proxmox vm's are not receiving networking details

[bbartelds]

Hi all - it seems my post was deleted. I'll try again.
I am running a proxmox cluster consisting of three intel nuc's. (HA, failover etc are not configured at the moment). Recently, the vm's running on the proxmox nodes no longer receive networking information. They do not receive an ip address, and seem to not be connected to the network at all. I am not sure what the root cause of the problem is, but at the moment I can only interact with the vm's running on proxmox nodes through the proxmox gui. I'm running Proxmox VE 7.4-17

network interfaces of proxmox node via gui:


network interfaces config on proxmox node:


ip a - on a proxmox node


ip a - on a vm



ens18/enp0s18 on the vm is what I'd expect to receive networking details on boot, but it does not receive anything.

What could cause this / how can I rectify this?

 

[gurubert]

Where should they receive the networking configuration from?
Is it a DHCP service? Is it still in operation?
Is it cloud-init? Does the VM use cloud-init?
Is it static configuration in the VM?

 

[bbartelds]

There’s a dhcp server running on 10.30.0.1 - the same gateway that also provides the proxmox node with an ip. This service is available and works as expected - no other hosts are experiencing problems.

There’s no cloud-init, the vm doesn’t use it.

I don’t recall ever setting the dhcp server/gateway location. The vm is running ubuntu. When installing the OS, it used to ask for ip configuration - but it’d already have received one via dhcp, so I assume(d) this info would be provided by the proxmox node by default. Currently that isn’t happening. I tried to create a new vm from scratch, but it has the same problem regarding ip/network configuration - no configuration populated, no ip/network details present.

 

[_gabriel]

I assume(d) this info would be provided by the proxmox node by default

no, it's your dhcp server provide ip to VM.
is ping dhcp/gateway from host works ?
try to set ip manually i
within your VM then ping your dhcp/gateway.

 

[gurubert]

I just saw that your Proxmox node has the address 10.30.10.1/24 and a gateway of 10.30.0.1.
This cannot work. The default gateway has to be in the same subnet as the host IP address.

But this also does not explain why the VM does not receive DHCP infos. You could run tcpdump on vmbr0 to watch the packets that are interchanged when the VM boots.

 

[bbartelds]

I just saw that your Proxmox node has the address 10.30.10.1/24 and a gateway of 10.30.0.1.
This cannot work. The default gateway has to be in the same subnet as the host IP address.

But this also does not explain why the VM does not receive DHCP infos. You could run tcpdump on vmbr0 to watch the packets that are interchanged when the VM boots.

That's true. That setting is the same for the other hosts, and does work, however. The 10.30.0.0/16 vlan is managed by my router - its gateway and dhcp service lives at 10.30.0.1.

I can also see the prox node receive and accept a dhcp offer from 10.30.01 upon boot

 

[bbartelds]

no, it's your dhcp server provide ip to VM.
is ping dhcp/gateway from host works ?
try to set ip manually i
within your VM then ping your dhcp/gateway.

Sorry, I expressed myself poorly. Since I haven't defined any additional vlans or networking in proxmox, I expected the dhcp server / gateway the proxmox host uses to be passed on to the vm. But that apparently is not happening.

I cannot ping or talk to the vm in any way, since it has no presence on the network. It has no ip address, no gateway, nothing.

If I set an ip address on the ens18 nic that the vm has, I can then ping the proxmox host - but nothing else. But the vm is not reachable from other parts of my network either.

 

[gurubert]

The 10.30.0.0/16 vlan is managed by my router

Then the IP of your Proxmox host has the wrong network mask. Instead of /24 configure a /16.

 

[bbartelds]

Then the IP of your Proxmox host has the wrong network mask. Instead of /24 configure a /16.

Done. Now the Proxmox host no longer connects to the cluster and is no longer reachable. Going to pull it out of the IT closet and see what the problem is. The issue as reported persists on my two other prox nodes.

Please don't take that as a slight - I'm really grateful for the help you're providing!

 

[_gabriel]

as @gurubert said : 10.30.10.1/24 can't talk to 10.30.0.1
double check your network configuration.

 

[bbartelds]

as @gurubert said : 10.30.10.1/24 can't talk to 10.30.0.1
double check your network configuration.

well .. it can. And it does - 10.30.0.1 was configured as dhcp server for 10.30.0.1/16 on my router.

Jan 7 12:31:23 prox1 dhclient[626]: DHCPREQUEST for 10.30.10.1 on vmbr0 to 10.30.0.1 port 67
Jan 7 12:31:23 prox1 dhclient[626]: DHCPACK of 10.30.10.1 from 10.30.0.1

you're right in the sense that there are obvious issues with this config. But there were no changes in the network configuration. And before, vm's I'd run on 10.30.10.1 / prox1 would actually get an ip address via dhcp. But now, there's no network configuration on the vm at all, and I don't understand why.

If this remains elusive I'll try to rebuild the proxmox cluster with saner network settings. It just bothers me that I don't understand what broke.

 

[_gabriel]

Jan 7 12:31:23 prox1 dhclient[626]: DHCPREQUEST for 10.30.10.1 on vmbr0 to 10.30.0.1 port 67
Jan 7 12:31:23 prox1 dhclient[626]: DHCPACK of 10.30.10.1 from 10.30.0.1

Proxmox VE do not use dhclient to set its ip. it use the ip defined in Network interfaces.
Have you manually started dhclient only to test ?

 

[bbartelds]

Proxmox VE do not use dhclient to set its ip. it use the ip defined in Network interfaces.
Have you manually started dhclient only to test ?

Running dhclient on the proxmox host made it unreachable from the network. So something is just very wrong all round.

 

[gurubert]

DHCP is an Ethernet protocol that is working within an Ethernet broadcast domain. I.e. IP addresses and IP subnets do not have much meaning here.

 

[bbartelds]

I think I am going to rebuild the proxmox cluster, and do it with sane networking. It bothers me that I haven't found a root cause or solution to get the current setup working again, but I don't think it's reasonable to expect finding it at this point.

Thanks so much for your efforts guys!