[Nfd-dev] [Operators] reachability of broadcast and guest prefixes

[Alex Afanasyev]

Hi John,

I have proceed with your suggestion and new package is on its way to be available soon.   I did basic tests, but could have screwed up again with upstart scripts, so be cautious :)

Small note.  ALL_FACES_PREFIXES should be only for /ndn/broadcast

/ndn/guest should be part of ON_DEMAND_FACES_PREFIXES only on spurs, since only spurs is home for guest users.  All other nodes should just have their site's prefix as part of on-demand.

---
Alex

On Aug 14, 2014, at 12:34 PM, John DeHart <jdd at seas.wustl.edu> wrote:

> 
> Alex and Junxiao,
> 
> Is it possible to expand nfd-autoreg to handle two sets of prefixes, one for on-demand faces and one for all faces?
> Right now in the config file for autoreg config file we have
> # Prefixes to register
> PREFIXES="/ndn/guest /ndn/broadcast /ndn/edu/memphis"
> 
> What if we had
> # Prefixes to register on All faces:
> ALL_FACES_PREFIXES="/ndn/guest /ndn/broadcast"
> 
> # Prefixes to register just on on-demand faces:
> ON_DEMAND_FACES_PREFIXES="/ndn/edu/memphis"
> 
> on-demand faces would get both sets.
> Non-on-demand faces would get just ALL_FACES_PREFIXES.
> 
> John
> 
> 
> On 8/14/14, 1:10 PM, Alex Afanasyev wrote:
>> I have proposed that before, but will repeat.  I don't think that broadcast prefixes (and setting the strategy) is what NLSR should do.  At most, this should be some other routing protocol.   What we could do instead is to write a tiny little daemon like nfd-autoreg, which job would be just to register /ndn/broadcast for every created face (not just on-demand ones).  May be we can ignore application faces, though it doesn't matter for NDN testbed much.
>> 
>> ---
>> Alex
>> 
>> On Aug 14, 2014, at 9:55 AM, Junxiao Shi <shijunxiao at email.arizona.edu> wrote:
>> 
>>> Hi John
>>> 
>>> fib.max-faces-per-prefix is the maximum number of Routes with same Name prefix that NLSR will install to the RIB.
>>> The parameter starts with "fib." due to historical reason; it concerns what NLSR would install to the RIB, not FIB.
>>> 
>>> I suggest setting this parameter to 60, so that expansions in the near future are also covered.
>>> When there are no more than 5 backbone links, setting it to 5 or 60 has same effect.
>>> 
>>> This short term solution would stop working when a HUB has more than 60 backbone links, but this day is unlikely to come within one year.
>>> 
>>> Yours, Junxiao
>>> 
>>> 
>>> On Thu, Aug 14, 2014 at 9:46 AM, John DeHart <jdd at seas.wustl.edu> wrote:
>>> 
>>> Junxiao,
>>> 
>>> Ahh. I was thinking of the fib.max-faces-per-prefix as a limit on all faces not just
>>> ones that NLSR is concerned with. 
>>> 
>>> In the testbed right now we have 5 nodes that
>>> each have 5 links to other nodes. Those are the most connected nodes.
>>> So, if we set the fib.max-faces-per-prefix for NLSR to >= 5 we should be ok.
>>> Right?
>>> 
>>> John
>>> 
>>> _______________________________________________
>>> Operators mailing list
>>> Operators at lists.named-data.net
>>> http://lists.named-data.net/mailman/listinfo/operators
>> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.lists.cs.ucla.edu/pipermail/nfd-dev/attachments/20140814/54558e70/attachment.html>