So I am testing with Fedora-Cloud-Base-AmazonEC2.aarch64-Rawhide-20250826.2 which has cloud-init-25.2-4.fc44.noarch and it is a bit better, at least that it how it looks like from our integration tests, but I still see some connection issues where sshd does not start.

In journal I see for example (here sshd did actually start):

Aug 27 02:19:35 localhost systemd-journald[503]: Runtime Journal (/run/log/journal/ec25151dd390da871f0b77c632b18850) is 8M, max 75.6M, 67.6M free. Aug 27 02:19:35 localhost systemd[1]: [🡕] sysinit.target: Found ordering cycle: systemd-update-utmp.service/start after auditd.service/start after cloud-init-local.service/start after cloud-> Aug 27 02:19:35 localhost systemd[1]: [🡕] sysinit.target: Job systemd-update-utmp.service/start deleted to break ordering cycle starting with sysinit.target/start Aug 27 02:19:35 localhost systemd[1]: [🡕] sysinit.target: Found ordering cycle: auditd.service/start after cloud-init-local.service/start after cloud-init-main.service/start after dbus.socke> Aug 27 02:19:35 localhost systemd[1]: [🡕] sysinit.target: Job auditd.service/start deleted to break ordering cycle starting with sysinit.target/start