wired ethernet disabled between reboots, was: when startup delays become bugs

[Chris Murphy]

On May 21, 2013, at 5:14 PM, Dan Williams <dcbw at redhat.com> wrote:

> On Tue, 2013-05-21 at 14:33 -0600, Chris Murphy wrote:
>> 
>> Is Gnome > Settings > Network leveraging Network Manager? If so, it's using a different naming convention than systemd, and it seems the two are involved in some non-deterministic confusion.
> 
> Yes, it does.  What's likely happening is that your saved network
> connections from before F19 have the name they were originally given,
> like "System p5p1".

No, in all cases these are completely new installs on baremetal. The effect isn't reproducible in a VM where I always get a p2p1 device whether booting from install media or from the installed OS. On baremetall ip addr when booted from live media is en5s0 (something like that) and when booted from the installed system it's p5p1 or p5p2. I don't have access to that hardware for another couple of weeks.d




>  If the ifcfg file has no "NAME=xxx" key, the name
> is taken directly from the ifcfg file name.  I bet you'll find you've
> got /etc/sysconfig/network-scripts/ifcfg-p5p1 on your system with HWADDR
> matching your en5s0 interface.
> 
> That name is never changed after the connection config is created; NM
> does not and will never rename your configuration files underneath you.

The configuration files weren't renamed. The ip addr  or ip link reported device is what's different. The effect is that when the device is p5p1, yet there's a configuration file created by anaconda ifcfg-en5s0, even though that file has a line ONBOOT=yes, the wired connection wasn't automatically being enabled. But again, even I'm not understanding the intended behavior so part of this is learning and testing and the ensuing confusion.


Chris Murphy