Session O323 - MVS/ESA Recovery

SHARE 73
August 21-25, 1989

Frank Rodegeb presented the recovery characteristics of MVS/ESA. It turns out that IBM has done a lot of work on recovery, and the differences between ESA and our little old MVS/370 system are becoming broader by the month.

First we had to endure a ritual incantation and gratuitous overhead display, to-wit:

"The following are trademarks of the IBM Corporation
MVS/SP MVS/ESA PR/SM MVS/XA"
Someday I will give my own presentation at a SHARE meeting, and I will take care to duly mention the owners of all trademarks referred to during my presentation. I can see it all now:
"Indiscriminate installation of OCO software on your MVS/SP
system can introduce trojan-horse programs beyond your control.
By-the-way, MVS/SP is a trademark of the IBM Corporation,
and Trojans is a trademark of..."
He reviewed the primary MVS recovery functions and services (a lot of alphabet soup including RTM, MCH, ACR, APR, MIH and others). Then he talked about the new stuff in ESA. Associated Recovery Routines (ARRs) are like ESTAE, but are available to programs entered through cross-memory program calls. Automatic Spin Loop Recovery (ASLR) tries to get the operator out of the loop when one CPU fails while holding a spin lock. The Disabled Console Communication Facility (DCCF) provides operator communication capability when the master console and its alternate are inop. It replaces the code that used to get control for restartable WAIT states. It uses the system console, and requires a new operator frame - "SCPMSG". Weird.

A new function in ESA is "Component Trace", or CTRACE. It provides a central mechanism for system components to record and analyze their own internal diagnostic traces. IPCS supports CTRACE.

ESA now supports a PAGEDEL command (you have always been able to dynamically add page datasets, but you couldn't get rid of them without IPLing).

ESA fixes a bug in WAIT/POST where VTAM would ABEND with an S202 when a VTAM application shut down. This is a continuing thorn in my side in MVS/370.

IPCS has gained new respect at IBM and is ballyhooed as "the strategic dump analysis tool for MVS/ESA". It now completely replaces AMDPRDMP, and contains a number of new canned routines to validity-check system dumps. SVC dump has gotten a bunch of new operands and can be tailored more completely. GTF has enjoyed a "major performance improvement".

Not much of it applies here. Yet.


Back to session index
Back to index of SHARE meetings
Read the disclaimer