[Battlemesh] Out of sync seqnos after reboot
hrogge at googlemail.com
Fri Mar 9 07:42:04 CET 2012
The important question is how many different (sequence numbers)
messages can be in transit at the same time?
The "remember the last 31 possible sequence numbers" is just a
uint32_t used as a binary mask that is shifted every times OLSRd
accepts a new message. So even a memory of 64 sequence numbers should
be trivial and efficient to implement this way.
On Thu, Mar 8, 2012 at 22:28, Juliusz Chroboczek <jch at pps.jussieu.fr> wrote:
>> Message sequence numbers in OLSRd are originator specific, so even if
>> the order becomes mixed up, OLSRd will just ignore the outdated one
>> that comes later.
> Hmm... I need to think it over. Babel is different from OLSR, as Babel
> routing messages can be delayed significantly (tens of seconds) and
> still be accepted, even not in seqno order, as long as the feasibility
> condition is obeyed.
> But you're right, there are probably some cases where I can detect an
> unexpected seqno and discard the feasibility state; I'll think it over.
>> Worst case might be a restart with a sequence number 31 less than the
>> last one,
> -- Juliusz
> Battlemesh mailing list
> Battlemesh at ml.ninux.org
Steven Hawkings about cosmic inflation: "An increase of billions of
billions of percent in a tiny fraction of a second. Of course, that
was before the present government."
More information about the Battlemesh