Session 2703 - TCP/IP Update for the VPS Administrator
February 22-27, 1998
This was the second of two VPS sessions I went to, hosted by
Beverly Weaver of
Levi, Ray and Shoup.
Truth to tell, the pickings were a little slim during this
time slot, and I came to this session expecting to know
everything about anything she was going to talk about. I've
spent much of my life this last year or so doing TCP/IP
printing, using VPS, RPM, NPF and other alphabet soup.
But the presentation was useful to me, and even lively in spots.
Several other folks in the audience had gone through much the
same shenanigans as I have, and others were contemplating it.
This session talked about updates to TCP/IP processing in the
VPS and DRS product lines. Among my notes:
- There is a fix level in release 6.2 (which fix level we haven't
applied) that finally does away with the requirement to have a
VTAM ACB defined. We don't have any VTAM printers defined
to VPS, yet we still have to define VPS to VTAM. I'm glad they've
finally gotten rid of this particular albatross.
- The staging file in VPS doesn't get used until you exceed the
QBUFSIZE specification in the printer definition member. Because
VPS doesn't use SAPI yet,
and the external writer interface to JES is single-threaded, VPS
destages one job at a time from the JES2 spool, and respools
to an intermediate OS dataset - one per printer.
- LPD printers must disconnect after each dataset; it's
in the protocol specification. VPS is unable to send an entire
JES2 OUTGROUP (read: job) as an LPD dataset because of difficulties
in handling JES2 checkpoints.
- A comment on performance: VPS exit #1 can kill your
system. Don't do any heavy processing in it. Don't use GETMAIN!
- Another audience member confirmed this for me: the "sockets"
protocol that VPS/TCPIP supports is most probably the same "telnet"
protocol that is supported by
Dave Brooks' RPM client.
- Have to check: Bev says that I should be setting RELREQ=I for
socket clients. Also to check our fix level - 6.2.214 has some
TCP/IP reconnection fixes.
- There exists an exit in VPS that can be used to send a PCL
"flush pages" command before the PCL ECHO at the end of an
output group. We probably should be using it.
Back to session index
Back to index of SHARE meetings
Read the disclaimer