[SOLVED] Can't connect to proxmox server directly using an ethernet cable from my pc but CAN connect IF proxmox server is connect to my router

[nefero]

Ok, this seems really silly, but I am really out of ideas...

I have a 6 port mini computer that works well with OPNSense but I wanted to use OPNSense in proxmox and replace my old router/switch.

It happens that during its configuration (when it's connected to my router), I can connect via web and SSH normally (and ping). BUT when I try to connect directly from my PC to proxmox, I can't.

Just to create a first summary:

• My PC has a static address of 192.168.1.100/24
• My proxmox ve server is at 192.168.1.4/24 (vmbr0)
• I have 5 of my 6 interfaces in vmbr0 and I can plug my router cable to any of these ports that I will be able to connect from my PC.

This is my /etc/network/interfaces file:

Spoiler

auto lo
iface lo inet loopback

auto enp9s0
iface enp9s0 inet manual
#ETH6

auto enp1s0
iface enp1s0 inet manual
#ETH1 (WAN)

auto enp2s0
iface enp2s0 inet manual
#ETH3

auto eno1
iface eno1 inet manual
#ETH2

auto enp7s0
iface enp7s0 inet manual
#ETH4

auto enp8s0
iface enp8s0 inet manual
#ETH5

auto vmbr0
iface vmbr0 inet static
address 192.168.1.4/24
gateway 192.168.1.1
bridge-ports enp9s0 eno1 enp2s0 enp7s0 enp8s0
bridge-stp off
bridge-fd 0
#Management

And this is the dmesg of my 'official' management port, enp9s0.

Spoiler

dmesg | grep -E "enp9s0|eth0"
[ 1.853173] igc 0000:01:00.0 eth0: MAC: 00:e2:69:12:24:ac
[ 2.467063] igc 0000:09:00.0 enp9s0: renamed from eth5
[ 2.481603] igc 0000:01:00.0 enp1s0: renamed from eth0
[ 5.935146] vmbr0: port 1(enp9s0) entered blocking state
[ 5.935155] vmbr0: port 1(enp9s0) entered disabled state
[ 5.935313] device enp9s0 entered promiscuous mode
[ 9.522794] igc 0000:09:00.0 enp9s0: NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX/TX
[ 9.523995] vmbr0: port 1(enp9s0) entered blocking state
[ 9.524025] vmbr0: port 1(enp9s0) entered forwarding state
[ 89.350105] igc 0000:09:00.0 enp9s0: NIC Link is Down
[ 89.350379] vmbr0: port 1(enp9s0) entered disabled state
[ 98.815072] igc 0000:09:00.0 enp9s0: NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX
[ 98.924857] vmbr0: port 1(enp9s0) entered blocking state
[ 98.924894] vmbr0: port 1(enp9s0) entered forwarding state
[ 258.264974] igc 0000:09:00.0 enp9s0: NIC Link is Down
[ 258.265229] vmbr0: port 1(enp9s0) entered disabled state
...

*I remember that we used to mind patch and cross-over cables when connecting two devices from the same layer, but I don't think it applies nowadays as I can connect directly IF I install OPNSense instead of proxmox, but MAYBE, there is something in the OS I should mind?

Anyway, I am not using fancy VLAN or anything like this. The most fancy thing I tried, was to use a different bridge to use the VMs but, ultimately, I wanted to:

• Be able to access the management from any interface as this mini pc will be my router/switch (using a virtualized OPNSense instead);
• Launch an OPNSense VM,assign it to a bridge so I can connect directly from my PC independent of the port I am connecting phisycally
• Launch a pi-hole CT, assign it to a bridge so I can connect directly from my PC independent of the port I am connecting physically;

I am thinking I am doing something very unusual as I couldn't see any similar scenario to mine, but still, I would really appreciate if I can get any insight from you.

Thank you!

 

[nefero]

Sorry, my bad.
Apparently it was ARP. I tested with another computer, and it worked.
After I cleaned ARP in my Windows PC with 'arp -d *' it started working.