On Tue 05 Jan 2016 at 16:14:51 +0000, Neil Williams wrote: > On Tue, 5 Jan 2016 15:20:41 +0000 > Brian Potkin <[email protected]> wrote: > > > On Tue 05 Jan 2016 at 12:52:34 +0100, Martin Pitt wrote: > > > > > Hello again, > > > > > > Martin Pitt [2016-01-05 12:43 +0100]: > > > > How is this? > > > > > > > > > > > > http://anonscm.debian.org/cgit/pkg-systemd/systemd.git/commit/?id=957002f > > > > > > > > > > It's obviously wrong, thanks to Marco for pointing out :-/ > > > Fixed harder in > > > > > > > > > https://anonscm.debian.org/cgit/pkg-systemd/systemd.git/commit/?id=f90b82c > > > > > Thanks, that looks good. > > > Is it worth mentioning that changes to interface rules > > in /etc/udev/rules.d need update-initramfs > > -u? > > The commit above does that: > + and rebuild the initrd with "update-initramfs -u". > > The reason why the symlink failed in vmdebootstrap was precisely > because the initramfs had not been rebuilt.
The advice to use "update-initramfs -u" is aimed completely at the symlink failure. But consider this rule in /etc/udev/rules.d: SUBSYSTEM=="net", ACTION=="add", ATTR(address)=="00:40:f6:77:3d:e1", NAME="eth-xxx" The user will have "eth-xxx" as the interface name. A short time later there is a kernel update. At some time after that the user user deletes the file containing the rule and reboots without doing "update-initramfs -u". The interface is still named "eth-xxx". Regards, Brian.
