[Battlemesh] Report on the testbed results

[Alessandro Gnagni]

The babeld version is 1.8.0-3
babel-pinger 0.1-1

2017-06-14 15:46 GMT+02:00 Leonardo Maccari <mail at leonardo.ma>:

>
>
> On 14/06/17 13:45, Juliusz Chroboczek wrote:
> >> https://github.com/battlemesh/battlemeshv10-testbed/blob/
> master/scripts/babel
> > Cool, thanks.  So Babel was run with the default 4s Hello interval and
> the
> > default 20s update interval, which is much higher than the other
> > protocols.  This might or might not explain the packet loss and the
> spikes
> > in latency, I'm not sure.
> >
> > If you've got some time, I'd really like to see the time distribution of
> > the packet loss and the CDF graph of the latency.
>
> Hi Juliusz,
>
> You can find all the data for the ping test here:
>
> https://github.com/battlemesh/battlemeshv10-testbed/tree/
> master/tests/failure_recovery/delay-one-run/BABEL
>
> each file is one flow, in each file you have an "entries" field with all
> the rtt.
>
> we used this script to parse them, which you can modify to extract cdf
> (and the parse_all_pings.sh in the same folder to run them all):
>
> https://github.com/battlemesh/battlemeshv10-testbed/blob/
> master/tests/failure_recovery/parser_scripts/parse_ping.py
>
> if you are wondering why the tests are in the failure_recovery folder,
> this  is because my idea was to do the tests and then let some node in
> between fail, but we never got to that point :-)
>
> l.
>
> >
> > -- Juliusz
> > _______________________________________________
> > Battlemesh mailing list
> > Battlemesh at ml.ninux.org
> > http://ml.ninux.org/mailman/listinfo/battlemesh
> >
>
>
> _______________________________________________
> Battlemesh mailing list
> Battlemesh at ml.ninux.org
> http://ml.ninux.org/mailman/listinfo/battlemesh
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://ml.ninux.org/pipermail/battlemesh/attachments/20170616/ccdb4803/attachment-0001.html>