[Battlemesh] Simplified testbed

Federico Capoano f.capoano at openwisp.io
Sun Jul 10 23:52:58 CEST 2022

On Sun, Jul 10, 2022 at 9:45 AM Claudio Pisa <clauz at ninux.org> wrote:
> Hi!
> I am replying below.
> On 7/8/22 10:31, ignifugo wrote:
> >
> >
> > On 07/07/22 20:08, Federico Capoano wrote:
> >> Hi everyone,
> >>
> >> I had a chat with some people regarding the possibility of setting up
> >> a testbed and running some simple tests to compare with the results
> >> obtained in the previous years.
> > thank that you open this topic.
> >
> > I participated only in 1 battlemesh when happen in C-base, in berlin.
> > and I arrived in late for a issue with ryanair and I just knew the
> > winner protocol. I remember was BMX6.
> >
> > I'm interested to know the procedure of the tests..  I find in the wiki
> > some description and map of the places.. but.. hem.. for me not enought
> > to understand..
> There is no agreed and standard procedure. In some editions there has
> been a person who designed the test, but there are no perfect tests and
> there have been discussions about whether is better to run all the
> protocols at the same time (to have the same environmental conditions)
> or to run the protocols sequentially, testing one protocol at a time (to
> avoid interference between protocols).

Here's the results of the tests collected in 2015:
You can find there a section where we wrote down the challenges faced
during the event.

In that edition we didn't run all the protocols at the same time, but
rather one protocol at time.

Whether it was a L2 or L3 we did not care much, we executed the same
tests on all of them.

> > https://battlemesh.org/HomePage?action=fullsearch&context=180&value=test
> >
> > for me also is not clear how is comparing protocol of the level 2 and 3
> > together...
> >
> > and.. I would like see the list of the winners of the past years.. if is
> > possible...?
> We don't always have a clear winner on all the tests. And in some
> editions the time was not enough to get a proper testbed. So I don't
> think we are collecting these results in a single place.

> > It's interesting because monitoring now is spread (we use prometheus)
> > and for example for our mesh network using libremesh (batman+babeld)
> > maybe could be possible have some feeback on the quality of routing
> > nodes. at them moment we have not idea of the performance..
> > out of raw tools, like ping and iperf, and dbm, but not viz in
> > prometheus on this.
> > but for example by humans feedback, we know that in supernode, we have
> > to turnoff "anygw" from all the antennas-1 because the routing became
> > crazy :) that's ok, but not automagic. hum...
> >
> > only to say that know the routing and have feedback can improve our life.
> To use monitoring tools to collect test data continuously, I think that
> we would need a separate network, either wired or on a separate channel.
> I don't know if we have multiradio devices that we can use.

Sounds complicated.

> > Ah.. I promoted a bit the event, and in some emails I explained that we
> > are not militaristic and the "battle" is a friendship, not agonistic
> > competition.  for us probably is clear, but.. better be explicit in this
> > years..
> >   :D
> >
> >>
> >> It seems the main reason for deciding to not run the tests was that
> >> this year the event is shorter so there wouldn't be enough time, I
> >> think I can deal with that problem.
> >>
> >> I am thinking of basing the firmware on OpenWrt 22.03 and test only
> >> the protocols which are still in use by the people actively following
> >> and participating in the event (so instead of attempting to test all
> >> the protocols, we focus on the ones which have active people using them).
> >>
> >> Are there any people interested in this idea who will be willing to
> >> help out?
> >>
> > +1
> > cheers
> +1 also from me

Thanks for the encouragement.

So what are the dynamic open source routing daemons are still in use
and maintained by this community nowadays?

I would definitely be interested in comparing how the standard 802.11
mesh protocol behaves compared to other protocols like babel.

Anyone else is interested in this?


More information about the Battlemesh mailing list