On 2019/12/05 11:34, Michael Biebl wrote: [...] > > The old 73-usb-net-by-mac.rules file contained: > > TEST!="/etc/systemd/network/99-default.link", \ > > This check can't be expressed via .link files. > The new 73-usb-net-by-mac.link link file does respect the net.ifnames=0 > kernel command line parameter, though. > > I don't see a good way of fixing this with .link files, aside from > documenting this change. We could say that users have to symlink > 73-usb-net-by-mac.link as well (or symlink 80-net-setup-link.rules, > which does the actual renaming).
I think documenting that 73-usb-net-by-mac.link needs to be symlinked as well is the best approach. It diverges from the upstream instructions in eg. https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames/#idontlikethishowdoidisablethis but then, Debian diverges from upstream by adding 73-usb-net-by-mac.rules or .link. > > The other alternative would be to revert the changes from #941636, but I > kinda like the new .link based approach and would prefer to keep it. > > Thoughts? Like I pointed out in 941636, that simple TEST for the existence of /etc/systemd/network/99-default.link is too basic and can lead to counterintuitive behavior as well: > This is an incomplete way of reimplementing udev functionality. In > particular, the user will get a different behavior if they do > $ cp /lib/systemd/network/99-default.link /etc/systemd/network/ > even though the naming rules are effectively the same. So I don't think it was good.
signature.asc
Description: PGP signature
