dhcp Server issues

And this is what tcdump see's in my PVE host when filtered as you suggested. I can see my test laptop etc.
 

Attachments

  • Screenshot 2025-07-26 at 19.18.34.png
    Screenshot 2025-07-26 at 19.18.34.png
    641.6 KB · Views: 3
Hi,

You need to filter on ports 67 AND 68 in wireshark (capture all, stop, and filter after capture).
And at the same time do it on the host when forcing client to request IP via DHCP, and post both results (in code tag preferably).

And make sure you have nothing that can block the request inside Windows (like a bad firewall rule or firewall in public profile).

EDIT :
Or else you can provide pcap files if you're not sur.
Code:
tcpdump -i {INTERFACE} port 67 or port 68 -vv -w mypcapfile.pcap

And the result of :
Code:
pve-firewall status && pvesh get /nodes/{node}/firewall/options

Best regards,
 
Last edited:
Ok. Though I can’t see how it will help. BTW as part of troubleshooting I disabled windows server firewall so it can’t be that, and I tested with an iPhone on Ethernet which also didn’t work, so it’s nothing on the client pc.

I’ll post requested details shortly, but I think we need to focus on why Proxmox isn’t letting the ports through
 
Hi,

I’ll post requested details shortly, but I think we need to focus on why Proxmox isn’t letting the ports through.
Well, it's a bridge so unless a firewall is activated on proxmox (or something was installed on the host), nothing should filter packet since bridge only forwards and nothing more.

I quote what a proxmox staff said :
the bridge configuration has nothing to do with DHCP for VMs. The bridge basically forwards everything that arrives on one port to all the other ports, so the bridge doesn't really know what it has to forward, it just forwards it.
Cf : https://forum.proxmox.com/threads/dhcp-on-vm.152401/#post-691180

Best regards,
 
  • Like
Reactions: MCal27
Hi,


Well, it's a bridge so unless a firewall is activated on proxmox (or something was installed on the host), nothing should filter packet since bridge only forwards and nothing more.

I quote what a proxmox staff said :

Cf : https://forum.proxmox.com/threads/dhcp-on-vm.152401/#post-691180

Best regards,
That’s my understanding too and my Proxmox 7/Server 2016 system works perfectly! It’s definitely not windows server as I tried with both vm’s on each of the two Proxmox hosts and also with a pfsense VM. I can access its login page fine but again DHCP doesn’t function on that (yes I disabled dhcp server service on windows server).
 
ok this is interesting. I added rules to proxmox firewall after turning it on, and I know get this. the server is offering dhcp, but the discover is still not happy. I again switched off the proxmox firewall and I still get the same. It's like simply adding the rules to the firewall helped abit. I used both port 67 and 68
 

Attachments

  • Screenshot 2025-07-26 at 20.56.16.png
    Screenshot 2025-07-26 at 20.56.16.png
    849 KB · Views: 3
Looking at the output on proxmox host. why is it showing the wrong subnet? I have 255.255.252.0 everywhere I'm sure..
 

Attachments

  • Screenshot 2025-07-26 at 21.12.56.png
    Screenshot 2025-07-26 at 21.12.56.png
    208.8 KB · Views: 3
It’s very clear to me that the firewall in Proxmox 8.4 is buggy. I’ve never had such problems with Proxmox before. Ever. I’ve wasted days in this buggy product and I’ve pretty much run out of time now. So I’m going back to my plan of using a seperate pfsense unit for dhcp and keep that service off Proxmox until they have it working reliably.

Very disappointed
 
Hi,

Something seems off with your result.

If I'm reading correctly, you say that you have to activate the firewall and allow DHCP to work and if you disable it it stops working ?

Can you please give the result of this command when it's not working :
Code:
pve-firewall status && pvesh get /nodes/{node}/firewall/options && iptables -L

Because if it's a bug it should be reported accordingly.

Best regards,
 
Last edited:
it's pointless.
"interfaces" means content of /etc/network/interfaces and please copy/paste into CODE tags.
auto lo
iface lo inet loopback

iface enp88s0 inet manual

iface enp87s0 inet manual

iface enp2s0f0np0 inet manual

iface enp2s0f1np1 inet manual

auto vmbr0
iface vmbr0 inet static
address 10.184.68.24/22
gateway 10.184.68.1
bridge-ports enp88s0
bridge-stp off
bridge-fd 0

iface wlp89s0 inet manual

source /etc/network/interfaces.d/*
 
Hi,

Something seems off with your result.

If I'm reading correctly, you say that you have to activate the firewall and allow DHCP to work and if you disable it it stops working ?

Can you please give the result of this command when it's not working :
Code:
pve-firewall status && pvesh get /nodes/{node}/firewall/options && iptables -L

Because if it's a bug it should be reported accordingly.

Best regards,
Just to be clear, nothing I do ever makes dhcp work. When I enabled the firewall and added options to allow ports 67 and 68 I got the last result i sent you in wireshark. Where DHCP discovery is still showing an IP of: 0.0.0.0 But DHCP Offer shows the IP of the DHCP server. Before that everything was ip 0.0.0.0
 
Is it possible to pay someone to remotely connect to my system and diagnose whats going on? It woukd really help me out nextweek when I have to install all this in a school..
 
Hi,

When I enabled the firewall and added options to allow ports 67 and 68 I got the last result i sent you in wireshark. Where DHCP discovery is still showing an IP of: 0.0.0.0 But DHCP Offer shows the IP of the DHCP server. Before that everything was ip 0.0.0.0
Your wireshark capture looks normal, DHCP discovers with "0.0.0.0" as the source is normal since your client is asking to get an IP.
And after that the offer is sending the result to your client who says, your client will have 10.184.68.10 as IP and 255.255.255.0 as subnet with a lease time of 300seconds and the capture seem not full as the gateway and DNS are missing in the "offer", and the client does not seem to acknowledge the DHCP offer.

Note : command fixed (typo error)
Note2 : you should edit messages instead of posting one after another for readability.

Best regards,
 
Hi,


Your wireshark capture looks normal, DHCP discovers with "0.0.0.0" as the source is normal since your client is asking to get an IP.
And after that the offer is sending the result to your client who says, your client will have 10.184.68.10 as IP and 255.255.255.0 as subnet with a lease time of 300seconds and the capture seem not full as the gateway and DNS are missing in the "offer", and the client does not seem to acknowledge the DHCP offer.

Note : command fixed (typo error)
Note2 : you should edit messages instead of posting one after another for readability.

Best regards,
I still get typo error when I try your command. Sorry I'm out of time.Nothing anyone has suggested has gotten us any closer to a solution. As far as I'm concerned 8.4 is a faulty product. I was going to get a subscription for these two servers but I don't think it's worth it now. I'll find something more robust. Thanks anyway.