Re: bridge carrier issue

[Dan Williams <dcbw@redhat.com>]

On Tue, 2011-09-20 at 15:29 -0400, David Miller wrote:
> I'm marking the patches in patchwork on this as "deferred".
> 
> It is my opinion that we need to push back on the virtualization stuff
> because otherwise it is impossible for network manager and friends to
> operate properly.
> 
> Without the carrier going on only when a device is actually attached
> and operating, NM has no "signal" to know when it actually should do
> DHCP and whatnot.

There's stuff we're planning for NM that'll make it much more friendly
to these use-cases, but that's mostly *after* the interface is
configured and running.  What the semantics of the bring-up behavior
should be need more discussion.  ie, in the case we're talking about
here, NM wouldn't even apply a static IP configuration to the bridge
until the bridge said it was up.  That's something we can change with a
flag in the bridge's config.

Dan

> On the other hand, the virtualization tools and libraries can operate
> properly within the constraints added by the carrier behavior we have
> now.
> 
> And I also believe they can do so in a way that allows them to work
> on kernels before the carrier change.
> 
> Therefore, the only way to move forward and get everything working is
> to change the virtualization tools.
> --
> To unsubscribe from this list: send the line "unsubscribe netdev" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html