Watch that dark ages stuff, Chuckie. By the time I saw that problem, I was
no longer keying on an 029 or maybe 026. I had graduated to a 3278 or
3279, altho I did then and still do carry 5081 cards in my pocket.
Jim
At 01:31 PM 6/9/2006, you wrote:
In the Dark Ages (stone knives and bear skins), CTCs were problematic for
SA programs because the interrupts are generated by the system on the
*other* end. The various SA programs that still depend on an I/O
interrupt in addition to, or instead of, LOADPARM were changed in the
Middle Ages (represented by the invention of Sense ID) to examine more
closely the cyberDNS of interrupting device. 3088s exacerbated the
problem because it was so easy to fully interconnect the attached systems.
Or someone decided that *now* would be good time to ENABLE one of the
adapters. :-)
For a 3088/CTC, the channel reset only affects *this* system's I/O status.
The other side can still restart the link and annoy your SA program.
If you find an SA program that gets confused by random interrupts and
cannot be overridden by LOADPARM, you should probably call it in. With
the XA I/O architecture there are all kinds of interrupts that can come in
that have nothing to do with "a tape was mounted" or "somebody flipped the
test/normal switch on the 3278" or "Attention was pressed on the 3215".
Alan Altmark
z/VM Development
IBM Endicott
Jim Bohnsack
Cornell Univ.
(607) 255-1760
