Ubuntu Server 22.04 LTS no DHCP on startup

I am wondering if there has been a regression with Ubuntu. Back when 18.04.3 came out, I noticed that my VM-based Ubuntu servers would no longer be able to get DHCP addresses from my Windows DHCP servers. I posted about it here.

21.10 seemed to resolve the issue, but of course, it is no longer supported. I created some more VMs with 22.04.3 and they now have the same issue. Stock install - everything seems to work, until you reboot. Then you get a 2-minute timeout and no network until you run sudo dhclient.

If I run ip a it shows the ens160 interface with an altname of enp3s0 (that's new) but no IP.

I've modified my /etc/netplan/00-installer-config.yaml to the following:

network:  renderer: networkd  ethernets:    enp3s0:      dhcp-identifier: mac      dhcp4: true      optional: true  version: 2

But it still won't connect until I manually get an IP after a reboot until I do sudo dhclient. I also tried using the original ens160 in place of enp3s0 with no effect.

sudo dmesg | grep dhcp shows the following entries:

[    7.568386] audit: type=1400 audit(1668726304.016:9): apparmor="STATUS" operation="profile_load" profile="unconfined" name="/usr/lib/NetworkManager/nm-dhcp-client.action" pid=742 comm="apparmor_parser"[    7.568396] audit: type=1400 audit(1668726304.016:10): apparmor="STATUS" operation="profile_load" profile="unconfined" name="/usr/lib/NetworkManager/nm-dhcp-helper" pid=742 comm="apparmor_parser"

I know there was an AppArmor bug that caused this behavior before, and that was where the evidence was. I don't know enough about the guts of Ubuntu to say if this is normal output or not, so I have included it.

Any help is much appreciated! I'd love to be able to have new VMs not need a systemd workaround, that manually runs dhclient at startup, which is the workaround I'm using.