vmbr switches doesnt work after restart network service

[student]

Hello to everyone,
when I restart the host, the network works fine, but when I make a change (for example: I add port forwarding for some VM) and when I restart network service on the host, then suddenly the communication on vmbr is lost, after which I restart the host and the network again starts to work properly. Has anyone had a similar problem? Do you have any advice on how to do this without restarting the entire host?

 

[Stoiko Ivanov]

This is (unfortunately) expected behaviour.
* When restarting the network service on your host the vmbr interface gets removed and created anew.
* Through this the associations of the guests' tap interfaces to the bridge are lost and not recreated.
* The recommended way for making network-changes is to change the config and reboot the node afterwards (that way you also know that the changes survive a reboot

PVE ships (since roughly 6 months) `ifupdown2` optionally. It is (for the greatest part) a drop in replacement for `ifupdown`, but adds quite a bit of functionality. One new feature is the tool `ifreload`, which reloads the interfaces without losing the association - you could try that

hope this helps!

 

[student]

Hello, I solved the problem. What was trick:
- when you add new vmbrX or change some options on current vmbrX, configuration is entered into file "interface.new"
- than you should do: cp interface.new interfaces
- next: rm interfaces.new (to avoid messages in GUI panel)
- finally: ifup vmbrX (if necessary, you should type ifdown vmbrX before ifup command)

ATTENTION: NEVER RESTART NETWORK SERVCIE because of #Stoiko Ivanov explanation .

Cheers!

 

[iparaskev]

Excuse me if i bring alive this thread again. Please advise me if i should have done something different.
New to proxmox, i came up to this problem today.

• Having installed version 6.3-3, realized that the workaround with the interface.new file does not apply to this version.
• Have not installed ifupdown2 in order to test ifreload (a little bit afraid that there might be a slight chance to loose connectivity if something goes wrong, but i will test it in a dev enviroment).
• Obviously did not want to reboot
  
  What i did, was to go to the hardware settings of the VMs and remove the network device. Then i added a network device again with the same settings (i guess this deleted the old tap and created a new one). Inside the VM, from the console, i did a networking restart in the VM and everything was working fine again.
  
  Could this workaround create any problems to the future regarding the VMs or the node?