IBM ZOS V1.12 Manual - page 20
IBM United States Software Announcement
210-008
IBM is a registered trademark of International Business Machines Corporation
20
a high amount of processor time, and analyze for patterns common to looping
address spaces.
• A new Timed Auto Reply Function is planned to provide an additional way for
the system to respond automatically to write to operator with reply (WTOR)
messages. This new function is designed to allow you to specify message IDs,
timeout values, and default responses in an auto-reply policy, and to be able to
change, activate, and deactivate auto-reply with operator commands. Also, when
enabled, it is designed to start very early in the IPL process, before conventional
message-based automation is available, and continue unless deactivated. An
IBM-supplied auto-reply policy in a new AUTOR00 parmlib member that you can
replace or modify is also planned. This new function is expected to help provide
a timely response to WTORs and help prevent delayed responses from causing
system problems.
• XCF Status monitoring will be designed to incorporate information about system-
critical XCF group members that identify themselves, and initiate termination
actions, including partitioning a system from the sysplex, if a monitored member
fails to respond when polled for status or indicates impairment. This function is
intended to help reduce the incidence of sysplex-wide problems that can result
from unresponsive critical components. GRS is planned to exploit these XCF
critical member functions in both ring and star modes. Additionally, GRS will be
designed to monitor key tasks and notify XCF if it detects that GRS is impaired.
• A new Predictive Failure Analysis check is planned to detect and automate the
system's response to tasks that are writing SMF records at unusually high rates.
Another new function, SMF record flooding automation, is designed to allow you
to define a policy for responding to these situations in the SMFPRMxx member
of parmlib, by specifying whether record flooding automation is to be active,
whether operators are to be warned, and the actions to take for specific SMF
record types if record flooding occurs. This is intended to limit the impact of such
problems by allowing less-important data to be discarded while keeping the data
from critical SMF records intact. Additionally, new function is planned for the SMF
dump program (IFASMFDL) to provide additional information to help you develop a
record flooding policy.
• Function is planned for Predictive Failure Analysis (PFA) to allow you to specify
that PFA ignore data related to certain jobs or address spaces when you expect
their behavior to be atypical. This can help you improve the overall accuracy of
PFA checks for logrec, message, and SMF record arrival rates. There are two types
of machine learning, supervised and unsupervised. In z/OS V1.12 PFA will support
both supervised and unsupervised learning. To support supervised learning,
function is planned for PFA to allow you to specify that PFA ignore data related to
certain jobs or address spaces when you expect their behavior to be atypical. By
providing supervision (insight), you can help improve the overall accuracy of PFA
checks for logrec, message, and SMF record arrival rates.
• Four changes to improve the quality of PFA modeling are planned for z/OS V1.12.
PFA will be designed to:
– Capture data when exceptions are issued to help you identify problems
– Use dynamic modeling intervals based on system stability
– Discard the last hour's LOGREC data from before a shutdown
– Monitor smaller increments of common storage assigned to system
• New functions are planned for recovery and termination processing (RTM). These
include a new option on ESTAEX to specify that SPIE or ESPIE exits be superseded
by ESTAEX, a new option on ESPIE to request percolation to RTM, and passing
information about held locks to ESTAE-type recovery routines.
• z/OS UNIX System Services file system processing will be designed to provide
better information when a DISPLAY GRS,ANALYZE command is issued by
identifying itself as the holder of held latches to GRS. This will be intended to
help you diagnose and take corrective actions for latch contention problems that
involve file system processing.
• Information about DFSMSrmm active and queued tasks is planned to be
available via the DFMSMrmm API and via TSO/E subcommand, enabling storage
applications to monitor and act on the available information.
• In z/OS V1.12, RSM and dump processing will be designed to improve capture
performance for large amounts of data during SVC dump processing. This is