LFS-5.x tarball for P2P distribution

Joel Miller cheeziologist at mail.isc.rit.edu
Mon Sep 22 12:28:14 PDT 2003

On Mon, 22 Sep 2003 02:47:51 +0200, Jeroen Coumans 
<jeroen at linuxfromscratch.org> wrote:

> Hi Joel Miller. You said the following on 09/21/03 20:02:
>> On Sun, 21 Sep 2003 19:52:20 +0200, Jeroen Coumans
>>> AFAIK no. There will be no P2P program on the LFS server, so this is 
>>> strictly a community-based effort.
>> If that's the case then we need either a) someone to host a tracker on 
>> one of their own systems, or b) to come to an agreement as to what sites 
>> tracker we will use. AFAIK we do need a tracker in order to be able to 
>> use a bittorrent file. It is possible that I can do the former of the 
>> options on my server back at home, but id prefer not to have to open 
>> more ports than are already open on that machine.
> Ok, what does it mean to run a tracker? And is there perhaps a public 
> tracker we can submit our package to?

Taken from the BitTorrent Docs:
"You need a tracker for downloaders to be able to find each other. Because 
a tracker should be on a very reliable net connection, it's strongly 
recommended that you use a tracker run by someone else, especially if you 
don't have a co-located machine. Everyone else making files available via 
BitTorrent will be using a tracker, you should ask one of them for the url 
of theirs. (A single tracker can handle a huge number of files with very 
little load.) Don't use anyone's tracker without their permission, that's 
rude. (Tracker urls are contained in .torrent files.)"

A tracker doesn't actually host the file, all it does is keep track of how 
many users are dl'ing and who has how much of the file and so on and so 
forth. There needs to be a tracker so that each client can find other 
clients to download from/upload to. The tracker also makes sure that each 
piece of a file a user dl's is valid by use of MD5 sums. There are public 
trackers we can use but my experience has been that all public trackers end 
up sharing copyrighted files and end up being shutdown eventually. Using a 
public tracker is not the most reliable of solutions IMHO. Ideally, I think 
we should run a tracker on the LFS server to track only this file (and 
possibly any others down the road). I can't vouch at all for how much 
bandwidth it takes, but since there is no actually part of the file going 
through the server I can't imagine it takes much.

Registered LFS User 6929
Registered Linux User 298182

More information about the lfs-dev mailing list