Problem with network loops and disconnection

[Sergiio]

Hello, it turns out that I have a Proxmox that was just installed a week ago, with a configuration that allows assigning a network with each provider (using vlans) on each machine, everything was going fantastic until last night when it started to lose ping and after restarting it, a loop appeared as I am going to show.





cat /etc/network/interfaces

auto lo
iface lo inet loopback

iface enp0s31f6 inet manual

auto provider1
iface provider1 inet static
        address 192.168.1.118/24
        gateway 192.168.1.1
        bridge-ports enp0s31f6
        bridge-stp off
        bridge-fd 0

# VLAN 20
auto provider2
iface provider2 inet static
        address 192.168.20.2/24
        gateway 192.168.20.1
        bridge-ports enp0s31f6.20
        bridge-stp off
        bridge-fd 0

# VLAN 30
auto provider3
iface provider3 inet static
        address 192.168.30.2/24
        gateway 192.168.30.1
        bridge-ports enp0s31f6.30
        bridge-stp off
        bridge-fd 0


# VLAN 40
auto provider4
iface provider4 inet static
        address 192.168.40.2/24
        gateway 192.168.40.1
        bridge-ports enp0s31f6.40
        bridge-stp off
        bridge-fd 0

# VLAN 50
auto provider5
iface provider5 inet static
        address 192.168.50.2/24
        gateway 192.168.50.1
        bridge-ports enp0s31f6.50
        bridge-stp off
        bridge-fd 0

source /etc/network/interfaces.d/*

Can someone tell me what's happening? It was working fine for almost a week, suddenly there is a problem.

To add more information use a Unifi Dream Machine Pro Max router. Thanks.

 

[shanreich]

You have multiple gateways set in your network configuration, which doesn't work that way. Do I understand correctly that every one of those gateways represents an uplink to a different provider? How do you decide which traffic gets sent via which provider? If you just want to have different VMs using different providers, then you set the VM to the providerX bridge and configure the gateway inside there accordingly and pick one provider for your Proxmox VE host and use that for the host. If you want to do failover across multiple providers for your host / VMs you need some more sophisticated configuration..

 

[Sergiio]

Tienes varias puertas de enlace configuradas en tu configuración de red, lo que no funciona de esa manera. ¿Entiendo correctamente que cada una de esas puertas de enlace representa un enlace ascendente a un proveedor diferente? ¿Cómo decides qué tráfico se envía a través de qué proveedor? Si solo quieres tener diferentes máquinas virtuales que utilicen diferentes proveedores, entonces configuras la máquina virtual en el puente del proveedor X y configuras la puerta de enlace dentro de ella en consecuencia y eliges un proveedor para tu host Proxmox VE y lo usas para el host. Si quieres hacer conmutación por error entre varios proveedores para tu host/máquinas virtuales, necesitas una configuración más sofisticada.

Hello, thank you very much for answering me.
I have 5 internet providers, and since I only have a 1-port network card, I use vlans to reach all 5 routers. So in each vm you configure a provider and this has been working perfectly for several days, now suddenly it seems that I have this problem.

With this configuration, the main proxmox host works with provider1 and the rest of the machines with provider1/provider2/provider3/provider4/provider5 depending on the machine configuration.

What configuration do you recommend for this?

 

[shanreich]

Depends on what you're trying to achieve. But if you are okay with VMs using a specific provider without redundancy then the way you set it up is perfectly fine. I would just remove the gateways (other than 1.1) from the network configuration, since they're unnecessary then and could cause your PVE host traffic to be accidentally routed via another provider.

How did you detect the loop, from your screenshots I cannot really tell that anything of the sort is happening. What symptoms are you seeing / what isn't working that was working before?

 

[Sergiio]

Depends on what you're trying to achieve. But if you are okay with VMs using a specific provider without redundancy then the way you set it up is perfectly fine. I would just remove the gateways (other than 1.1) from the network configuration, since they're unnecessary then and could cause your PVE host traffic to be accidentally routed via another provider.

How did you detect the loop, from your screenshots I cannot really tell that anything of the sort is happening. What symptoms are you seeing / what isn't working that was working before?

Well, I detected this when I saw that proxmox stopped returning my ping. I have deleted all the configuration and done this configuration again and it has started working again. I notice some occasional ping loss. I'm starting to think it's the network card.

As a curiosity, I was previously on Windows, and the Dockers lost connectivity within hours of being turned on. It may be that something wrong is happening to the network card.

And it only happens on this computer, the rest of the network works correctly, although I only use proxmox with vlans on this pc.

Thank you

 

[shanreich]

Then it might be best to check if the same problem occurs with a different network card (don't forget to switch the cable as well!) and see if that resolves the problem.