Re: hanging up for the previous P2P user
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On 05/25/09 15:34, Michael M. Moore wrote:
> On Mon, 2009-05-25 at 09:50 +0800, jidanni@jidanni.org wrote:
>> Tony Baldwin writes:
>>
>>> I'm just guessing here, but I honestly thought killing the client
>>> should stop the incoming connections from seeking the ip, so I'm a
>>> little confused, and curious about the matter, now that you've brought
>>> it up.
>> I've used transsmission(1). When one wants to stop torrenting,
>> transsmission spends several seconds saying proper goodbyes to the
>> tracker or whatever, then exits.
>>
>> What I assume is happening here is when I connect to my ISP, the
>> previous user of that IP address has not said these proper goodbyes,
>> hence 'the liquor store has been converted to a church but not all the
>> previous customer know that so they keep on knocking on the door'.
>
> I was under the impression -- perhaps incorrect -- that it takes time
> for information about which nodes are still available on the network at
> any given time to spread across the network. I wouldn't necessarily
> presume that the previous user hasn't said his "proper goodbyes," just
> that the fact that he has signed off hasn't registered everywhere yet.
> Likewise, when you shut down transmission, what it does is send your
> transfer totals to the tracker (which is really only relevant if you're
> using a private tracker, or some other tracker that needs to know your
> ratios for some reason) and disconnects. But your availability on the
> network has already spread to other clients connected to the same
> trackers you're connected to, and in turn from them to still more
> clients that will connect to those trackers. The clients don't
> broadcast information about who has signed on or off -- they only
> attempt to establish a connection when they're needing connections. So
> someone coming on to the network, say, 15 minutes after you've left
> might still try to connect to you because his client is getting
> information from another client that doesn't know yet that you've left
> (because it has not attempted to establish a connection with your client
> in the time since you left). In a reasonably short period of time,
> every client that knew about you will figure out that you are no longer
> available, and will stop passing on your information.
>
> It's kind of a daisy chain of information, where everybody's information
> is not exactly up-to-the-minute, but it's current enough for the network
> to function effectively. If everybody on the network knew, instantly,
> when any given person had signed on or off, then it wouldn't really be a
> decentralized network.
>
While I agree with everything you've said, if the torrent client were to
send a sign-off signal to all nodes connected to it which would then in
turn *immediately* rebroadcast that to all the nodes connected to it,
wouldn't it still be a decentralised network but with no problem's like
the OP's? Although, I suppose that would create a lot of extra traffic
which would cause slower download speeds.
- --
Many thanks
Harry Rickards
- -----BEGIN GEEK CODE BLOCK-----
Version: 3.1
GAT/GCM/GCS/GCC/GIT/GM d? s: a? C++++ UL++++ P- L+++ E--- W+++ N o K+
w--- O- M- V- PS+ PE Y+ PGP++ t 5 X R tv-- b+++ DI D---- G e* h! !r y?
- ------END GEEK CODE BLOCK------
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iEYEARECAAYFAkoasDEACgkQ1kZz3mRu0GoYdACg0DhhNgcTWbYDwPTVual6Vc/0
8xEAoN3eIugCOEAlPb9cTdNBFHlPu9sZ
=22co
-----END PGP SIGNATURE-----
Reply to: