Machines slow to get DHCP over PXE

  • Everything was working excellent and then I am unaware of any changes it just started taking 2+ minutes to boot PXE to the Fog menu and fail. The initial process of getting a DHCP address is super slow, finally after ~2 minutes the fog menu appears, I select full host registration, and it launches that and tries to get a new DHCP lease, which it times out after 20~30 seconds and gives up.

    I can get tftp files from the fog server very fast using tftpd32. Booting windows normally I get dhcp leases instantly. I am on a gig switch directly connected to the windows dhcp server and fog server, I moved dhcp to the router on the same switch, same behavior… Ping times are 1ms on all devices, tried another downstream switch.

    Fog server: Debian 10.8, Fog 1.59. The ipxe.efi file is dated 03/16/21 - I am pretty sure I have imaged some machines just fine around 04/05/21 just fine. Nothing has changed in my switching and routing environment, I rebooted all switches, router, servers. At a total loss, gonna try loading a fresh install of fog.

  • @sebastian-roth In the back of my mind I was knowing that its something at a lower layer … I am the only admin on this network so I dont know how it self destructed from 1 day to the next, might consider a windows update, but moving dhcp off windows would rule that out. Will try a pcap.

  • Moderator

    @totalimpact said in Machines slow to get DHCP over PXE:

    At a total loss, gonna try loading a fresh install of fog.

    I guess this is not going to help. FOG is not involved much in that early PXE boot process. Sure it serves iPXE binaries and config via TFTP and HTTP but all the DHCP related stuff is handled by other systems in your network. So setting up FOG from scratch won’t help.

    It’s best if you can configure a monitoring port on your switch and capture all the packets on that port where you PXE boot a machine using wireshark/tcpdump. Save it as PCAP file, upload to a file share and post a link here.