Re: Yocto Linux Project Sync notes

[Tom Zanussi <tom.zanussi@intel.com>]

On Tue, 2010-10-19 at 14:44 -0700, Hart, Darren wrote:
> On 10/19/2010 09:04 AM, Zanussi, Tom wrote:
> > On Tue, 2010-10-19 at 08:40 -0700, Stewart, David C wrote:
> >
> >>
> >> * demo
> >>
> >> ** Alex will put the hardware bits together, Darren to write down the description of what needs to come together and wiki-fy it. RP suggests that maybe getting all the pieces running on one hardware might be a good way to try it out just in case.
> >>
> >
> > I can try getting everything running on one machine too - would it make
> > sense to try that all on e.g. emenlow, or all on a laptop using qemu, or
> > both?
> 
> Everything on One machine doesn't make a lot of sense to me, for two 
> reasons:
> 
> 1) This is heavily network dependent, making sure the machines can find 
> eachother is a critical aspect of testing this.
> 2) Qemu does some magic with tun/tap devices, you have to configure 
> things to get multicast working between guests. I've tested the 
> mediatomb image in qemu, it works. We need to test on real hardware.
> 3) We don't an image that will install everything on the same machine, 
> and making one doesn't make a lot of sense to me.
> 4) Running a NAS to serve up media to itself doesn't test the networking 
> side of things as well.
> 
> I'd rather use 3 real devices (even if they are all emenlow or 
> blacksand) rather than qemu or all on one.
> 

Yeah, it doesn't make sense as a demo per se, but for the folks who only
have one system, to get it all going.  Another motivition I thought was
to have 'something' as a worst-case fallback, e.g. with all the 'moving
pieces' Dave is worried about.

Actually, it could make a little bit of sense to have a single
nas-mediatomb-xxxclient image, pointing to each other on localhost, that
could be easily tweaked in the field and swapped in/out as any one of
the components - just a thought and probably not worth it at this point.

Tom

> --
> Darren
> 
> >
> > Tom
> >
> 
>