headers
Gil Hirsch | 1 Dec 2009 13:16

Re: Multi-tracker not multi-tracking?

I'll clear things up:

For this multi-tracker torrent:
- 0.14 attempts to connect to the first tracker on the list. If successful,
then that tracker's peers are the only ones added as candidates (not
including DHT). This is according to spec.
- 0.15 with announce_to_all_tiers+trackers set to false: the detailed
session logs shows that libtorrent announces on all trackers, and the
torrent_status.current_tracker value keeps updating with the last one
attempted. However, it doesn't seem to have gathered all the the peers from
all of the trackers as candidates, and reports very low swarm numbers.
- 0.15 with announce_to_all_tiers+trackers set to true: same behavior as
0.15 with announce options set to false.
- utorrent - connects to all trackers on the list, continuously updates all
of them, and gathers all peers from all trackers to a single list of > 1500
peers (most of which are registered on only one tracker out of this 30
trackers list...)

Re. Tier bug - it seems that it was uninitialized - we didn't run a debugger
against it though to fully assess it, just removed it from our printouts...

We also hit another bug - seems that upload throttling wasn't working, it
the upload was always maxed out (using a single torrent).

Thanks, and let me know if we can help further in analyzing and ironing out
issues!

On Sun, Nov 29, 2009 at 7:51 PM, <arvid <at> cs.umu.se> wrote:

> Quoting Gil Hirsch <birsch <at> tvitty.com>:
(Continue reading)

Permalink | Reply | 1 comment |
headers
Anis Abdul | 3 Dec 2009 15:56
Picon

Deleting torrent

Hi,

I have implemented torrent on Tonido platform using libtorrent library. 

When a user wants to delete a existing torrent, I use the remove_torrent api to delete it from the session. Is
there anything else that needs to be done ?

 I have multiple users who inform that the system's torrent port (default at 6881) is still getting pounded
by Torrent requests from other Torrent clients. 

Any help on this is highly appreciated.

Fyi: Here is a form post about this http://tonido.com/forum/viewtopic.php?f=33&t=476

Regards

------------------------------------------------------------------------------
Join us December 9, 2009 for the Red Hat Virtual Experience,
a free event focused on virtualization and cloud computing. 
Attend in-depth sessions from your desk. Your couch. Anywhere.
http://p.sf.net/sfu/redhat-sfdev2dev
Permalink | Reply | 2 comments |
headers
arvid | 3 Dec 2009 18:15
Picon
Picon
Favicon

Re: Deleting torrent

Quoting Anis Abdul <anismo <at> gmail.com>:

> Hi,
> 
> I have implemented torrent on Tonido platform using libtorrent library. 
> 
> When a user wants to delete a existing torrent, I use the remove_torrent api
> to delete it from the session. Is there anything else that needs to be done
> ?
> 
>  I have multiple users who inform that the system's torrent port (default at
> 6881) is still getting pounded by Torrent requests from other Torrent
> clients. 
> 
> Any help on this is highly appreciated.
> 
> Fyi: Here is a form post about this
> http://tonido.com/forum/viewtopic.php?f=33&t=476

It's perfectly normal that other peers will keep trying to connect to you even
after you stop the torrent, it shouldn't go on for much longer than one tracker
announce interval though.

>From the posts, it's not clear what kinds of requests they are talking about.
The fact that they assume it's caused by torrent's not quite being stopped
suggests that they see TCP SYN packets for that port.

However, I don't get the impression that they have done any significant
reasoning behind their observations, so they might very well see DHT traffic.
(Continue reading)

Permalink | Reply | 1 comment |
headers
Anis Abdul | 3 Dec 2009 18:55
Picon

Re: Deleting torrent

Thank you Arvid. I will expose an UI option to turn DHT off.

Regards
On Thu, Dec 3, 2009 at 11:15 AM, <arvid <at> cs.umu.se> wrote:

> Quoting Anis Abdul <anismo <at> gmail.com>:
>
> > Hi,
> >
> > I have implemented torrent on Tonido platform using libtorrent library.
> >
> > When a user wants to delete a existing torrent, I use the remove_torrent
> api
> > to delete it from the session. Is there anything else that needs to be
> done
> > ?
> >
> >  I have multiple users who inform that the system's torrent port (default
> at
> > 6881) is still getting pounded by Torrent requests from other Torrent
> > clients.
> >
> > Any help on this is highly appreciated.
> >
> > Fyi: Here is a form post about this
> > http://tonido.com/forum/viewtopic.php?f=33&t=476
>
> It's perfectly normal that other peers will keep trying to connect to you
> even
> after you stop the torrent, it shouldn't go on for much longer than one
(Continue reading)

Permalink | Reply | 0 comments |
headers
arvid | 5 Dec 2009 18:46
Picon
Picon
Favicon

Re: Multi-tracker not multi-tracking?

Quoting Gil Hirsch <birsch <at> tvitty.com>:

> I'll clear things up:
> 
> For this multi-tracker torrent:
> - 0.14 attempts to connect to the first tracker on the list. If successful,
> then that tracker's peers are the only ones added as candidates (not
> including DHT). This is according to spec.
> - 0.15 with announce_to_all_tiers+trackers set to false: the detailed
> session logs shows that libtorrent announces on all trackers, and the
> torrent_status.current_tracker value keeps updating with the last one
> attempted. However, it doesn't seem to have gathered all the the peers from
> all of the trackers as candidates, and reports very low swarm numbers.

What is the symptom you see to make you think that libtorrent didn't "gather all
the peers"?

The log should tell you how many peers each tracker returned. Did the num_want
parameter limit the number of peers?

> - 0.15 with announce_to_all_tiers+trackers set to true: same behavior as
> 0.15 with announce options set to false.

Are you sure? You're essentially saying that the settings don't have an affect?
That's not the case for me.

> - utorrent - connects to all trackers on the list, continuously updates all
> of them, and gathers all peers from all trackers to a single list of > 1500
> peers (most of which are registered on only one tracker out of this 30
> trackers list...)
(Continue reading)

Permalink | Reply | 0 comments |
headers
Chris Morley | 12 Dec 2009 22:39
Picon
Favicon

rtorrent to use custom Curl libs (CentOS 5.4 uses Curl 7.15.5)


Hi, i am getting the time-out issue when trying to connect to known working trackers.

I believe this is because my curl libary is outdated.

I would like to download the latest curl version and compile and link to rtorrent without affecting system files.

How can i statically link the latest Curl libs to my rtorrent? I belive its to do with LD_LIBRARY_PATH but i am
not sure howto proceed:

wget http://curl.haxx.se/download/curl-7.19.7.tar.gz
cd curl-7.19.7./configuremakemake install

The above make install would overwrite or upgrade the existing curl, worried about breaking existing
system! 

Many thanks in advance, Chris 		 	   		  
_________________________________________________________________
Have more than one Hotmail account? Link them together to easily access both
 http://clk.atdmt.com/UKM/go/186394591/direct/01/
------------------------------------------------------------------------------
Return on Information:
Google Enterprise Search pays you back
Get the facts.
http://p.sf.net/sfu/google-dev2dev
Permalink | Reply | 0 comments |
headers
imin imup | 13 Dec 2009 22:39
Picon

Re: How to install libtorrent after it is built by bjam

I don't expect this is a new issue. any advice?
------------------------------------------------------------------------------
Return on Information:
Google Enterprise Search pays you back
Get the facts.
http://p.sf.net/sfu/google-dev2dev
Permalink | Reply | 2 comments |
headers
Jori Paananen | 14 Dec 2009 12:56
Picon
Favicon

More of redundant traffic with streaming media


Hello Arvid,

This is a continuation to the thread, where you and Milton discussed
about redundant traffic occurring in Milton's streaming media
implementation:

http://sourceforge.net/mailarchive/forum.php?thread_name=5a997a610907070704h45e53bc7g4218d4f9a1b9ff8d%40mail.gmail.com&forum_name=libtorrent-discuss

I too encountered redundant traffic, when testing Milton's
implementation and managed to get an example of it into debugger. Here
is my description of it (the libtorrent release 3508 is still used, so
line numbers correspond to that version):

> Hi Milton,
>
> I caught a case of redundant traffic in debugger! Since it seems, that
> almost every received UNCHOKE message during the session causes
> redundant traffic, I set a breakpoint to policy::unchoked. After the
> initial few UNCHOKEs at the start of the session, I caught an UNCHOKE,
> that was sent by a peer from port 6882 in the same machine as the
> seeder in port 7701 (the debugged application was in a different
> machine).
>
> The .6882 peer log of the debugged Livestreamer looks like this:
> ...
> Dec 09 10:02:16 <== HAVE    [ piece: 3479]
> bandwidth [ 0 ] + 5
> Dec 09 10:02:16 *** ASYNC_WRITE [ bytes: 5 ]
> Dec 09 10:02:16 *** REQUEST_BANDWIDTH [ upload: 5 prio: 3]
(Continue reading)

Permalink | Reply | 2 comments |
headers
Max Rafferty | 15 Dec 2009 00:36
Picon

Re: How to install libtorrent after it is built by bjam

There may be a way to install using bjam depending on which compiler you are
using and which build settings you have chosen (i.e. shared vs static), but
the easiest way is probably to just copy the .lib or .dll you built into the
appropriate location. If your using visual studio this is "../visual
studio/vc/libs" or something simmilar. I believe with the dll you can also
get by by simply having libtorrent.dll in a system path folder (I use
windows, so ../windows/system32 works for me).

Hope this helps some,
Max Rafferty

On Sun, Dec 13, 2009 at 4:39 PM, imin imup <iminimup <at> gmail.com> wrote:

> I don't expect this is a new issue. any advice?
>
> ------------------------------------------------------------------------------
> Return on Information:
> Google Enterprise Search pays you back
> Get the facts.
> http://p.sf.net/sfu/google-dev2dev
> _______________________________________________
> Libtorrent-discuss mailing list
> Libtorrent-discuss <at> lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/libtorrent-discuss
>
------------------------------------------------------------------------------
Return on Information:
Google Enterprise Search pays you back
Get the facts.
http://p.sf.net/sfu/google-dev2dev
(Continue reading)

Permalink | Reply | 1 comment |
headers
arvid | 15 Dec 2009 13:34
Picon
Picon
Favicon

Re: More of redundant traffic with streaming media

Quoting Jori Paananen <jori.paananen <at> vtt.fi>:
> You mentioned in previous discussion:
> 
> > If this happens consistently throughout the download, it would
> > appear to be a bug. It can still happen if most of your peers have
> > the exact same pieces over an extended period of time. If you can't
> > download anything new from a peer (because it doesn't have any other
> > pieces), libtorrent will request something that has been requested
> > from another peer.
> 
> In the streaming application, all the peers probably have mostly the
> same pieces, at least in networks, where delays are short. So could it
> be, that the construct should not be used with streaming media?  What
> are the consequences disabling it?

I will have to look this up in the code and catch up on the previous thread to
be sure, but off the top of my head, I believe the logic you are referring to
is where a "busy" block is picked if no other block can be picked from a peer.

The main reason why this logic is there is instead of the traditional end-game
mode. This prevents downloads to get stuck at the end because of a few slow
peers blocking the last pieces to finish.

I believe the proper fix would be to somehow limit the number of redundant
requests that are sent.

--

-- 
Arvid Norberg

------------------------------------------------------------------------------
(Continue reading)

Permalink | Reply | 1 comment |

Gmane