Hosts are looking for tftp server.

[Sebastian Roth]

@marted @george1421 Nice things you guys worked out. I am at a loss with this. Can’t really believe that dnsmasq is not able to respond to 20 clients “at the same time”. Though I don’t have a really good idea on how to tackle this issue.

Well here is one other thing you can try. Check to see if the switch you have all the clients hooked to with some running into the issue. See if you can setup a port mirror to receive all the communication from one (or a couple) of the client port on a mirror port. Hook up a laptop/PC to that mirror port, assign a static IP to it (doesn’t matter which because we don’t need it to comuinicate with the network, we just want to make sure it does not send out DHCP requests itself). Install wireshark and capture all the traffic up to the point where you have DHCP information captures with hosts that show the “enter tftp” message. Now filter the wireshark capture to only see information with that MAC address. I would be really interested to actually see what DHCP information it received.

[marted]

@george1421
I created 4 interfaces

inet 192.168.149.43/22 brd 192.168.151.255 scope global dynamic noprefixroute ens32
    inet 192.168.148.254/22 brd 192.168.151.255 scope global eth10:0
    inet 192.168.149.254/22 brd 192.168.151.255 scope global eth11:0
    inet 192.168.150.254/22 brd 192.168.151.255 scope global eth12:0
    inet 192.168.151.254/22 brd 192.168.151.255 scope global eth13:0

I changed the dnsmasq conf like this

dhcp-range=eth10,192.168.148.254,proxy
dhcp-range=eth11,192.168.149.254,proxy
dhcp-range=eth12,192.168.150.254,proxy
dhcp-range=eth13,192.168.151.254,proxy

but nothing changed, I mean I still have 5 to 7 posts asking for tftp server.
Could you check the config. Maybe I missed something.

dnsmasq status

root@foglabunix:/var/log# clear
root@foglabunix:/var/log# nano /etc/dnsmasq.d/ltsp.conf
root@foglabunix:/var/log# nano /etc/dnsmasq.d/ltsp.conf
root@foglabunix:/var/log# systemctl status dnsmasq
● dnsmasq.service - dnsmasq - A lightweight DHCP and caching DNS server
   Loaded: loaded (/lib/systemd/system/dnsmasq.service; enabled; vendor preset: enabled)
   Active: active (running) since Tue 2020-03-10 13:35:28 EDT; 10min ago
  Process: 4589 ExecStop=/etc/init.d/dnsmasq systemd-stop-resolvconf (code=exited, status=0/SUCCESS)
  Process: 4606 ExecStartPost=/etc/init.d/dnsmasq systemd-start-resolvconf (code=exited, status=0/SUCCESS)
  Process: 4597 ExecStart=/etc/init.d/dnsmasq systemd-exec (code=exited, status=0/SUCCESS)
  Process: 4596 ExecStartPre=/usr/sbin/dnsmasq --test (code=exited, status=0/SUCCESS)
 Main PID: 4605 (dnsmasq)
    Tasks: 1 (limit: 4915)
   CGroup: /system.slice/dnsmasq.service
           └─4605 /usr/sbin/dnsmasq -x /run/dnsmasq/dnsmasq.pid -u dnsmasq -7 /etc/dnsmasq.d,.dpkg-dist,.dpkg-old,.dpkg-new --local-service --trust-anchor=.,19036,8,2,49aac11d7b6f6446702e54a16073

Mar 10 13:46:06 foglabunix dnsmasq-dhcp[4605]: 2327500257 available DHCP subnet: 192.168.150.254/255.255.252.0
Mar 10 13:46:06 foglabunix dnsmasq-dhcp[4605]: 2327500257 available DHCP subnet: 192.168.151.254/255.255.252.0
Mar 10 13:46:06 foglabunix dnsmasq-dhcp[4605]: 2327500257 vendor class: MSFT 5.0
Mar 10 13:46:06 foglabunix dnsmasq-dhcp[4605]: 2327500257 client provides name: C42145
Mar 10 13:46:10 foglabunix dnsmasq-dhcp[4605]: 2327500257 available DHCP subnet: 192.168.148.254/255.255.252.0
Mar 10 13:46:10 foglabunix dnsmasq-dhcp[4605]: 2327500257 available DHCP subnet: 192.168.149.254/255.255.252.0
Mar 10 13:46:10 foglabunix dnsmasq-dhcp[4605]: 2327500257 available DHCP subnet: 192.168.150.254/255.255.252.0
Mar 10 13:46:10 foglabunix dnsmasq-dhcp[4605]: 2327500257 available DHCP subnet: 192.168.151.254/255.255.252.0
Mar 10 13:46:10 foglabunix dnsmasq-dhcp[4605]: 2327500257 vendor class: MSFT 5.0
Mar 10 13:46:10 foglabunix dnsmasq-dhcp[4605]: 2327500257 client provides name: C42145
lines 1-22/22 (END)

[george1421]

@marted I can’t speak for this setting since I’ve never used it.

dhcp-range=eth10,192.168.148.254,proxy
dhcp-range=eth11,192.168.149.254,proxy
dhcp-range=eth12,192.168.150.254,proxy
dhcp-range=eth13,192.168.151.254,proxy

I can say for 99% of the dnsmasq installs this line is all that is needed.

dhcp-range=<fog_server_IP>,proxy

Just for clarity this is a new vm and not the fog server where dnsmasq is installed?

So if you are on a subnet where these 5-7 workstations ask for the tftp server. If you have wireshark loaded on a third computer with the capture filter of port 67 or port 68 do you only see one or 2 offers? There has to be still something else going on here. Dnsmasq should be fast enough to respond to 100s of requests per second.

[marted]

@george1421 @Sebastian-Roth
Please somebody to help me with the options 66 67
on windows DHCP (Infoblox). Finally the University accepted to add the info for tftp on the main dhcp and like this I can stop the dnsmasq.
On 66 I put the IP of the server 192.168.149.43
but on 67 I don’t know how to provide the syntax of the boot file. Do I have to write /tftpboot/default.ipxe or I write directly default.ipxe
Thanks

[Sebastian Roth]

@marted said:

Finally the University accepted to add the info for tftp on the main dhcp and like this I can stop the dnsmasq.
Ohhh wow. Keeping my fingers crossed this will fix your issue!

but on 67 I don’t know how to provide the syntax of the boot file.

They need to add several ones for different vendor classes. Hope infoblox is able to do this.

• undionly.kpxe for legacy BIOS machines (vendor class PXEClient:Arch:00000)
• ipxe.efi for most UEFI machines (vendor class PXEClient:Arch:00006 and PXEClient:Arch:00009)
• i386-efi/ipxe.efi for some weird 32 bit UEFI devices (vendor class PXEClient:Arch:00007 - but seldomly used at all).

The first two are important!

[marted]

@Sebastian-Roth just from curiosity— why when a client boot I see default.ipxe like a file boot

[Sebastian Roth]

@marted Because it’s the first file we tell iPXE to load. Take a look at the ipxescript we embed into the iPXE binaries: https://github.com/FOGProject/fogproject/blob/master/src/ipxe/src/ipxescript

[marted]

@Sebastian-Roth thanks for the answer.
Now I need just to put the files without /tftboot in front right?

[Sebastian Roth]

@marted said in Hosts are looking for tftp server.:

Now I need just to put the files without /tftboot in front right?

Yes

[george1421]

@marted I remember infoblox from another thread. I guess we need to be mindful of this thread if we see issues here

https://forums.fogproject.org/topic/11749/uefi-pxe-not-downloading-ipxe-efi-file?_=1583888282658

[marted]

@george1421 you’re right. It was a challenge to figured out the way to boot BIOS and UEFI clients with Infoblox… but We did it and I’m so happy. The info I saw in fogwiki for Windows 2012 server it wasn’t work well on Infoblox. Creating a general roule for Bios clients with .kpxe file and add a filter for UEFI clients didn’t work at all. But what we figured out was that we can simply delete all general config IPV4 and BOOTP and create three IPV4 filters for BIOS 00000 , UEFI 00007 and UEFI 00009. And it works. 5 hours tests but it works . And I was wondering if you want guys to add this info in your wiki or I can add it.
@george1421 @Sebastian-Roth I can prepare a page with all screenshots for all config we did on Infoblox server and I think will be quite handy for people who want to use the principal DHCP for delivering information for boot file and tftp server.
Thanks again for your help and I’m so happy it works now without any proxy DHCP !

[george1421]

@marted Any hints you could provide the next guy would be appreciated. If one person has the problem there will be 10 other later. So if you have a step by step or at least a general guide that can be added to the wiki it would be well accepted.

I’m glad you have it working now so you can get back to imaging and not messing under the hood with networking infrastructure.

[Sebastian Roth]

@marted Just want to keep this topic up as it might be very helpful for others. Not sure if you have time and access to systems right now though.