On May 12, 2011 10:02 AM, "Brian Utterback" <brian.utterb...@oracle.com> wrote: > On 05/12/11 09:53, Vladimir Kotal wrote: > As to the second, two ideas come to mind. First, if the data packet was > delayed due to congestion or Nagle, so that the close syscall was > processed before the data was sent. The second, if the data was sent but > the packet was lost, the retransmission could combine the data and FIN.
So pause the app after connecting, add a -blackhole route for the peer, have the app send data and close the socket, then remove the blackhole route. Hopefully TCP won't just retransmit but send a FIN+data. Nico --
_______________________________________________ networking-discuss mailing list networking-discuss@opensolaris.org
