Very slow speeds using linux X64 alpha client (debian variant)

[CyberMonk Titan]

During experiments with the Linux utorrent client I have come across a strange situation:

- One machine (A) running utorrent with the seed of a 500Mb file;

- One machine ( downloading that file in different ways.

Three different scenarios, depending on how B downloads the file, and their outcomes:

- Download using KTorrent, forcing TCP download only: full speed (CPU maxes at over a megabyte per second; KTorrent isn't that efficient, apparently);

- Download using KTorrent, forcing uTP download only: full speed, again;

- Download using utorrent x64 alpha client: 3.5 kbps maximum.

The seeder was started fresh for every single experiment, just like the private tracker I use. I have verified that machine A was the only peer during each download. The same file and same torrent file were used every time. Between some other machines I have been able to get to maximum speeds using only utorrent. Firewalls aren't in play here: the subnets of A and B are completely exposed to each other, and any firewalls would play up with the KTorrent cases as well. Both subnets are on the same university network (same building, as a matter of fact).

I'm not sure what is happening here, but at this point I am assuming an error somewhere in the utorrent code. One possibility my research group (we're researching P2P applications) could come up with was the congestion control.

If you would like to know some more of the situation, or would like me to run some tests: please let me know.

[mcdonald]

- Download using utorrent x64 alpha client: 3.5 kbps maximum.

Try disabling uTP to see if that changes the download rate.

[CyberMonk Titan]

Thanks for picking this up, mcdonald!

I tried to recreate the situation today to try your suggestion. I started the clients using the original setup (same client files, same torrents, same data files, same tracker, same machines, same network). While looking through the options to disable uTP I noticed the unexpected: the file was being transferred at nice speeds.

So: same setup, problem no longer occurs. I've tried again, reversed the transfer, tried some more, but I can no longer reproduce this problem. Alas, I do not control every aspect involved (one of the machines is administered by the university, as is the complete network), so I can't tell whether something has changed beyond my view.

Hence: I can no longer reproduce this problem, but this does not mean it's not there. If anyone can reproduce this, please continue investigation.