udhcpc: sending discover

[jhalbert]

Hello folks

I have few sites I can’t get ip quick enough before fog gives up. Boot to PXE fine. No issues, gets IP address. Fog runs down some scripts and when it normally starts sending the image part it trys to get IP again from DHCP. During that part of this, does it give sending discover. I watch the switch and it turns green right when it assay failed to get IP via DHCP! Tried on interfaces enp1s0
Please check your network setup and try again!

I don’t see any choice within fog to address this problem. Suggestions?

[george1421]

When this typically happens, its not a fog issue but an infrastructure one. This is symptomatic of having spanning tree enabled on the business switch but not having one of the fast STP protocols enabled (RSTP, MSTP, fast-STP, portfast, or what ever your switch mfg calls it).

A quick test to see if its a spanning tree issue is to put an unmanged (read cheap dumb switch) in between the pxe booting computer and the building network switch.

The issue is FOS Linux boots so fast, it gives up before traditional spanning tree starts forwarding data on the switch port. For traditional STP, it waits for 27 seconds listening for a BPDU packet. If it doesn’t hear one then it starts forwarding data on the port.

[jhalbert]

@george1421 That is what I was reading. I know its spanning tree because I am imaging computers with my netgear switch that is $30. I was wondering if there is any work around. Maybe make Linux wait longer? Even 10 more secs would solve my problem. I am on site because I could not reimagined this pc remotely. I know its my whole site but I can’t turn that feature off. It has already save me few times.

[george1421]

@jhalbert Understand I’m NOT recommending you turn off spanning tree, only change its mode to one of the fast spanning tree protocols. The fast protocols use optimistic blocking (forward first then listen for BPDU) instead of pessimistic (block while listening for BPDU) that standard stp uses.

Is there a way to do it without changing the network, yes but it will require a little work on your part (honestly you need to enable RSTP to fix it for all). You will need to unpack the init.xz file. Its located on the fog server in /var/www/html/fog/service/ipxe directory. The instructions for unpacking and packing the inits are here: https://wiki.fogproject.org/wiki/index.php/Modifying_the_Init_Image Follow the 1.0.0 and up instructions. Once they are unpacked in the unpacked directory you need to edit a file in ./etc/init.d called S40Network. Just before the udhcpc command add in a line that says sleep 27 Save the file and then repack the inits and move to the /var/www/html/fog/service/ipxe directory.

What that will do is have FOS Linux pause for 27 seconds on every boot to let the network settle before trying to dhcp boot.

[tech49]

@george1421 I hate to ask this, but could you write up a detailed guide on how to do this? I’m beginner level at linux. I unpacked the file and when i tried to open up the init file i could not save the file after changing the values. It was being edited in Terminal.

I also noticed in fog that there is a setting called “Remit Hello Interval” in the storage settings. Would this be the same settings?

Thanks for any advice you can give!

[george1421]

@tech49 What version of FOG are you using?

I don’t have a step by step because this is a bit more advanced function than beginners should approach. But if you unpack the inits it will create a directory structure that mirrors the virtual hard drive. You can than traverse that directory. You can edit the file using the host OS gui editor if you are using a gui based OS like ubuntu. Or you can edit the file using a character based editor. vi is the most common but also the most cryptic to use.

I’ll go through the steps with vi to show you at least how to edit.

You will need to navigate to the etc/init.d directory where the inits were unpacked to.
vi S40Network
Use the down arrow to find the line with udhcp there will be more stuff after but you want the line with the text that starts out as udhcp. At the beginning of the line key in
i to insert mode
cr for a new line
esc key to edit insert mode
up arrow to go to the inserted blank line
i to enter insert mode
space over until cursor is just above the u in udhcp then key in
sleep 27
esc to exit insert mode
:wq colon write quit

That will drop you back to the linux command prompt. Go back up to the root of where you unpacked the inits and then repack them back into init.xz

[tech49]

@george1421 Thanks, I’m using fog 1.5.7 on Ubuntu 18 LTS

for notes reasons this is what I did. I was unable to open it in VI for some reason. The screen was blank, SO I did a sudo gedit and opened it up in the text editor software. This is what I did for the sleep 27 line.

 for retry in $(seq 3); do
       sleep 27
       /sbin/udhcpc -i $iface --now

About to give this puppy a whirl. Will let you know if it worked.

[tech49]

@tech49 No go

[Sebastian Roth]

@george1421 @tech49 Shouldn’t it be:

for retry in $(seq 3); do
       sleep 27
       /sbin/udhcpc -i $iface --now
done

[tech49]

@Sebastian-Roth Oh no, I was just typing in the little bit of it freehanded described in the instruction and guide. I’m not sure if there’s a done there or not. I will check asap.

[george1421]

@tech49 The other thing I would recommend is that you add an echo statement just above the sleep 27 like echo "Sleeping for a bit"; to be sure that the code is executing like we think. Let me go and get the github page so we can be sure.

The other thing is to (with the fog delivered inits) pxe boot into debug mode and then key in /sbin/udhcpc -i $iface --now replacing $iface with the name of your physical network adapter collected with the ip addr show command.

This is the line in question https://github.com/FOGProject/fos/blob/master/Buildroot/board/FOG/FOS/rootfs_overlay/etc/init.d/S40network#L38, where the echo and sleep command needs to be inserted.

[tech49]

@Sebastian-Roth

Wait till the interface is fully up and ready (spanning tree)

timeout=0
linkstate=0
until [[ $linkstate -eq 1 || $timeout -ge 35 ]]; do
    let timeout+=1
    linkstate=$(/bin/cat /sys/class/net/$iface/carrier)
    [[ $linkstate -eq 0 ]] && sleep 1 || break
done
[[ $linkstate -eq 0 ]] && echo "No link detected on $iface for $timeout seconds, skipping it." && continue
for retry in $(seq 3); do
    sleep 27
    /sbin/udhcpc -i $iface --now
    ustat="$?"
    curl -Ikfso /dev/null "${web}"/index.php --connect-timeout 5
    cstat="$?"
    # If the udhcp is okay AND we can curl our web
    # we know we have link so no need to continue on.
    # NOTE: the link to web is kind of important, just
    # exiting on dhcp request is not sufficient.
    [[ $ustat -eq 0 && $cstat -eq 0 ]] && exit 0
    echo "Either DHCP failed or we were unable to access ${web}/index.php for connection testing."
    sleep 1
done
echo "No DHCP response on interface $iface, skipping it."

done

If we end up here something went wrong as we do exit the script as soon as we get an IP

[tech49]

@george1421 How do I pxeboot into debug mode?

[tech49]

@george1421 Is it worth modifying anything in init_32.xz?

[george1421]

@tech49 said in udhcpc: sending discover:

@george1421 How do I pxeboot into debug mode?

When you schedule a task (capture or deploy) before you hit the schedule task button tick the debug checkbox. Then pxe boot the target computer.

The idea with this route is to manually wait the 27 seconds before manually issuing the udhcp command. Your issue may be totally different if time doesn’t fix the dhcp issue (i.e. your computer uses one of the new realtek nic chip sets) .

[george1421]

@tech49 said in udhcpc: sending discover:

@george1421 Is it worth modifying anything in init_32.xz?

Not unless you are working with 32 bit hardware (not likely now days)

[tech49]

@george1421
Ok, I entered in the Echo line and repackaged the file with the guide on the wiki.

The echo command did not work. I saw nothing saying “Sleeping for a bit”

[george1421]

@tech49 What version of FOG are you using? Oh I see it never mind. Let me do this to the inits.

[george1421]

OK here is the modified init: https://drive.google.com/open?id=1ZDLcftq_yx3c0BiIAVhNwwpn2j-gDzCE

If this fails to print the sleep 27 we will need a screen shot of what you see when it fails because it should print the waiting a bit .

[tech49]

@george1421 No echos seen

I saved the file to my desktop and did sudo cp -r /home/username/desktop /var/www/fog/…/ipxe… I browsed to it on the GUI and confirmed on properties it was modified the time id id so.

The output I got after a quick host registration was
populating .dev: done
initializing random number done
starting haveged: ok
starting enp1so interface and waiting fof the link to come up
udhcpc: started v1.29.3
udhcpc: sending discover
udhcpc: sending discover
udhcpc: sending discover
udhcpc: no lease, failing
Either DHCP failed or we were unable to access http://fogserveripaddress/fog//index.php for connection testing.
repeat 3 times
No DHCP respawn on interface enp1s0 , skipping it.
failed to get an ip via dhcp, tried on interfaces enp1s0

Note: I’m using a laptop for the fog server , Dell latitude e5420 Bios: LegacyBoot Secure off
I’m trying to do a Quick or Full host registration on a Dell Optiplex 3050. Bios: Uefi on SecureBoot on

Both confirmed pxe on