As I mentioned in the upstream systemd bug, it's still not clear what exactly is causing this, and the inability to reproduce it after the image first boot, or with another image with debug added, or with the latest image, means it will be impossible to verify any fix to this problem.
However, my best guess is udevd is somehow missing the 'add' uevent for the network device, as I described in the upstream bug. While I don't know why it would miss the event, the upstream commit e0e789c1e97e2cdf1cafe0c6b7d7e43fa054f151 should fix the problem, as it will setup the proper udevd device parameters not only for 'add' but also 'change' uevents.
So, based on those assumptions/guesses and on the relative safety of the upstream commit, I plan to SRU that commit for this bug.
As I mentioned in the upstream systemd bug, it's still not clear what exactly is causing this, and the inability to reproduce it after the image first boot, or with another image with debug added, or with the latest image, means it will be impossible to verify any fix to this problem.
However, my best guess is udevd is somehow missing the 'add' uevent for the network device, as I described in the upstream bug. While I don't know why it would miss the event, the upstream commit e0e789c1e97e2cd f1cafe0c6b7d7e4 3fa054f151 should fix the problem, as it will setup the proper udevd device parameters not only for 'add' but also 'change' uevents.
So, based on those assumptions/guesses and on the relative safety of the upstream commit, I plan to SRU that commit for this bug.