Plüm, Rüdiger, VF-Group wrote:
Without the patch the container side will presume the next packet
(can be new request or ping) is first body packet.
So with the patch the connection to the container can be reused and
the state of the connection in the container is that it waits for a
new request on this connection?
Yes.
It's quite straightforward. Per AJP protocol spec,
whenever there is C-L > 0 header present the container
expects two packets:
a) request itself and b) the first body packet.
Additional body packets are then requested by container
by GET_BODY_PACKET message. If the first body packet
is zero-length or container decides it doesn't
wish to read rest of the body it terminates the request
(without closing the physical connection of course)
Regards
--
^(TM)
