[Battlemesh] Battlemesh Digest, Vol 24, Issue 13

[Markus Kittenberger]

On Thu, Mar 8, 2012 at 1:00 PM, <battlemesh-request at ml.ninux.org> wrote:

> Date: Thu, 08 Mar 2012 10:20:54 +0100
> From: Axel Neumann <neumann at cgws.de>
> Subject: Re: [Battlemesh] Battlemesh v5 tests
>
> 2)  intuitively deactivated and reactivated nodes to save power.


> Power outages are rather seldom. Especially short-time ones.
> But what happens is that nodes are turned down intuitively to save power
>

I agree with axel, that real (or administrative) power outages last a
little longer,..
(so protocols usually have some time to timeout stuff)

But i would vote to not ignore the "reboot instead of debug" mentality of
many mesh network users,..

So its a valid testcase, but to keep realism do not perform it too often (-;
(As its hopefully not the typical mesh-network users full time "hobby" to
power on/off there routers)

And luckily many mesh (non-)users also have the, "just ignore it" (as it
will fix itself) mentality,.. (common especially in older/settled networks)


>
> Therefore I would actually vote for something like decreasing tx-power,
> turn off the interface
>

hmm a good option, as its works fast and is scriptable.
(but i agree its not a real-live test)


>
> If this is not possible, what about turning them down for 5 minutes?
>
I guess this could make these tests take some time,..

Markus

p.s. regarding olsrd, if it knows that is has to shutdown (e.g. receives a
SIGTERM) it immedialtely sends out some goodbye messages (infact just
normal oslrd messages saying i have no links, and no neighbours = forget me
(-;

Depending on firmware (how and how fast it shuts down) this also works on a
reboot, and there are usually no issues with old seqno, topology
information, for olsrd on a reboot.

Of course plugging out power does not give olsrd (or other protocol with
similar features) time to say goodbye.

And i fear if installation allows it (i.e. easy user-accesible power source
of the routers)
such real powercycles might be more common as reboots (via gui or shell)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://ml.ninux.org/pipermail/battlemesh/attachments/20120308/840b6cb2/attachment-0001.html>