We have used HWMP as well. My comments mirror Mike's. In our experience any devices connected to a MESH node via ethernet or on another wireless interface (think hotspot clients) would become unreachable. I managed to pinpoint this down to entries in the HWMP FDB (Forwarding Database) becoming stale, but never actually timing out. This would mean that HWMP would often try and route the traffic to the previous mesh node that had that MAC address as a client on one of it's ports, effectively black holing it. We worked around the problem using the same trick as Mike, e.g. EoIP tunnels over the HWMP mesh network. On Mon, May 2, 2016 at 9:01 PM, Mike Everest <mike@duxtel.com> wrote:
Hi!
Sure - happy to share any potentially useful information. Like I said, we have used it may times - most simple use case is public access hotspot network (e.g caravan park) where APs are deployed at various locations that have power only. In this way, mesh works perfectly because there's no devices connected to the ethernets of any AP (except for the 'mesh portal' device/s)
Another interesting one is for a digital signage network in a major city railway station - again, there are no opportunity for network cables, and sometimes power to those signs fails for whatever reason. So wireless mesh is a good way to ensure all units are always online. In this case, every mesh node has something connected to the Ethernet socket (being signage controller) so this is where we first discovered that 'bug'
I worked that bug like a champion, documented it and even demonstrated to them exactly what was happening (i.e. some nodes did not learn new optimised path properly, so packets ended up bouncing back-and-forth between two nodes) but they never did anything about it. At least they never did anything about it within a few months after I first reported it (back 2-3 years now) and I never noticed any change to that behaviour since.
So actually, it might even be fixed by now, but I don't trust it enough to deploy a production network to find out if it IS fixed ;)
Sure, adding 'star' topology over the top of the mesh adds a layer of complexity, but only at deploymemt time. After that, everything works normally - potentially 'forever' ;)
Cheers!
Mike.
-----Original Message----- From: Public [mailto:public-bounces@talk.mikrotik.com.au] On Behalf Of Karl Auer Sent: Monday, 2 May 2016 5:36 PM To: public@talk.mikrotik.com.au Subject: Re: [MT-AU Public] mesh network using MikroTik?
On Mon, 2016-05-02 at 16:42 +1000, Mike Everest wrote:
We've done it several times, and apart from one significant caveat, it works very well indeed! :-)
Can you share the use case?
And is that caveat likely to be fixed any time soon?
An effective work-around is to connect all mesh nodes back to the 'mesh portal' by EoIP tunnel
"Effective" I suppose, but another layer of complexity. Rather not :-(
Regards, K.
-- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~ Karl Auer (kauer@nullarbor.com.au) work +61 2 64957435 http://www.nullarbor.com.au mobile +61 428 957160
GPG fingerprint: 6D59 8AE6 810D 44E3 7626 7040 4DD6 F89F 3053 4774 Old fingerprint: 9DCA 0903 BCBD 0647 BCCC 2FA7 A35C 57A1 ACF9 00BB
_______________________________________________ Public mailing list Public@talk.mikrotik.com.au http://talk.mikrotik.com.au/mailman/listinfo/public_talk.mikrotik.com.au
_______________________________________________ Public mailing list Public@talk.mikrotik.com.au http://talk.mikrotik.com.au/mailman/listinfo/public_talk.mikrotik.com.au