[00:09:42] *** Goldfisch has joined #bittorrent
[00:11:22] <Goldfisch> I have been trying to setup a bittorrent server of my own, in to share big files with my colleague. I was using http://ashughes.com/?q=node/85, but I can't seem to get the server-side instance of transmission to seed anything. Searching google is tricky because 99% of all things seems to come back involving client-side stuff or bittorrent sites.
[00:12:01] <Goldfisch> I pointed the download location to be the same folder as the source content, and also switched it to start seeding despite ratios.
[00:12:28] <Goldfisch> But when I look at transmission's info, it sees 1 peer, 0 seeds, and 1 leech, with 0% download.
[00:12:53] <Goldfisch> Err...when I look at the statistics of what should be a seed, I see 1 peer, 0 seeds, and 1 leech, with 0% download.
[00:14:56] <burris> first off, if your seed is showing 0% then you have one problem, peers not connecting to each other may be a different problem
[00:15:44] <burris> on the seed you have to load the torrent file then have the client save it over the original copy, it should check that everything matches and start up at 100%
[00:16:24] <burris> 0% indicates that the data is not where the client thinks it is
[00:16:39] <burris> Goldfisch, ^
[00:19:55] <Goldfisch> burris: I put the original data in /Public. I then used transmission to generate a .torrent for that, and put it on the Desktop. Next, I opened my torrent file using transmission, and changed download destination to folder Public where the original data is, and it just sits there at 0%. Did I miss a critical step there?
[00:21:11] <burris> does it appear as if Transmission put a new directory in /Public?  maybe selecting the original data when you save will do the trick...  the cilent is supposed to detect if you have selected the directory with the data or the directory containing the directory of data (in the case of multifile torrents)
[00:22:48] <Goldfisch> burris: No, I don't see a new directory inside /Public. Does this have anything to do with the fact that I built the torrent out of a folder rather than an individual file?
[00:23:27] <burris> that should be ok, dunno whats up with Transmission
[00:24:26] <Goldfisch> burris: Bing! I created a torrent from my sample.txt file inside /Public, loaded it, and now it reports 100%.
[00:25:16] <Goldfisch> burris: I can see it in the seeding folder now. Let me see if I can now distribute it to another machine I have.
[00:26:51] <Goldfisch> burris: Works great now. Tnx.
[00:27:02] <Goldfisch> burris: I was able to distribute it to another client.
[00:27:15] <burris> so now you just need to get it to work with your folder full of stuff
[00:30:34] <burris> also, just because it worked with another client on your network doesn't mean it will work across the net, clients do multicast discovery of clients on the same subnet...
[00:32:06] *** Andrius has quit IRC
[00:32:29] <Goldfisch> Sure. That's the next step.
[00:41:18] *** Goldfisch has quit IRC
[00:48:39] *** RubberMonkey has quit IRC
[01:16:31] *** Switeck has quit IRC
[01:29:47] *** ari-__-e has joined #bittorrent
[01:33:18] <K`Tetch> http://torrentfreak.com/us-government-told-piracy-losses-are-exaggerated-100616/
[01:33:36] *** ari-_-e has quit IRC
[02:12:30] *** ari-__-e is now known as ari-_-e
[02:37:27] *** [diablo] has quit IRC
[02:58:33] *** K`Tetch has quit IRC
[03:16:10] *** init0 has quit IRC
[03:17:49] *** init0 has joined #bittorrent
[04:01:46] *** The_8472 has quit IRC
[04:07:22] *** The_8472 has joined #bittorrent
[04:10:10] *** omn0mz has joined #bittorrent
[04:12:35] <omn0mz> how does bittorrent verify that the data received is correct?
[04:14:51] <alus> by hashing the pieces and comparing them to the hashes from the torrent file
[04:15:23] <omn0mz> oh, ok, makes sense :P
[04:15:43] <omn0mz> What's the DHT, dynamic hash table, stuff?
[04:17:42] <omn0mz> Oh, ok, let me read over the wikipedia page first and see if I have a specific question about it.
[04:18:31] *** goussx has quit IRC
[04:20:22] <omn0mz> how does DHT protect against malicious participants? For example changing the values in the hash table.
[04:24:55] <alus> the BT use of DHT only allows you to store your own IP, which if pulled from the udp packet
[04:25:02] <alus> s/if/is
[04:25:39] <alus> so you can't overwrite anything without a lot of computers
[04:25:57] <alus> and sure, if you have a lot of computers you can do damage to a DHT, but that only cripples the DHT, not BT
[04:29:22] <omn0mz> what's do you mean by the "udp packet"?
[04:30:08] <omn0mz> doesn't bt use tcp?
[04:30:18] <omn0mz> sorry, I'm a noobie :(
[04:46:03] <alus> the BT DHT uses UDP
[04:49:48] *** ari-__-e has joined #bittorrent
[04:52:44] *** ari-_-e has quit IRC
[05:00:41] *** goussx has joined #bittorrent
[05:26:39] *** omn0mz has quit IRC
[05:27:02] *** omn0mz has joined #bittorrent
[06:12:10] *** ari-__-e is now known as ari-_-e
[06:23:52] *** KyleK_ has joined #bittorrent
[06:37:58] *** Kitsoran has quit IRC
[06:56:39] *** KyleK_ has quit IRC
[07:29:38] *** NightOath has quit IRC
[07:29:48] *** uau has quit IRC
[07:29:53] *** NightOath has joined #bittorrent
[07:33:31] *** uau has joined #bittorrent
[07:35:06] *** MassaRoddel has joined #bittorrent
[08:02:52] *** ProperN[out] has joined #bittorrent
[08:06:38] *** ProperNoun has quit IRC
[08:06:40] *** ProperN[out] is now known as ProperNoun
[08:06:55] *** ari-_-e has quit IRC
[08:24:37] *** ygrek has joined #bittorrent
[08:42:10] *** ari-_-e has joined #bittorrent
[09:24:38] *** ygrek has quit IRC
[09:40:31] *** goussx has quit IRC
[09:48:37] *** goussx has joined #bittorrent
[10:47:00] *** [diablo] has joined #bittorrent
[10:55:54] *** Andrius has joined #bittorrent
[11:20:23] *** goussx has quit IRC
[12:22:27] *** ProperN[out] has joined #bittorrent
[12:22:32] *** ProperN[out] has joined #bittorrent
[12:26:38] *** ProperNoun has quit IRC
[12:26:42] *** ProperN[out] is now known as ProperNoun
[13:42:58] *** ygrek has joined #bittorrent
[14:35:02] *** ygrek has quit IRC
[15:13:58] *** BitTorrentBot has quit IRC
[15:14:43] *** DeHackEd has quit IRC
[16:09:16] *** WhatMan has quit IRC
[16:15:08] *** burris has quit IRC
[16:15:39] *** BitTorrentBot has joined #bittorrent
[17:31:56] *** KyleK_ has joined #bittorrent
[17:44:52] *** edigaryev has joined #bittorrent
[17:46:31] *** Kitsoran has joined #bittorrent
[17:54:28] *** KyleK_ has quit IRC
[18:33:10] *** e-manuel has joined #bittorrent
[18:36:35] *** K`Tetch has joined #bittorrent
[18:36:35] *** K`Tetch has joined #bittorrent
[18:38:37] *** e-manuel has quit IRC
[19:01:52] *** goussx has joined #bittorrent
[19:29:16] *** KyleK_ has joined #bittorrent
[19:49:58] *** burris has joined #bittorrent
[20:06:56] *** burris has quit IRC
[20:32:52] *** [diablo] has quit IRC
[20:43:48] *** DeadTorrent has joined #bittorrent
[20:43:51] <DeadTorrent> hi
[20:44:18] <DeadTorrent> is ExtraTorrent.Com any good?
[20:47:16] <DeadTorrent> HoMM 3 dloads on TPB are dead and/or have trj in them. HoMM 3 Complete on ExtraTorrent downloaded. But I want to know your opinions about that tracker.
[20:49:56] <andar> go away
[20:53:09] <charles> DeadTorrent: see /topic
[20:53:15] <charles> *then* go away
[20:53:30] <andar> or just go away
[20:58:41] *** ygrek has joined #bittorrent
[21:09:31] *** edigaryev has quit IRC
[21:22:56] *** ygrek has quit IRC
[21:27:43] *** DeadTorrent has left #bittorrent
[21:28:38] *** [diablo] has joined #bittorrent
[21:58:56] <alus> we really need to finish this last seen complete feature
[21:59:26] <alus> hydri and I came up with a draft, and actually put the data collection piece in uT already
[22:04:10] <The_8472> you mean on a global level?
[22:08:10] <alus> yes
[22:08:20] <The_8472> via trackers or dht?
[22:08:32] <alus> gossip
[22:08:42] <The_8472> mhh
[22:09:20] <The_8472> not cascading though, right?
[22:11:27] <alus> cascading, actually
[22:11:45] <The_8472> then misreporting peers could spoil the stats
[22:11:55] <alus> there is not much point in attacking this metric.
[22:12:08] * The_8472 points at hanlon's razor
[22:12:59] <The_8472> if the swarm is small you'll see all peers anyway, no cascading needed. if it's big... well... you'll see a significant subset at least, and if nobody of them has seen a complete copy then you're most likely screwed anyway
[22:13:03] <The_8472> again, no cascading needed
[22:13:22] <alus> imagine small swarms with NAT involved
[22:13:49] <alus> or, late-comers to a recently seeded torrent
[22:14:07] <The_8472> well, those late-comers would learn from their neighbors
[22:14:33] <alus> if they knew first-hand, which maybe they don't
[22:14:51] <The_8472> swarm swarms with nat would justify 1hop cascading. i.e. sending 2 stats. "what i see", "best my immediate neighbors see"
[22:14:55] <alus> I'd be open to a hop count or something on the cascade if you like, but I think cascading is more helpful than not
[22:15:07] <The_8472> *small
[22:15:56] <alus> the only downside is if some client writes a bug, which is not something I care about planning for. they could just as easily write a bug which says they are seeds when they are not, or uploads bad pieces (hey, that happens!) or what have you
[22:16:16] <alus> and a simple timestamp is easier to manage than all that
[22:16:29] <alus> they have a motivation to get it right, so that their client is more informative
[22:16:44] <The_8472> 1hop cascading. i.e. sending 2 stats. "what i see", "best my immediate neighbors see" <-
[22:17:14] <alus> yeah, that's almost as good :P
[22:17:27] <The_8472> 90th percentile instead of "best" maybe
[22:18:23] <alus> you'll have to re-calculate what your connected neighbors see every time you want to send it out to other peers
[22:18:30] <The_8472> but i think that's not really needed. since you can use the direct stat of all your neighbors to reference
[22:19:21] <The_8472> <alus> you'll have to re-calculate what your connected neighbors see every time you want to send it out to other peers <- send it synchronized across all peers... just calculate for all peers, then send to all peers
[22:20:32] <alus> well, you'd want to send it in the handshake too
[22:20:57] <alus> it's not a complicated thing, just mentioning
[22:22:37] <The_8472> you can update your stat in an event-driven way too if you want. or just calculate it every N minutes and send slightly outdated stats
[22:22:59] <alus> anyway, if there is some client screwing the pooch, he'll screw it for all his neighbors too.
[22:23:10] <alus> enough of those, and it doesn't matter you're not broadcasting it
[22:23:17] <The_8472> only the 2nd stat
[22:23:20] <The_8472> not the 1st one
[22:24:03] <alus> but which will you display?
[22:24:08] <alus> to the user
[22:25:04] <The_8472> well, you can do some sanity-checking with all that stuff and then evict outliers
[22:26:32] <alus> throwing away valid data because you are concerned someone wrote a bug...
[22:26:59] * The_8472 doesn't trust cascading. it just seems wrong
[22:27:33] <alus> you were going to put it in the DHT :P
[22:27:45] <alus> peer writable global data == cascading
[22:28:00] <The_8472> no, i wouldn't have done that
[22:28:47] <alus> well, the tracker isn't an option either.
[22:28:54] <alus> since no one would ever implement it
[22:29:17] <The_8472> just need to knock on erdgeist's door ^^
[22:29:19] <alus> so gossip seems to be the answer. and cascading is the easy way to build globalish data into gossip
[22:29:47] <alus> I think the erdgeist has had enough knocking on his door from more frightening people than us
[22:30:36] <The_8472> yeah, but if he implements it then we woudl see it on some of the biggest trackers
[22:31:31] <alus> also, a peer checking in to the tracker doesn't mean it's connectable or even connecting to anyone
[22:31:38] <alus> and someone could poison that very easily
[22:31:48] <The_8472> anyway, why aren't seed/peer stats not good enough? they're almost always indicative of the swarm's state
[22:32:30] <alus> it mostly matters on smaller swarms, where either the seed is not connectable or comes and goes because of queueing or daytime hours or what have you
[22:32:40] <The_8472> mhh
[22:32:58] <alus> you might think these swarms are dead if you used seed/peer stats, but often they are not.
[22:33:11] <alus> telling the difference is something you can only do by sitting on it for a few days, and that's boring
[22:33:32] <The_8472> well, stuff has to die eventually anyway
[22:33:59] <The_8472> but yeah, in those cases it might make sense
[22:34:03] <alus> right, but these are often not dead at all.
[22:34:30] <alus> imagine a private tracker where there are potential seeders not on the torrent at all, who look for torrents with peers and jump on them
[22:34:52] <alus> that happens at some frequency you can either write down and gossip about, or sit on the torrent and observe
[22:35:25] <The_8472> afaik on private trackers it's quite the opposite problem
[22:35:56] <The_8472> that many seeds sit on torrents without peers, lurking for a lonely peer to come along and force bandwidth down their throat
[22:37:28] <alus> you would need to force-start all your torrents to keep the queueing from moving on
[22:37:56] <The_8472> that's what they do :(
[22:39:12] <The_8472> the problem is that they don't want to wait for scrapes
[22:39:17] <The_8472> because other seeds might be faster
[22:39:40] <The_8472> and they want to get credits for uploading => the sooner you join a swarm with a lonely peer the more bandwidth you can cram down their throat
[22:46:37] <The_8472> that basically voids the queueing :/
[22:50:57] *** K`Tetch has quit IRC
[23:26:06] *** DeHackEd has joined #bittorrent
[23:32:17] *** KyleK_ has quit IRC
[23:32:55] *** jonasl has quit IRC
[23:35:35] *** jonasl has joined #bittorrent

   June 17, 2010  
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | >