On Feb 8, 1:47*pm, Jorgen Grahn <grahn+n...@snipabacken.se> wrote:
> On Mon, 2011-02-07, Washington Ratso wrote:
> > When I have connection tracking enabled in the kernel and use the
> > Space Communications Protocol Standard (SCPS)
>
> * * * * * * * * * * * * * * * * ^^^^^^^^
> The name sounded silly, so I googled it. It's "Space Communications
> Protocol Specifications", which sounds much more serious.
>
> http://en.wikipedia.org/wiki/Space_C...col_Specificat...
>
> > which is a protocol to
> > speed up TCP over satellite
>
> It's many different things. *Do you mean the sender-side modifications
> to TCP, or the alternative to IP, or the FTP extensions, or ... ?
>
> I have never heard of it before so I can't help -- but due to the
> vague information, neither can anyone else at this point.
>
> /Jorgen
>
> --
> * // Jorgen Grahn <grahn@ *Oo *o. * . *.
> \X/ * * snipabacken.se> * O *o * .
It is on the receive side. The call originates on a Polycom
ViewStationFX (10.175.1.3). It then goes through a Linux box
(10.185.1.1), over a satellite link, then through another Linux box
(10.175.2.1), then to a PC (10.175.2.2). While setting up the call, I
can see the "Connect" message is sent from the PC (10.175.2.2) but the
Linux box (10.175.2.1) intentionally drops it, as mentioned above, in
the kernel's connection tracking software.
Note, it is the kernel that is dropping the packet, not the SCPS
software, but enabling the SCPS does cause the "Connect" packet to be
dropped because this does not happen if SCPS is not enabled.
What other details would you like to know?