RE: Sniff TCP/IP traffic on iSeries?

Shutdown():
Usage Notes
Issuing a shutdown() with a how parameter of 0 causes any new data
received for the socket to be discarded. Any input functions for this
socket complete with a 0, meaning that end-of-file has been reached. On
a BSD implementation, if the socket is being shared across multiple
processes, any blocking input operations are deblocked by this action.
However, the OS/400 sockets implementation of shutdown() does not cause
these blocked functions to be deblocked.


Issuing a shutdown() with a how parameter of 1 results in all output
functions being failed with an error of [epipe]. The process issuing the
output operation will receive a synchronous sigpipe signal. This also
sends a normal close sequence to the partner program. Receive operations
issued by the partner program receive a return value of 0 once all
previous data has been received. On a BSD implementation, if the socket
is being shared across multiple processes or threads, any blocking
output functions are deblocked with a return value of -1 and an error
code of [epipe]. However, the OS/400 sockets implementation of
shutdown() does not cause these blocked functions to be deblocked.


Issuing a shutdown() with a how parameter of 2 results in the actions
listed for a how parameter of 0 being performed first, followed by the
actions listed for a how parameter of 1.


Issuing a shutdown() on socket connected through a SOCKS server is not
supported.


When you develop in C-based languages and an application is compiled
with the _XOPEN_SOURCE macro defined to the value 520 or greater, the
shutdown() API is mapped to qso_shutdown98().  


Chris Bipes
Information Services Director
CrossCheck, Inc.

-----Original Message-----
From: midrange-l-bounces@xxxxxxxxxxxx
[mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of David Gibbs
Sent: Monday, October 24, 2005 2:26 PM
To: Midrange Systems Technical Discussion
Subject: Re: Sniff TCP/IP traffic on iSeries?

On Mon, October 24, 2005 4:04 pm, Scott Klement said:
> a) Sometimes they forget to add a CR/LF.  This causes the whole
process
> to stop and wait for the CRLF.
> b) the send() and recv() APIs (or, the read() and write() APIs, which
> work exactly the same) do not send fixed-length data on a stream
socket.

My code uses shutdown() to indicate that the sending is done.  We don't
rely on fixed length or even data terminators to determine when the data
is finished.

david