Re: [Libguestfs] [v2v PATCH 4/4] convert_linux: install the QEMU guest agent with a firstboot script

On 06/07/22 14:59, Richard W.M. Jones wrote: After some investigation: - NetworkManager and systemd-networkd-wait-online seem mutually exclusive, and switching from the former to the latter (which is obviously only possible on a systemd-based system) takes (mostly) one "systemctl disable" and another "systemctl enable" command: https://fedoraproject.org/wiki/Cloud/Network-Requirements#Switching_from_... - When one of { NetworkManager, systemd-networkd-wait-online } is *disabled* (not running), and we invoke the "wait online" utility for that (disabled) service, the utility will wait (uselessly) for the timeout, and then report failure / timeout. I've tested this with "systemd-networkd-wait-online" on my F35 workstation (where NetworkManager is enabled -- the default), and with "nm-online" in a RHEL9 guest, where I could disable NetworkManager without risks. - This useless waiting (for NetworkManager that's not running) can be disabled with the "-x" switch for nm-online, but I found no corresponding switch for "systemd-networkd-wait-online". What works is "systemctl is-active systemd-networkd", which is reasonable, because "systemd-networkd" already depends on systemd. Additionally the default timeout for nm-online seems to be 30 seconds already. So ultimately I believe we should do: nm-online -x -q || ( systemctl -q is-active systemd-networkd && /usr/lib/systemd/systemd-networkd-wait-online -q --timeout=30 ) If the final exit status is nonzero, I think that's not a problem for the firstboot script. The whole command seems to make sense also when nm-online, systemctl, or both, are missing (a missing command results in exit status 127). Thanks Laszlo