allow user to specify source port for outgoing connections

[wfaulk]

Allowing the user to select a source port (or set of source ports) for outgoing connections would be helpful in matching packets for prioritizing traffic.

[charles]

I'm afraid I don't understand what you mean by "source port for outgoing connections".

[bgilmore]

Speaking on behalf of the reporter:

In libtransmission/net.c:110, tr_NetOpen would need to be modified to bind(2) to a sockaddr.

From the ​Advanced Network Settings page on the Azureus wiki:

This sets outgoing connections to a fixed source port. This can be used with full cone NAT routers to reduce the number of NAT table entries and thus prevent crashes on some router models. When the outgoing port is bound to the same as the incoming port that might even solve NAT problems on full cone NAT routers.

note: not endorsing the enh request, just clarifying it

-bg

[wfaulk]

Correct. Right now when Transmission opens a TCP connection to a remote peer, the source port is chosen randomly by the OS. Having Transmission specify it would allow me to match connections based on that port so that my QoS router can throttle those connections. Right now, it cannot, since those connections have a random port on my end and whatever port the remote peer has chose on the other end. And since connections are largely encrypted, they cannot be inspected to determine that they are bittorrent connections.

[charles]

Sounds like a good idea... anyone want to submit a patch?

[charles]

This is disabled by default in Azureus, and the wiki contains this note: "Binding to the same port as the incoming port works under Windows 2k and greater only. Unix users should use a different port. It's not working at all under Mac OS X. " Maybe there are portability problems with this?

[wfaulk]

It need not be the same port. Specifying a different port (or, more likely, set of ports) is a perfectly acceptable solution. The user just needs to know in advance what outgoing bittorrent connections are going to look like.

[charles]

Is this this same as #776?

[charles]

10:29 < Chinstrap> charles_: #776 and #583 are different ways of saying the same thing