suddenly network doesnt work anymore

[cerise]

Hi there
i have running a proxmox machine for a year or so. The machine has a 1GB interface (vmbr0 -> 10.0.0.11) and a 10GB interface (vmbr1 -> 10.0.0.10).






Everything used to run without any problems. But since this week, the vmbr1 is not able to communicate anymore in a strange way:
If i assign a VM to vmbr1, it cannot access the network, it gets no IP from the DHCP server. vmbr0 works fine (but only 1g)

I can ssh / https to 10.0.0.10 and to 10.0.0.11 ... i can also ping 10.0.0.10 and 10.0.0.11 from an external device.

But when im on SSH on the machine and then try to ping a host, like google.de or so, it tells me it cant reach the host. The funny thing is, that DNS request like with dig are still working.

I dont really know what happend, nothing has changed in the config or elsewhere, it simply stopped working.
Does anyone have an idea how to troubleshoot this? would be really happy about your help

best regards and thank you

 

[Moayad]

Hi,

Have you tried to stop the firewall on the vmbr1?

I dont really know what happend, nothing has changed in the config or elsewhere, it simply stopped working.

If nothing is changed, it could be the Ethernet cable?

But when im on SSH on the machine and then try to ping a host, like google.de or so, it tells me it cant reach the host. The funny thing is, that DNS request like with dig are still working.

Have you checked the /etc/resolv.conf in the VM, if correct?

 

[B.Otto]

Hmm, that sounds like a routing problem.

I see one thing that might cause this: you should never assign IP addresses of the same subnet to two different interfaces, that screws up the routing table. Both 10.0.0.10/22 and 10.0.0.11/22 are in the same subnet.

Maybe the DNS requests are cached? How did you check them with dig?

 

[cerise]

Hi, thanks for your reply.

stopping the firewall on the VMs interface didnt worked ... stil gets no IP. I think the problem is not the VM, cause the issue also effect the host.
heres the results or "ip route", "dig google.com" and "ping google.com"

 

[aaron]

May I ask why you have two IPs in the same subnet on two different interfaces? If it is, so that you can differentiate where 1Gbit and 10 Gbit traffic should go, be aware that the Linux kernel does not guarantee that only the interface with the IP is used. It is possible, that the other interface is used if it has an IP in the same subnet!

To really make sure that traffic is using a specific interface, it is best to use different subnets for each.

Also, for your problem, have you tried to remove the IP on one of the interfaces? If only VMs need to access the physical network, there is no need to specify an IP address on a bridge port. Only if the Proxmox VE host itself needs to be reachable.

 

[melroy89]

In my case I was updating some packages on Proxmox (see attachment). After which the network interface number changed. I was kinda in panic, but eventually the problem was clear. I execute: systemctl status networking.service, which gave me:

error: vmbr0: bridge port enp46s0 does not exists

So what happened was somehow, something changed the number of this network interface. Execute ip addr showed me the correct number which is now: enp45s0 (notice the 45 instead of 46). Then I manually edit: nano /etc/network/interfaces.

This allowed me to both change the iface interface to enp45s0 as well as the bridge-ports line to the same enp45s0.
Finally I executed:

ifreload -a

Now everything works fine again.