Re: [PATCH] bonding: Prefix bonding connections with "Bond" and append slave suffix

[Thomas Graf <tgraf redhat com>]

commit 9dd270966edb294452eadac34ec48615f31a851a
Author: Dan Williams <dcbw redhat com>
Date:   Thu Nov 17 18:05:35 2011 -0600

    ifcfg-rh: revert some bits of a2b1afb1e41805f3e0bf7e7c956c65d8374f1210

    Revert the DEVICE and suffix bits for the connection name; there's
    a few problems with this.  It adds the DEVICE value for connections
    regardless of what type they, even in cases where it's not hugely
    useful (ie basic wired).  We used to do this, but stopped doing it
    because it has zero relevance to a large number of users.  Instead,
    the UI itself should do this where appropriate.  That probably means
    that 'nmcli' and other tools should give more information about
    the components of a connection (like a slave device's master) and
    GUI tools would show that in detailed connection information but
    not in the at-a-glance status or tooltips.  Second, if more
    more advanced users wish this information to show up in the name
    they can always set the name themselves, or name the ifcfg file
    something like "ifcfg-bond1-slave-of-eth0" too.

I'm OK with that although even with this is very confusing. Assume the
following situation:

[...]/ifcfg-eth0:
  DEVICE=eth0
  HWADDR=MAC_1

[...]/ifcfg-eth1:
  DEVICE=eth1
  HWADDR=MAC_2

Given eth0 and eth1 use the same driver (which is very common especially
in virt guests) the order of eth0 and eth1 may change frequently, i.e.
eth0 will have MAC_2 and eth1 will have MAC_1.

What happens in this situation is that we name the connection
"System eth0" which get auto connected to the device eth1 and connection
"System eth1" which gets auto connected to eth0. This is especially bad
because the name "System" indicates that we are referring to the kernel
device name.

I realize that when people properly rename their ifcfg files and use
TYPE=ethernet instead of relying on DEVICE=eth+ for detection this can
be avoided but the above is still pretty much the default way of
configuring interfaces.

I guess we could step away from using the "System <ifcfg_name>" template
and just enumerate ethernet connections "Ethernet 1", "Ethernet 2", ...
but that doesn't solve the DEVICE= conflict.

It's just weird that connection with a DEVICE=eth0 line ends up using
eth1 especially since the inittools do it the other way around.