I would be interested to in participating in both direct and HTTP
seeding using gentoo.chem.wisc.edu. The server is on a shared 100Mbps
connection, with about 80% of bandwidth being typically available. I
could also seed bittorrent-only files from a machine on a different
network for a limited time (no HTTP seeding for that one though) if
On 5/3/07, Robin H. Johnson <email@example.com> wrote:
> On Thu, May 03, 2007 at 04:34:19PM -0700, Robin H. Johnson wrote:
> > Would all mirrors that carry the experimental tree please see my
> > follow-up email about BitTorrent HTTP-seeding?
> Hi folks,
> So now stuff on BitTorrent for the 2007.0 release.
> There's two portions to this email, firstly stuff distributed by
> BitTorrent only, and then everything else.
> We have 94 torrents in total, 2 of which are only media that is only
> being distributed by BitTorrent.
> We needs seeds for all of it, but the bittorrent-only stuff more than
> the rest. I have one sponsor offering 50Mbit * 2 weeks for seeding
> everything so far.
> In terms of basic structure, there is a .torrent for every .tar.bz2 or
> .iso in releases or experimental, containing the core file, plus it's
> DIGESTS, signature, and CONTENTS files.
> 3.7G livedvd-amd64-installer-2007.0/
> 3.8G livedvd-i686-installer-2007.0/
> If you're interested in direct BitTorrent seeding (esp. for the above
> two, even if you aren't seeding anything else), please give me a shout
> so that I can arrange to get them to you.
> Now for the experimental part...
> To help out long-term seeding of the majority of torrents (those that
> aren't bittorrent-only), I'd like to take advantage of the newish
> HTTP-seeding support of BitTorrent .
> HTTP-seeding kicks in at a MUCH lower priority than regular BitTorrent
> peering, basically helping clients that cannot contact any other source
> for missing pieces. It effectively converts the BitTorrent client to a
> slightly-smart HTTP fetcher that then shares the blocks back over
> BitTorrent to other clients.
> For all experimental/ mirrors, that means we're adding a directory tree
> under experimental/ that has directories containing only symlinks. Your
> rsync options will need to have the various symlink-transfer options,
> which I think should be in place already.
> Here's an example of one.
> lrwxrwxrwx stage3-x86-2007.0.tar.bz2 -> ../../releases/x86/2007.0/stages/stage3-x86-2007.0.tar.bz2
> lrwxrwxrwx stage3-x86-2007.0.tar.bz2.CONTENTS -> ../../releases/x86/2007.0/stages/stage3-x86-2007.0.tar.bz2.CONTENTS
> lrwxrwxrwx stage3-x86-2007.0.tar.bz2.DIGESTS -> ../../releases/x86/2007.0/stages/stage3-x86-2007.0.tar.bz2.DIGESTS
> lrwxrwxrwx stage3-x86-2007.0.tar.bz2.asc -> ../../releases/x86/2007.0/stages/stage3-x86-2007.0.tar.bz2.asc
> This in itself doesn't make the BitTorrent clients contact you, so even
> if you aren't participating, it doesn't cause harm to have it.
> For part two of the HTTP-seeding (eg making the BT clients actually use
> the mirrors), I'd like to know which of the experimental mirrors are
> interested in being used for HTTP seeding.
> Your webservers will need to support HTTP Range requests for this to
> work. The BitTorrent clients will send range requests for 32K..256K
> blocks (powers of two only).
> I'll be placing the base potion of your URLs into the .torrent files, as
>  The actual specification for the HTTP-seeding is here:
> Robin Hugh Johnson
> Gentoo Linux Developer & Council Member
> E-Mail : firstname.lastname@example.org
> GnuPG FP : 11AC BA4F 4778 E3F6 E4ED F38E B27B 944E 3488 4E85
email@example.com mailing list