes (with the type–null) that task was not scheduled / ready for that specific host when FOS was booted (or something is going sideways with hostinfo.php).

bmaster001

I first tried the last menu entry, which is the debug mode. That’s when I wrote my previous “yes, progress!” post. Then I did schedule a capture task on the server, booted the host, selected the first menu item (capture/deploy), and then it gave me the “type null” error. At that point I added the update to my previous post.

@george1421 As far as I remember, I didn’t downgrade the fog server… but I’ll make sure to check that on monday!

I can’t test anything until after the weekend (I’m already home and don’t have access to this device of course), but if you guys can think of more things to try on monday, just let me know! Always happy to test things with this weird piece of hardware… and hopefully it can help someone with the same issues in the future!

Sebastian Roth

@bmaster001 Great to hear you are making progress. If I remember correctly George and Tom have changed some code in the init files lately. Please upgrade to the very latest trunk version and see if you still get that fatal error: unknown request type :: Null thing.

bmaster001

Ok, now I am very confused:

First I updated to the latest trunk version
I tried to choose the first grub-menu item on the FOS stick, with and without a capture task in the server, and both failed with the same error
I tried running “fog” from within the debug menu item -> with the same error
then I noticed that the MAC address that I see with ifconfig is totally different from what I have in the fog server, so I created a new host on the server, with that MAC address, but that doesn’t help either
Then I removed the USB stick, and re-enabled UEFI boot in the bios to check the MAC address, and suddenly the network boot works?!

Did you guys change something in the latest trunk version to make this thing boot correctly?
Is it normal that a pxe boot gives me a different MAC address than an USB-boot ?

Sebastian Roth

@bmaster001 said:

Did you guys change something in the latest trunk version to make this thing boot correctly?

Possibly a newer kernel that does handle this hardware properly? @Tom-Elliott?

Is it normal that a pxe boot gives me a different MAC address than an USB-boot ?

Should not be the case if both are using the same NIC! Would you mind posting both MACs here so we can have a look. Usually MAC addresses have a vendor part and we might shed a light on this if we see the two different addresses.

bmaster001

I already did a lookup for both addresses, and it’s the second one that’s strange because it couldn’t find a manufacturer:

00:40:fd:0a:41:a8 is the one I see when network-booting, which is LXE. That’s correct, because the manufacturer of this VM3 device was called LXE a few years ago
f6:c5:25:ec:b1:ff is the address I see when booting from the usb stick. No vendor can be found for this one… maybe it’s something virtual?

george1421

@bmaster001 said in another init.xz issue:

Ok, now I am very confused:

Did you guys change something in the latest trunk version to make this thing boot correctly?

Tom said that the kernels were updated last monday or tuesday to the latest release. My fos-usb.img file has the older kernels on it. If you built your own then it should have the latest kernels on it.

I did wonder myself if the fos client was picking the wrong network adapter as compared to what was captured during inventory.

bmaster001

I didn’t build the img myself. But apparently the newer kernels seem to help me a lot

The inventory task didn’t work either so I had to create the host manually in the fog server, and I used the MAC address I saw during network boot for that. No idea where the other address comes from. It seems to work though, when I ping from another pc to this device, I see that same mac address in my ARP table. Very strange. EDIT: If you want me to do more tests with your image, just let me know, I have this device on my desk for a few more days

Next problem: there seems to a problem with the NTFS partition on the device, so the capture doesn’t start. I have to boot to windows and run “chkdsk /f”. Can’t do that of course, since I want to capture the image as it is
Quick registration and Full Host Registration still don’t work. After init.xz the screen goes black
memtest gives me “Exec format error”

So we’re not entirely there yet…

bmaster001

I’m now on version 8185, and pxe boot hangs again on “init.xz…ok”. Weird, I’m sure it worked yesterday.
Disabled uefi again, booted the fos usb stick, and noticed that the MAC address is totally different as before. It seems that the mac address is different after each reboot. That makes this fos image pretty hard to use for deployment/capture tasks with fog because it gives the “fatal error: unknown request type :: Null” error each time. @george1421: any idea why this happens? I tried starting from the stich in debug mode, update the host on the server so that it matches the virtual mac, and ran ‘fog’ on the host, but that gave the same error. Maybe this is fixed when the real mac is used when booting … ?

george1421

@bmaster001 This is totally strange, I can’t understand why the mac address is coming from random places. I’m not going to rule out something strange going on in the FOS engine, but this hardware is the only one doing this (so far)

What I’m going to suggest is that you update the kernel and inits on that usb stick. Just download the files using these urls
https://fogproject.org/inits/init.xz
https://fogproject.org/inits/init_32.xz
https://fogproject.org/kernels/bzImage
https://fogproject.org/kernels/bzImage32

And replace the files in the /boot folder on the stick. This will put the latest kernels and inits on that stick.

Once that is done boot from the usb stick and select the debug boot. That will drop you to a command prompt on the device.

Then key in ip addr show and record the network mac addresses
Key in . /usr/share/fog/lib/funcs.sh
Key in set | grep mac
Record the value
Reboot
Test 2 more times. See if the mac address is dynamic (for some reason).

Hopefully some pattern will show up. The mac variable is what fog uses to identify the target to the FOG server.

I’ll tell you a secret for debugging the FOS engine. if you use the ip addr show you can find the IP address of the target. Then if you set a password for the logged in user (root) with passwd you can then use putty to ssh into the FOS engine. This makes it easier to copy / paste take screen shots of what is going on in the FOS environment.

another init.xz issue

170

11.6k

17.1k

154.6k