Ktorrent Optimization

[D. R. Evans]

adams said the following at 04/19/2007 04:06 PM :
> I realize that there is a "conditioning" time envolved with Ktorrent.
> So I have a T1 line and 24/7 access and I will leave the DVD images
> on after they have completely downloaded.
> 
> Question that I have.  I searched the Internet via google and haven't
> a clear answer as to what, if any optimization, to do locally to
> insure that I have the right ports open for greatest throughput on
> uploads and downloads.  There are close to 100 peers for both distros
> but the download rate of about 5KB/s and estimated download times of
> 44 days or greater doesn't seem right.

>From the BT FAQ:

----

Prior to version 3.2, BitTorrent by default uses ports in the range of
6881-6889. As of 3.2 and later, the range has been extended to 6881-6999.
(These are all TCP ports, BitTorrent does not use UDP.) The client starts
with the lowest port in the range and sequentially tries higher ports until
it can find one to which it can bind. This means that the first client you
open will bind to 6881, the next to 6882, etc. Therefore, you only really
need to open as many ports as simultaneous BitTorrent clients you would
ever have open. For most people it's sufficient to open 6881-6889.

----


I run with 6881 to 6889 open. But I almost never run with more than one
client, so that's probably overkill.

> 
> I'll plot the data as it goes and see if there is a clear trend.
> I would think the more the merrier as to the rates, unless 
> every one else is doing dialup...... :-)
> 

Took me about 2 hours (total) on a fixed wireless connection yesterday
(supposedly 256k up, 512k down, but commonly rather more than that) to
download the 32-bit and 64-bit ISOs. I didn't look closely, but I think I
saw that the download rate was about 3Mbps at one point. So I would expect
your T1 to sit pretty close to 1.5 Mbps once the download got properly
started (it usually takes anything from 10 seconds to maybe five minutes to
stabilise, is my experience, depending on the the other nodes in the swarm).