Re: [PATCH 5/8] bonding: Implement bonding slaves

[Thomas Graf <tgraf redhat com>]

On Sun, 2011-11-06 at 10:45 -0600, Dan Williams wrote:
> On Fri, 2011-11-04 at 19:12 -0500, Dan Williams wrote:
> > Could we drop this property and just use the 'type' setting of the
> > connection to determine what type of slave this is?  ie, if 'master' is
> > set and 'type == 'bond'' then this is a bond slave, no?  It seems like
> > most of the time the 'slave-type' value will correspond to the same
> > value as 'type'.  Thoughts?
> 
> After re-reading the patches, disregard this...  slaves won't be
> type=bond since that's what the master is; the slave would just be
> type=ethernet or whatever (well, until we get vlans on top of bonds on
> top of bridges :)

The fact that we are going to support multiple levels of nesting is
exactly the reason why I have put in "slave-type".

BTW, feel free to notice me before you start merging the patches so I
can repost. I have been rebasing the patches and made small corrections
and enhancements in the meanwhile.