On Thu, Sep 07, 2006 at 09:30:05AM -0400, Greg Troxel wrote:
> 
> 
> After a timeout, which I suspect is RTO rather than persist, it sends
> a single 32k packet into a 16k window.

Note that 32K is the interface MTU.  The sender is using the interface MTU
as the initial window size, even though the receiver wants less.  Is this
the only bug here?  (The receiver, in this case, should not have such a
small window, but perhaps Amanda has yanked the socket buffer size down;
if not, there's another bug on the receive side, I think.  But the sender
should not send data past the end of the window...)

Thor