[SOLVED] Unable to access PVE Management Interface if VMs or CTs are started

[infotechjrf]

Hi Proxmox Community.

I would like to report an issue I've been through since yesterday. Been banging my head ever since!

I'm using 4 tagged VLANs between my Proxmox Baremetal Server (PVE 8.1.3) to my HPE Switch, VLAN 4 is my Production VLAN as well at the VLAN where I intend to access manage my Proxmox Management interface. My server uses an Intel X520-DA2 card, 1st Port connected to a 10G DAC Cable has been in use since the start without any issues at all...

The whole problem started after I attempted to add a 10G RJ-45 Copper SPF+ module and wanted to add it as a Slave Interface to my vmbr0 Linux Bridge. I had the following setup configured like this it prior.

Before: VMBR0 slaves interfaces ---> ENO1 (10G DAC) + ENO4 (Backup 1GB RJ-45 Port on Mainboard) ;

After: VMBR0 slaves interfaces ---> ENO1 (10G DAC) + ENO2 (10G RJ-45 Copper SPF+) + ENO4 (Backup 1GB RJ-45 Port) on Mainboard.

Rebooted server, noticed something very awkward... My PVE Management Interface Port was "pingable" over it's respective IP address for maybe 6-7 consecutive times until it's stops replying right after booting up... Investigating the issue furthermore, I've realized that as soon as a single CT or VM starts, I would loose connectivity to my Proxmox PVE Management interface, but any traffic coming off my CTs or VMs would work as it normally would between them on bridge and over my HPE Switched Network.

I decided to revert my changes and go back to VMBR0 slaves interfaces ---> ENO1 (10G DAC) + ENO4 (Backup 1GB RJ-45 Port on Mainboard) ; But issue keeps persists moving forward.

The only way I've been able to restore PVE Management Interface Access was by doing the following in /etc/network/interfaces ;

1st: Removing IP Address from vmbr0.4 interface. Interface config was the below;

auto vmbr0.4
iface vmbr0.4 inet manual
address: 192.168.4.51/24
gateway: 192.168.4.1
#VLAN 4 - LAN

2nd: Adding / assigning IP to interface using vmbr0v4, I've noticed this interface (vmbr0v4) by using command : ip -br a

auto vmbr0v4
iface vmbr0v4 inet static
address 192.168.4.51/24
gateway 192.168.4.1
#PVE Management Temp Access

////

Would somebody have any suggestions or ideas on how to resolve this issue?

I've been using Proxmox on my Server for quite a while now (Fresh Install happened on PVE 6.3 initially), but I would like to understand what caused this issue and also avoid the need for a reinstall. Would it be possible that the newly introduced SDN feature on PVE 8 is causing my this grief?

My setup is still usable, but I'm afraid it looks more like a temporary solution than a real fix.

Thank for your time,
JR

 

[infotechjrf]

UPDATE : Finally, after doing a fresh install to PVE 8.1.1 ; It turns out that there was some errors in my Network Configurations in my PVE setup itself.

1. I was missing the "VLAN-Aware" setting on my vmbr0 bridge.

2. I had set many different "Linux-VLAN IDs" in my setup for vmbr0, but it seems this configuration wasn't valid anymore following the latest upgrade to PVE 8.1.3. I thought I had to call or flag every single VLAN I would like to get passing through vmbr0, but that wasn't necessary at all. I only needed a Linux-VLAN entry for my PVE Management interface only (VLAN4).

I'm surprised that my previous configurations has been working for the last 3 years. Maybe something changed "Network-wise" in Proxmox 8, which made my current configs deprecated / invalidated from now on.

So this thread is resolved. Hopefully this can help somebody how encounters the same issue as I faced.

Thanks,
JR