libvirt net ostestbm won't start after reboot

[SirPhuttel]

After a reboot of the hypervisor machine, libvirt's 'ostestbm' network refuses
to start for several reasons:

• Interface ostestbm exists already
• External interface's accept_ra != 2

Observed on CentOS8.4.2105. To reproduce:

• Standard install, the only customization I did was to set 'IP_STACK=v4v6'
• reboot
• virsh net-start ostestbm

I could eliminate the first cause above by removing ifcfg-ostestbm file.
AFAICT, it is not needed as libvirt creates the bridge by itself anyway. It
works like this with virbr0, for instance.

I spent a while trying to change ifcfg-eno3 (the external interface in my case)
so that ifup-ipv6 sets accept_ra=2 but failed. Eventually just settled with

echo 'net.ipv6.conf.eno3.accept_ra = 2' >/etc/sysctl.d/01-eno3-accept_ra.conf

With these two changes applied, all three libvirt networks start upon boot. The
nodes still need to be started manually, though. Maybe make them autostart?
('virsh autostart ' should do)

[lalan7]

I have the same issue. Can you share how to recover from this situation. I try the echo 'net.ipv6.conf.eno3.accept_ra = 2' >/etc/sysctl.d/01-eno3-accept_ra.conf did'nt work.

[SirPhuttel]

Erik, the described measures were sufficient for me. The external device name (eno3 in my case) is hardware-specific, are you sure your external interface has the same name?
Apart from that, I won't be much help I fear - I had the problem two years ago and I don't work with dev-scripts anymore. So, good luck? ;)