Comment 11 for bug 1869796

So we start&restart plenty of sockets (not just libvirtd-ro.socket), and all others are happy.
So what is wrong/special with libvirtd-ro.socket?

Comparing the normal with the -ro socket:

one can be started without error:
root@f2:~# deb-systemd-invoke start libvirtd.socket
root@f2:~# deb-systemd-invoke start libvirtd.socket
root@f2:~# deb-systemd-invoke start libvirtd.socket

But the other one fails on the same:
root@f2:~# deb-systemd-invoke start libvirtd-ro.socket
Job failed. See "journalctl -xe" for details.
root@f2:~# deb-systemd-invoke start libvirtd-ro.socket
Job failed. See "journalctl -xe" for details.
root@f2:~# deb-systemd-invoke start libvirtd-ro.socket
Job failed. See "journalctl -xe" for details.

But with a restart I could break the formerly working one:
root@f2:~# deb-systemd-invoke restart libvirtd.socket
Job failed. See "journalctl -xe" for details.

And from that moment on:
root@f2:~# deb-systemd-invoke start libvirtd.socket
Job failed. See "journalctl -xe" for details.

So it seems as soon as we do "restart" we can no more "start".
That is reasonable, the socket is up then the service is up.
And a "start" just makes it "no-op as it is there" but the restart stops it.