[Libtorrent-devel] Incorrect bandwidth calculation?
dgerow at afflictions.org
Mon Feb 2 23:47:07 UTC 2009
Josef Drexler wrote:
: >> /Is/ it
: >> actually transferring data? (My cacti box died last week, and I've
: >> not yet had a chance to replace it.)
: It's definitely network traffic, not necessarily torrent data though.
: If you have a network monitor it should also show this amount of
: traffic, actually somewhat higher due to protocol overhead (ACKs and
: retransmissions included).
Figures that my network monitor should die just before I develop a need for
it. It'll take me a few days to get the replacement network monitor in
place; I'll have a closer look at network traffice then.
: I'd bet on it being DHT traffic which for a stable node can indeed
: reach many KB/s after a while. Check the log, it shows how many bytes
: DHT is transferring over a 2-hour period, that should match the
: numbers you're seeing. On average DHT as a whole won't use more
: bandwidth than a regular peer though, unless there are no busy
: torrent peers in which case it's allowed to use all left-over bandwidth.
DHT seems to be a viable culprit The numbers it provides for reading and
sending (roughly 30 and 60MB) do work out to the sustained traffic numbers
I'm seeing. I'm going to have to read up on DHT, though, as all of my
active torrents are done downloading (i.e. they're all seeding), so I
wouldn't have thought I'd have much downstream bandwidth utilization.
More information about the Libtorrent-devel