auto upgrade unintentionally updates /etc/udev/rules.d/70-persistent-net.rules

[Oliver Grawert]

i am using a statically set up interface through the new snappy config ubuntu-core mechanism, setting the detected eth0 device to a static IP and GW ...

my snappy install did an auto-upgrade over night and considered that my eth0 device needs a new name, so it added the same device line that already exists in /etc/udev/rules.d/70-persistent-net.rules with a new name (eth3) ... this should indeed never happen, the machine is not reachable anymore now after it tried to boot into the newly upgraded system-b partition ...

[Oliver Grawert (ogra)]

note in syslog:
Sep 14 09:16:01 aleph2 kernel: [ 43.590490] r8169 0000:01:00.0 eth3: renamed from eth0

[Martin Pitt (pitti)]

At first sight this sounds like https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=765577 which got fixed in http://anonscm.debian.org/cgit/pkg-systemd/systemd.git/diff/?id=f9cb380a . I . e. it's fixed in Debian jessie and sid, and Ubuntu wily (via completely removing this rules generator), but Ubuntu vivid already has that fix as well.

[Martin Pitt (pitti)]

It's very suspicious that the kernel gets written twice. Can you please check if this is yet another case of the boot racing with any potential bind-mounting of /etc/udev/rules.d or the 70-persistent-net.rules file?

[Martin Pitt (pitti)]

IOW, don't bind-mount /etc/udev/rules.d/ in the running system -- you have to mount it (and in general, everything in /etc) in initramfs.

[Leo Arias (elopio)]

Hello ogra. This says it's fix released for 15.04. I suppose it's in master too, but I wonder why it still says work in progress. Do you know if there's some work missing for fixing this issue?

[Michael Vogt (mvo)]

We no longer have 70-persistent-net.rules in 16