Barbara Nitz wrote:
I've had some seemingly good results by issuing a cancel three times,
waiting about a minute between each command. Almost like each cancel
takes out a layer.
It does. Kinda sorta. Jim, please step in, because I am even more fuzzy on
this. Here goes:
In the ninetees RTM was rewritten to avoid the address space hang problems
after cancel (usually deadlocks between terminating TCBs). From then on when an
address space wouldn't come down, the bottom tcb is always the one holding off
shutdown, as the top ones will not terminate anymore until all daughters are
gone.
In some cases repeated CANCELs go and cause re-entry to RTM. RTM then knows to
'escalate' the termination, taking away control from the recovery routine that
is just supposed to be running. This doesn't always work, as the cancel command
isn't necessarily propagated down to the bottom tcb a second or third time. I
am not really clear anymore *why* this sometimes works and sometimes doesn't.
My experience with the callrtm program has always been that I needed to run the
same program several times to see a result. Sometimes even in combination
callrtm'ing the top application tcb and the bottom tcb.
So yes, in a way each cancel command takes away a 'layer'.
RTM parallel detach was a major enhancement in MVS/ESA V5. Incredibly,
it has never been documented in the books! (Should I be surprised?) If
one experiments by placing a never-ending WAIT (or WTOR or similar) into
an ESTAE routine, you'll find out in a hurry what it's all about.
I remember learning how parallel detach worked by spending an afternoon
reading APAR descriptions on IBMLink -- only because IBM had to fix some
issues with it.
To learn how it works, search for "RTM Parallel Detach", read the APAR
descriptions, and then do some follow-up searching based on additional
keywords and field names you find in those APARs.
--
Edward E Jaffe
Phoenix Software International, Inc
5200 W Century Blvd, Suite 800
Los Angeles, CA 90045
310-338-0400 x318
[EMAIL PROTECTED]
http://www.phoenixsoftware.com/
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
