[systemd-devel] Debugging systemd in a container

[Mark Clarkson]

Hi!
I have been writing a tool that creates kubernetes clusters in
containers. I had just about finished when I upgraded to Fedora 32
yesterday and it dropped docker and moved to cgroupv2.

So, I have Podman and Cgroupv2 now.

Using a centos 7 container with systemd, I cannot get systemd to start.

The container starts and /usr/sbin/init is running, but no dbus, or
systemd - only init.

For the container a script starts and sets machine ID, product name,
product uuid, mounts /sys /dev etc. It all worked with cgroupv1, but
now with the new cgroupv2, it doesn't.

I have been trying to debug this but I can't get any logs out of
systemd.

I can add cmdline options to systemd. I currently have the following in
the Dockerfile:

ENTRYPOINT ["/usr/local/bin/entrypoint", "/usr/sbin/init",
"systemd.unified_cgroup_hierarchy", "debug"]

I also start the container with:

podman run -d \
  -v /sys/fs/cgroup:/sys/fs/cgroup:ro \
  --privileged \
  --tmpfs /run \
  -e container=docker \
  --tmpfs /tmp \
  c7-systemd

I would really like to get some information out of systemd to work
through the errors, but I can't work out how to do it. I expected it to
output to stdout and be accessible through 'podman logs <ID>'.

This was another line I tried:

ENTRYPOINT ["/usr/local/bin/entrypoint", "/usr/sbin/init",
"systemd.showstatus", "systemd.default_standard_output=tty",
"systemd.unified_cgroup_hierarchy", "debug"]

Can anyone point me in the right direction? Of course, 'journalctl -xe'
says:

No journal files were found.
-- No entries --

I need logs!! :)

Thanks!
Mark