We use a systemd-nspawn container to set up a test environment, among others. With F36 this is currently not workable for us because the login process is broken.
I set up a container as usual using dnf to create a container file system, e.g. using
[…]# dnf --releasever=36 --best --setopt=install_weak_deps=False \ --installroot=/var/lib/machines/testf36 install dhcp-client dnf \ fedora-release glibc glibc-langpack-en iputils less ncurses passwd \ systemd systemd-networkd vim-default-editor
I can set a password using
[…]# systemd-nspawn -D /var/lib/machines/testf36 passwd
I create a container with
[…]# systemd-nspawn -D /var/lib/machines/testf36 -b
The container starts fine and finally displays a login prompt. You can enter a name as usual. There follows no password prompt, but after a while it still displays a login prompt again.
The container even keeps running if I exit the login process (is shows up in machinectl).
It’s not a SELinux issue, the system was in permissive mode.
If I start a system service container using […]# systemctl systemd-nspawn@testf36 the container start fine, shows up in machinectl, but a […]# machinectl login testf36 produces the same phenomenon of repeated display of the login prompt without asking for the password. And […]# machinectl shell testf36 results in „Connection to machine testf36 terminated.“
== My questions: == * Does anyone has the same problem * or does anyone use a F36 systemd-nspawn container without such a problems? * Or did I miss a change, so I have to use a different procedure?
Several releases ago we had a similar issue because auf /etc/securetty, but that doesn’t exist anymore.
A F35 container works fine on a F36 host. But a F36 container doesn’t work on a F35 host either.