[lime-users] setting one mesh device as gateway? (and nanostation M2 update)

Nicolas Pace nico at libre.ws
Wed May 10 17:03:27 UTC 2017


On Wed, 2017-05-10 at 18:45 +0200, Nk via lime-users wrote:
> Fantastic thank you so much.
> 
> Last question I promise: I’ve noticed in traceroute that when the
> first router has wifi off and is ethernet meshing with the second
> one, and I connect my laptop to the first router’s lan, like so:
> 
> Laptop <client over eth> router 1 <mesh over eth> router 2 <mesh over
> wifi> router 3 <client over eth> ISP router
> 
> There is no extra “hop” in the trace route output. I assume this is
> due to the layer 2 meshing, correct?

Exactly!

> However, is this exactly comparable, in terms of performance and
> route selection, bandwidth, latency, and everything, in every way, to
> router 1 meshing over wifi directly with router 3 without going
> through router 2, provided 1 and 2 have the exact same wireless link
> quality toward 3?

Batman-adv calculate which of the links it has from router 1 to 3 is
the best to send information through.

> For instance, let’s say router 1 has a 5GHz AC interface, and router
> 2 has a 2.4GHz interface. If this were a single dual-band router,
> LiMe would automagically select the best wireless interface to mesh
> with every other node on an individual basis,

The best interface (wireless or wired)

> if I understand correctly. Does this dual-router setup perfectly
> replicate this behavior in your opinion, letting the C7 decide
> whether to use its own faster but shorter range 5GHz interface or
> whether to use the 2.4 interface of the 1043 just as if it were its
> own, without being influenced by the fact there’s a hardware-level
> "extra hop”? Or does it de-prioritize the 2.4 interface of the 1043
> just because it’s one extra “hardware-level hop” away [please excuse
> the terminology]?

I believe batman-adv uses latency to determine which link it uses.

> I’m asking this because I’m doing some hardware hacking to build what
> we’re calling an openNODE, essentially a TL Archer C7 and a TL 1043
> ND stuck together in a single package we’re making. It’s an
> experiment to build a high performance dual band, triple-wifi-
> interface node under 150€.

Awesome. Do tell us how it works on the C7, and if you have those
cheaper C3 to test on also!

>  [I’m simultaneously testing out lime-sdk to see if we can get an
> open mesh A60 working, thanks to you ;]

What's an open mesh A60 ?

> The C7 will mesh over 5GHz [provided Gui enlightens us as to how to
> get this working ;] and the 1043 will mesh over 2GHz. The only
> wireless interface available to clients in AP mode will be the 2.4
> secondary interface of the C7.
> 
> If the two act exactly as one, we’re golden. If not, I have to
> understand the cons in having two mesh routers instead of a single
> dual-band mesh router per node.

If they both use libremesh, then they act as one.
There is nothing to do manually, except giving them meaningful names :)

> 
> Thank you so much again!
> 
> Nk
> 
> On May 10, 2017, 6:31 PM +0200, Nicolas Pace <nico at libre.ws>, wrote:
> > On Wed, 2017-05-10 at 15:50 +0200, Nk via lime-users wrote:
> > > Ok sorry so just to confirm, the options then are
> >  
> > > 1] WAN to WAN and
> >  
> > There is no WAN-to-WAN between LibreMesh routers... this is an
> > unexpected behaviour, but not desired nor recommended.
> > 
> > > 2] LAN to LAN, and both routers can also have traditional LAN
> > > clients
> > > on the remaining 3 eth1 ports that receive DHCP as usual without
> > > any
> > > interference?
> >  
> > Yes, this is the default and intended behaviour.
> > 
> > > At this point my setup would be the one you can find attached. If
> > > you confirm it works, I’ll test it out tonight. Thank you so
> > > much.
> >  
> > That is exactly the way you should use it!
> > 
> > Awesome, tell us how it went!
> > 
> > Regards,
> > 
> > > Nk
> > > 
> > > On May 10, 2017, 3:37 PM +0200, Nicolas Pace <nico at libre.ws>,
> > > wrote:
> > > > On Wed, 2017-05-10 at 15:34 +0200, Nk via lime-users wrote:
> > > > > AHAHAHAH epic.
> > > > > 
> > > > > Thanks Ilario, sorry to bother. My tests were first router’s
> > > > > LAN
> > > > > to
> > > > > second router’s WAN, and that wasn’t working. Are you saying
> > > > > they
> > > > > should be connected LAN to LAN [as an alternative to WAN to
> > > > > WAN
> > > > > which
> > > > > we know works perfectly]? Wouldn’t that cause a DHCP
> > > > > conflict?
> > > >   
> > > > Not at all. That is the default behaviour of LibreMesh (crazy,
> > > > right?
> > > > :) ).
> > > > 
> > > > 
> > > > > Thank you.
> > > > > 
> > > > > Nk
> > > > > 
> > > > > On May 10, 2017, 3:33 PM +0200, Ilario Gelmetti <iochesonome@
> > > > > gmai
> > > > > l.co
> > > > > m>, wrote:
> > > > > > On 05/10/2017 02:20 PM, Nikksno via lime-users wrote:
> > > > > > > If I use the eth1 of the first router to mesh with other
> > > > > > > routers
> > > > > > > as
> > > > > > > you're suggesting, the problem is that the meshing
> > > > > > > doesn't
> > > > > > > seem
> > > > > > > to
> > > > > > > happen correctly, as bmx6 shows the second router's route
> > > > > > > as
> > > > > > > being eth1,
> > > > > > > but batman shows all routes to other nodes being over
> > > > > > > wlan
> > > > > > > mesh,
> > > > > > > even to
> > > > > > > the second router.
> > > > > >    
> > > > > > It's not lan, it's not wan...
> > > > > > It's BATWAN nananananananananananananananana batwaaaaaannn
> > > > > > 
> > > > > > No, seriously, I expected batman was present on ethernet
> > > > > > LAN
> > > > > > interfaces...
> > > > > > They're also included in br-lan which is included in bat0,
> > > > > > no?
> > > > > > 
> > > > > > Can you post these?
> > > > > > # brctl show
> > > > > > # batctl if
> > > > > > 
> > > > > > I agree with nicopace, lan to lan should work.
> > > > > > @devs pls? Are ethernet ports with LAN proto not used by
> > > > > > batman-
> > > > > > adv?
> > > > > > _______________________________________________
> > > > > > lime-users mailing list
> > > > > > lime-users at lists.libremesh.org
> > > > > > https://lists.libremesh.org/mailman/listinfo/lime-users
> > > > >   
> > > > > _______________________________________________
> > > > > lime-users mailing list
> > > > > lime-users at lists.libremesh.org
> > > > > https://lists.libremesh.org/mailman/listinfo/lime-users
> > >  
> > > _______________________________________________
> > > lime-users mailing list
> > > lime-users at lists.libremesh.org
> > > https://lists.libremesh.org/mailman/listinfo/lime-users
> 
> _______________________________________________
> lime-users mailing list
> lime-users at lists.libremesh.org
> https://lists.libremesh.org/mailman/listinfo/lime-users
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 801 bytes
Desc: This is a digitally signed message part
URL: <http://lists.libremesh.org/pipermail/lime-users/attachments/20170510/14c4666e/attachment.sig>


More information about the lime-users mailing list