[OpenDHT-Users] servers.txt empty

[Sean Rhea]

AFAIK, there aren't any DHTs that fully support nodes behind NATs.  I
think they all treat such nodes as second-class participants, each
hanging off some non-NATed node.  But I could be wrong.

In any case, it seems to me like NATed nodes *should* be second class,
since they are definitely less powerful than non-NATed nodes in some
ways.  For example, you can't initiate a connection to them without
the help of some third node to which they're already connected.

No DHT that I know of is robust to malicious behavior.  They're all
pretty easy to subvert.  The proposals to make them hard to subvert
(e.g., Miguel Castro's work) make them slow.

Sean



On Mon, Dec 1, 2008 at 4:01 PM, Dr Bob <drbob at lunamutt.com> wrote:
> Hi Sean,
>
> Thanks for the reply.
>
> Yes I understand that OpenDHT was your PhD project... and that its not 100%
> reliable. Thanks for keeping it running.
>
> I was using KadC initially for our DHT, but got frustrated with it.
> OpenDHT seemed like a easy to use alternative... and I especially
> liked the Key Timeout feature.
>
> I'm contemplating writing my own DHT/Overlay network for Retroshare,
> and was wondering - as this is your area of expertise - whether you
> could recommend a robust overlay algorithm (or source package)
>
> I've looked at various open-source packages at one point. The most
> promising seemed to be Chimera. But when I actually looked into the
> code I realised that it couldn't handle NAT/firewalls - and was
> primarily just research code.
>
> Best Regards.
>
> Bob.
>
>
>
> Sean Rhea wrote:
>>
>> OpenDHT is now, and will always be (at least from time to time), "on the
>> blink".
>>
>> It's up to the degree that it's up because it mostly works without me
>> messing with it.  When it stops doing so, I have to go fix it, and
>> that takes time.  It was a Ph.D. project, and I have another job now.
>>
>> If you want something rock solid, you'll probably have to pay money
>> for it.  Sadly, such is life.  If you can pay for it, S3 is pretty
>> rad, and it's an ideological descendant of all the DHT research.  If
>> you can't pay for it, you'll have to wait for me to get around to
>> figuring out what's wrong and fixing it.
>>
>> Sean
>>
>> P.S.  I'd love for there to be a way to have users help keep it
>> running, but that would probably involve something along the lines of
>> a Byzantine-fault-tolerant DHT, which to my knowledge no one has ever
>> actually built.
>>
>>
>>
>> On Sun, Nov 30, 2008 at 9:44 AM, DrBob <drbob at lunamutt.com> wrote:
>>>
>>> Hi Sean,
>>>
>>> OpenDHT seems to be on the blink recently. Any news?
>>> Can we (the opendht-users) help at all to keep it running?
>>>
>>> Thanks.
>>>
>>> Bob.
>>>
>>>
>>>
>>> Sean Rhea wrote:
>>>>
>>>> It lists 32 for me right now.
>>>>
>>>> The number is falling though.  Some of the nodes are suffering out of
>>>> memory errors, and I'm not sure why yet.  Work is busy this week, too.
>>>>
>>>> Sean
>>>>
>>>>
>>>>
>>>> On Mon, Nov 17, 2008 at 2:48 PM, Gabriel Maganis <gymaganis at gmail.com>
>>>> wrote:
>>>>
>>>>> Hello Sean,
>>>>>   http://www.opendht.org/servers.txt is not listing any nodes.
>>>>>
>>>>> Thanks,
>>>>> Gabriel
>>>>>
>>>>>
>>>>
>>>>
>>>>
>>> _______________________________________________
>>> OpenDHT-Users mailing list
>>> OpenDHT-Users at opendht.org
>>> http://opendht.org/cgi-bin/mailman/listinfo/opendht-users
>>>
>>
>>
>>
>



-- 
"We're borrowing money from China to buy oil from the Persian Gulf to
burn it in ways that destroy the planet.  Every bit of that's got to
change."  -- Al Gore