help with lxc networking

[coolmike8902]

I made a custom configuration in order to have a larger storage pool. Here is the profile I made:

config: {}
description: B970Pro LXD profile
devices:
eth0:
nictype: bridged
parent: lxdbr0
type: nic
root:
path: /
pool: B970Pro
type: disk
name: B970Pro
used_by:
- /1.0/instances/handy-pipefish
- /1.0/instances/faithful-bedbug
- /1.0/instances/stable-ostrich

I am able to ping the instances from the host, but not outside of the host. If I perform: lxc network show lxdbr0, I get this:
config:
ipv4.address: 10.93.5.1/24
ipv4.nat: "true"
ipv4.routes: 10.122.135.135/32
ipv6.address: fd42:22b0:3deb:3bc6::1/64
ipv6.nat: "true"
description: ""
name: lxdbr0
type: bridge
used_by:
- /1.0/instances/handy-pipefish
- /1.0/instances/stable-ostrich
- /1.0/profiles/B970Pro
- /1.0/profiles/default
managed: true
status: Created
locations:
- none

Can someone please help me troubleshoot this issue?

Thanks.

[TheFu]

please edit and wrap output in code-tags to retain indentation. Make it easier for someone to help you, if they can. Show the exact commands used - many people don't touch this stuff often enough to remember how to show details.

I've never used lxdbr0 nor have I added any ipv[46].* settings for LXD.

Since I'm using a separate bridge, not the one created, and I seldom touch any networking once the lxd init is done. It is listed as unmanaged. Doubt I can help your question. A few years ago, I set the defaults to use a bridge I'd manually created and never touched it again. Inside each container, setup a static IP. I'm running 7 lxc containers that way. Also, I disable all IPv6 on my hosts. I started doing this when IPv6 started being enabled by default. After attending a few IPv6 training classes, I decided I didn't know how to secure it, so it was best to just disable it for my needs.

[ian-weisser]

LXD bridging is a two step process.

1) Netplan (create the bridge and connect it to the real LAN)
2) LXD Profile (use the bridge)

If you skip step 1, then LXD creates a bridge which cannot talk outside of the host because it's not connected to the LAN.

I wrote an example of a simple working setup last week at https://ubuntuforums.org/showthread....2#post14188192

[coolmike8902]

Forgive the late reply. I will try and figure out how to do that. Thanks.

[coolmike8902]

I had it working, and now it doesn't. I believe I have followed your instructions, so I'm at a loss. Here's my host netplan:

network:
  version: 2
  renderer: networkd
  ethernets:
    enp7s0:
      dhcp4: no
    eno2:
      dhcp4: no
    eno1:
      dhcp4: no
    enp9s0f1:
      dhcp4: no
    enp9s0f0:
      dhcp4: no
  bridges:
    bridge0:
        addresses:

and then the profile I'm using:

config: {}
description: B970Pro LXD profile
devices:
  eth0:
    name: eth0
    nictype: bridged
    parent: bridge0
    type: nic

and lastly the netplan on the container:

network:
    version: 2
    ethernets:
        eth0:
            dhcp4: true

Am I missing something? Thanks.

[Tadaen_Sylvermane]

network:
  version: 2
  renderer: networkd
  ethernets:
    enp7s0:
      dhcp4: no
    eno2:
      dhcp4: no
    eno1:
      dhcp4: no
    enp9s0f1:
      dhcp4: no
    enp9s0f0:
      dhcp4: no
  bridges:
    bridge0:
        addresses:

You need to assign the bridge to interfaces, and give it a static ip if you need. Here is my netplan from my home server. Use it as a template / example as needed.

network:
 version: 2
 renderer: networkd
 ethernets:
  enp4s0:
   dhcp4: false
   dhcp6: false
 bridges:
  br0:
   dhcp4: false
   dhcp6: false
   interfaces: [enp4s0]
   addresses: [192.168.1.2/24]
   gateway4: 192.168.1.240
   nameservers:
    addresses: [8.8.8.8, 8.8.4.4]

Once you have the bridge properly configured on the host you can re-run

lxd init

and specify the bridge, in my example

br0

At this point they will be on the lan like any other physical machine.

[TheFu]

My stance is, if it is important enough to have a container, then it is important enough to have a static IP.
I'd never use DHCP for a container.