For performance problems that are not related to access
path issues, collect a description of the problem symptoms, SMF data, RMF data, and a supervisor call (SVC)
dump.
Before you begin
To diagnose performance problems, you must have the appropriate
authorization to collect data from different z/OS® components.
Procedure
To collect diagnostic data for general performance
problems:
- Write a description of the problem and include answers
to these questions:
- What are the names of packages, user IDs, and subsystems
that are experiencing performance problems?
- If a job abended, what were the abend codes?
- Has anything recently changed in the environment that
caused the problem?
- How do you reproduce the problem?
- Collect SMF type 100-102 trace records by issuing Db2 START
TRACE commands for both a successful and an unsuccessful run.
For example, start both accounting and statistics traces:

-START TRACE(ACCTG) CLASS(1,2,3,7,8) DEST(SMF)
-START TRACE(STAT) CLASS(1,3,4,5,6) DEST(SMF)

Separate accounting data from other transactions to prevent accumulation. In a Resource
Recovery Services (RRS) or distributed data facility (DDF) attach environment, set the ACCUMACC
subsystem parameter to NO.
- Collect RMF Monitor
I and RMF Monitor II SMF type
70-79 trace records by issuing RMF commands
for both a successful and unsuccessful run:
- Start RMF Monitor
I:
- Start RMF Monitor
II:
MODIFY RMF, START session-id
In
this example,
session-id is two alphanumeric characters
(not ZZ) to start RMF Monitor
II.
- Issue the z/OS DUMP
command to dump Db2 associated
address spaces:
DUMP COMM=(title),
JOBNAME=(ssnmIRLM,ssnmMSTR,ssnmDBM1,ssnmDIST),
SDATA=(RGN,CSA,SQA,LPA,LSQA,SWA,PSA,ALLNUC,XESDATA,TRT,GRSQ,SUM),END
where
title is the title of the dump
data set and
ssnm is the
Db2 subsystem
name.
Collect service SQL documentation that describes the environment for the SQL statement, as
described in Collecting service SQL documentation.
