Bridge won't start because of bridge-ports issues

Marathon

Member
Apr 7, 2019
40
1
6
54
I have a server with my WAN interface for my Proxmox server where I also have routed subnet on for my VM's.

The issue I run into is that this works for the IP's fine but I have trouble with the bridge-ports as a bridge doesn't start without any but also complains when it doesn't exist of you use "none" twice, so that is why I now use none1 and none2

Am I overseeing something here ?

Code:
auto lo
iface lo inet loopback
iface lo inet6 loopback

auto eth0
iface eth0 inet static
        address  172.16.0.1
        netmask  255.255.255.0
        gateway  172.16.0.254

auto eth1
iface eth1 inet manual

auto eth2
iface eth2 inet manual

auto bond0
iface bond0 inet manual
        bond-slaves eth1 eth2
        bond-miimon 100
        bond-mode balance-rr

auto vmbr11
iface vmbr11 inet static
        address  10.0.0.1
        netmask  255.255.255.0
        bridge-ports none1
        bridge-stp off
        bridge-fd 0

auto vmbr101
iface vmbr101 inet manual
        bridge-ports bond0
        bridge-stp off
        bridge-fd 0

auto vmbr1001
iface vmbr1001 inet manual
        bridge-ports none2
        bridge-stp off
        bridge-fd 0
        bridge-vlan-aware yes
        bridge-vids 2-4094
 
I have a server with my WAN interface for my Proxmox server where I also have routed subnet on for my VM's.

The issue I run into is that this works for the IP's fine but I have trouble with the bridge-ports as a bridge doesn't start without any but also complains when it doesn't exist of you use "none" twice, so that is why I now use none1 and none2

Am I overseeing something here ?

Code:
auto lo
iface lo inet loopback
iface lo inet6 loopback

auto eth0
iface eth0 inet static
        address  172.16.0.1
        netmask  255.255.255.0
        gateway  172.16.0.254

auto eth1
iface eth1 inet manual

auto eth2
iface eth2 inet manual

auto bond0
iface bond0 inet manual
        bond-slaves eth1 eth2
        bond-miimon 100
        bond-mode balance-rr

auto vmbr11
iface vmbr11 inet static
        address  10.0.0.1
        netmask  255.255.255.0
        bridge-ports none1
        bridge-stp off
        bridge-fd 0

auto vmbr101
iface vmbr101 inet manual
        bridge-ports bond0
        bridge-stp off
        bridge-fd 0

auto vmbr1001
iface vmbr1001 inet manual
        bridge-ports none2
        bridge-stp off
        bridge-fd 0
        bridge-vlan-aware yes
        bridge-vids 2-4094


the /etc/network/interfaces entry for e.g. vmbr11 is correct (apart from "none1", should be none), if the interface is not up after reboot something is not configured properly in your system.
 
Hi,

The same goes for vmbr1001 then, I can remove them and see no errors with ifreload -a (ifupdown2) but my tapinterfaces, or at least my VM's are not reachable after the reload.

Any idea about that ?

Thanks so far!

/edit because of update

Proxmox adds 'bridge-ports none' to a bridge where you don't set a bridgeport for. And if you add multiple like that, if updown complains about none not there or only can be used once.

ifreload works when the birdge-ports are not set, otherwise it errors and the tapinterfaces are not attached anymore

So actually this is a proxmox-bug, just don't add the bridge-ports line at all when you set nothing there!
 
Last edited:
Hi,
I'm trying to reproduce here, but I don't see any problem with reload.
tap/veth interfaces are still in vmbrX after a networking reload (brctl show vmbrX).

I'm only seeing an error in "ifreload -a"
error: 'NoneType' object has no attribute '__getitem__'
error: 'NoneType' object has no attribute '__getitem__'

but this is not reload to "bridge-port none", this is because we have a patch in ifupdown2 to not handle tap/veth interfaces not defined in bridge, and I think it's missing a check in some part of reload. (and python code think that tap|veth object exist in ifupdown2 cache, and throw an error on this).
I'll fix that, but I don't think they are any impact.

Is is this error message that you seen on reload ?


Edit:
I'm also seeing a breaking, if a bridge-xxxx is changed, the vms/ct tap/veth interfaces are not in bridge after reload. (taking default bridge-ports ...) . I'll check that.
 
Last edited:
Good we are making some progress, nice you checked this out.

The error I get is the following:

# ifreload -a
error: vmbr11: bridge port none1 does not exist

After that non of the hosts are pingable anymore and I need to bring down/up the hosts for a working tap again.
 
Hi,

could you verify in
/etc/network/ifupdown2/ifupdown2.conf

the value of
ifreload_down_changed.

and try to setup it to 1 if it's 0.

ifreload_down_changed=1

and use "bridge-ports none" in your config. (not none1)


the thing to check, is after a reload, do a "brctl show", and check if tap interfaces are still in the bridge.
 
Hi,

Thanks so far!

I made the setting and did a reload but the same issue occures, same error, even on none. I changed that for the setup as I hoped it checked like you can also bridge only once to an interface and none was used multiple times, I now use it only on one bridge

Code:
# brctl show
bridge name     bridge id               STP enabled     interfaces
vmbr1001                8000.6cb3113c80a6       no              bond0
vmbr11          8000.000000000000       no

It now errors on vmbr11 as that one has no bridge interface, so is set to none and vmbr1001 doesn't give a ping anymore as well.

Does this help ?
 
I have build an updated version of ifupdown2, with changes to handle tap|veth interfaces,

http://odisoweb1.odiso.net/ifupdown2_1.2.5-1+pvetest1_all.deb

can you test it ?


Also, about your configuration, do you have now:

auto vmbr1001
iface vmbr1001 inet manual
bridge-ports none
bridge-stp off
bridge-fd 0
bridge-vlan-aware yes
bridge-vids 2-4094


auto vmbr11
iface vmbr11 inet static
address 10.0.0.1
netmask 255.255.255.0
bridge-ports none
bridge-stp off
bridge-fd 0

?

are can you post your vms configuration ? (/etc/pve/qemu-server/<vmid>.conf)


Edit: for bridge vlan aware, it seem than vlan tag is lost on ifreload
(#bridge -c vlan to verify )

Edit: Ok, I have fixed it, and reuploaded again
http://odisoweb1.odiso.net/ifupdown2_1.2.5-1+pvetest1_all.deb

I have tested with/without none, with/without bridge vlan aware, and I don't have breakeage anymore.
Also, it's really need "ifreload_down_changed=0" in ifupdown2.conf
 
Last edited:
  • Like
Reactions: KORG
Hi,

I have tested it and it seems to work perfectly now, no breakage here as well anymore, thanks for that!

The only message I got was:

Code:
# ifreload -a
error: /etc/network/interfaces: line36: iface bond0: unsupported keyword (bond-primary)
error: /etc/network/interfaces: line37: iface bond0: unsupported keyword (bond-primary-reselect)

I thought that was supported (and works).
 
Hi,

I have tested it and it seems to work perfectly now, no breakage here as well anymore, thanks for that!

Great :)

patches have been applied in proxmox git and official package should be available soon.

Thanks again for the report and tests.


The only message I got was:

Code:
# ifreload -a
error: /etc/network/interfaces: line36: iface bond0: unsupported keyword (bond-primary)
error: /etc/network/interfaces: line37: iface bond0: unsupported keyword (bond-primary-reselect)

I thought that was supported (and works).

seem than bond-primary is not yet implemented in ifupdwon2
https://github.com/CumulusNetworks/ifupdown2/issues/9

here a work around

Code:
auto bond0
iface bond0 inet dhcp
        bond-slaves eth0 wlan0
        bond-mode active-backup
        up echo eth0 > /sys/class/net/$IFACE/bonding/primary
 
OK, everything works so far but traffic between vmbr11 and eth0 stops floating after a ifreload -a.

On the proxmox machine itself I can ping to the outside world and the IP's on tap interfaces of vmbr11 and vmbr11 itself but not over eth0 anymore, so form the outside world.

Maybe I'm missing something but you might have an idea as well ?
 
OK, everything works so far but traffic between vmbr11 and eth0 stops floating after a ifreload -a.

On the proxmox machine itself I can ping to the outside world and the IP's on tap interfaces of vmbr11 and vmbr11 itself but not over eth0 anymore, so form the outside world.

Maybe I'm missing something but you might have an idea as well ?

Hi

what do you mean by traffic between eth0 and vmbr11 ? (eth0 is not in bridge, and they are on differents subnets)
 
can you verify that ip_foward is still enabled after reload ?

cat /proc/sys/net/ipv4/ip_forward

?

also, do you have any kind of static routes in /etc/network/interfaces ?

I need to check, not able at the moment, sorry for that and inform you later about it!

I don't have any static routes in the interfaces file (that is where routers are for ;))
 

About

The Proxmox community has been around for many years and offers help and support for Proxmox VE, Proxmox Backup Server, and Proxmox Mail Gateway.
We think our community is one of the best thanks to people like you!

Get your subscription!

The Proxmox team works very hard to make sure you are running the best software and getting stable updates and security enhancements, as well as quick enterprise support. Tens of thousands of happy customers have a Proxmox subscription. Get yours easily in our online shop.

Buy now!