Oracle and Centos 8 Images - NetworkManager not able to manage macvlan Device

LXD
1

There may be nothing the LXD folks or the team that creates the images for Oracle/Centos 8 can do about it, but I am unable to get a stock Centos8/Oracle8 (cloud or vanilla) image booting to usable network with or without cloud-init.

Seems I’m hitting this bug:

https://bugzilla.redhat.com/show_bug.cgi?id=1832603

lxd version 4.9

profile for container:

config:
  boot.autostart: "true"
  snapshots.expiry: 2d
  snapshots.schedule: 0 */6 * * *
description: Default LXD profile
devices:
  eth0:
    name: eth0
    nictype: macvlan
    parent: enp4s0f0
    type: nic
  root:
    path: /
    pool: lpool
    type: disk
name: oracle

[root@oracle ~]# nmcli
lo: unmanaged
        "lo"
        loopback (unknown), 00:00:00:00:00:00, sw, mtu 65536

eth0: unmanaged
        "eth0"
        macvlan, 00:16:3E:8E:7E:B9, sw, mtu 1500

lxc exec oracle bash and using the following commands:

nmcli dev set eth0 managed yes
nmcli con up 'System eth0'

results in:
Error: Connection activation failed: No suitable device found for this connection (device lo not available because device is strictly unmanaged).

2

Yes there was a fix for veth devices added to our images recently see CentOS 8 Stream container not assigned IPv4/IPv6 addresses

However there is some ongoing discussion in that thread about getting it to work on a Ubuntu bionic host.

Also I note you are actually using macvlan and not veth and for this case there is an issue open to address this (hopefully it will be similar to the fix for veth).

See

1 Like