Le 30/05/2017 à 21:46, Branko Majic a écrit : > On Tue, 30 May 2017 21:23:58 +0200 > Emmanuel Kasper <m...@debian.org> wrote: > >> Than you for the detailed bug report. >> >> I understand the unstable network card ordering can perturbate the >> network config if the NIC is assigned the name eth1 instead of eth0. >> But I am wondering why is that changing the type of the second NIC >> make the device ordering stable. >> Are you sure this fixes the issue ? > > I've been thinking about it a bit. It could be a "race" between which > kernel module gets initialised first? This is highly hypothetical from > my side, mind you, just one thing that pops to my mind. I would need to > read up quite a bit on the topic to be sure.
Comparing the output of dmesg | grep between multiple boots it might very well be that, I was able to reproduce the bug. I have to think about the best way to solve this. In theory udev persistent device naming should be here for this kind of issues, but I don't use it since I build via qemu, and the devices on Virtualbox get another position in the pci bus. I will have a thought on using udev rules or setting all NICS to the same hardware type.
