* copyright (C) 1984-2019 merrill consultants dallas texas usa

Change 05.69 In these sample reports, transactions with blanks for

Jun 9, 1987 OPERATOR were not counted in some reports, because by

ANALCICS default PROC SUMMARY treats a blank value of a CLASS

variable as a missing value. These transactions will

be counted if you add the MISSING operand to line 50:

PROC SUMMARY DATA=CICSTRAN MISSING;BY APPLID;

Thanks to Jim Enia, Allied Signal, USA.
Change 05.68 Added support for ACF2 user (default=196) SMF record.

Jul 21, 1987 Ten ACF2... data sets are created. The ... suffix of

EXACFAR data set name is the same as the parameter name in the

EXACFCR ACF2 Utilities Manual, page 74 (1/15/85 revision). No

EXACFDR reports are written or planned. Half of the data sets

EXACFER have been validated with actual data. Only Release 4.0

EXACFJR and later ACF2 releases will be supported.

EXACFNRA

EXACFNRB ACF2 is a product of UCCEL Corporation, maintained by

EXACFPR their Chicago Development and Support Center.

EXACFTR

EXACFVR

IMACACF2

TYPEACF2

VMACACF2

Thanks to Jill Raudabaugh, Whirlpool Corporation, USA.

Change 05.67 Changed references to SMF59TID to correct variable

Jun 4, 1987 TRANTYID to eliminate "uninitialized variable" note.

VMAC59

Thanks to John Mueller, ARMCO Steel, USA.

Change 05.66 Support for IMS 2.1 log record processing and a major

Jun 4, 1987 redesign of the logic for building IMSTRAN. Previously

EXIMSOUT the algorithm threw away transactions that were not

IMACIMS perfectly matched. Now, an observation will be placed

TYPEIMS in IMSTRAN if a '07' IMS log record is found, even if

VMACIMS not all of the other possible log records exist. This

provides complete accounting for the existence and the

resource accounting of the transaction, but may cause

missing value for response time and some time stamps.

Since IBM does not put a version number in the IMS log

you must change the _IMSVERS macro (which is defined

in member IMACIMS) from its default of 1.3 if you wish

to process IMS 2.1 log records. This re-design has

been moderately well validated by comparison with data

from the IMF product, but as with any new design (and

especially with the undocumented state of the IMS log)

you should be alert for any idiosyncrasies. This

change includes and superceedes Change 5.46. The code

in EXIMSOUT which formerly replaced IMSCPUTM with the

average cpu time per transaction has been eliminated.

Now, IMSCPUTM is the actual message region CPU time.

Thanks to John D. Pike, American Airlines, USA.

Change 05.65 Support for MVS 2.2 changes as announced in the MVS/XA

Jun 1, 1987 Conversion Notebook, Volume 2 (GC28-1411-0). Untested.

1. VMAC30. DIVRREAD, the 'ASID TOTAL*DIV REREAD*COUNT' is

added to the TYPE30_V, TYPE30_4, TYPE30_5 and TYPE30_6

data sets to report the new DIV (Data in Virtual) use

at the interval, step, and job level of detail.

2. VMAC7072. Three new variables added to TYPE72 data set

ACTFRMTM='ACTIVE*FRAME*TIME'

PERFACCT='PERFGRP*SET FROM*ACCOUNT?'

PGPAGEIN='PAGE INS*BY THIS*PERF GROUP PERIOD'

Note that ACTFRMTM and PGPAGEIN report memory and page

in counts by each period of each performance group!

3. VMAC90 and FORMATS. New subtype 17 (SET PFK) added to

the TYPE90 data set.

Change 05.64 Variable UCBTYPE was increased from 4 bytes to 5 bytes

Jun 1, 1987 to eliminate truncation of the last byte. Even though

VMAC21 the field is only 4 bytes, because it is stored as a

VMAC64 numeric in SAS, 5 bytes are required for resolution of

VMAC74 all four hex bytes. The change to a 4-byte character

VMAC75 variable was considered, but would be catastrophic on

any report program now using UCBTYPE. If you combine

your daily, weekly and monthly PDBs as recommended in

Chapter Thirty-Five using SET statements, there will

be no impact. However, if you use PROC APPEND on the

TYPE21 or PDB.TAPES, TYPE64, TYPE74, or TYPE75 data

sets, you will encounter an error message due to the

change in length. You could specify FORCE on the PROC

APPEND statement to force the joining of the old and

new, but that will keep the old length of 4 bytes. To

use PROC APPEND without error, you must re-create the

BASE= data set(s) with length 5:

DATA BASE.xxxx; LENGTH UCBTYPE 5; SET BASE.xxxx;

and then the PROC APPEND will succeed. (These are

some of the reasons MXG does not recommend the use of

PROC APPEND.)

Note: UCBTYPE is not a commonly needed variable, and

since it has never been right, probably is not used in

your report programs. DEVICE is decoded from UCBTYPE

and should be used instead.

Thanks to Bill Mullen, BGS, USA.

Change 05.63 TYPE30 variables JINTTIME,INTETIME,JTRMTIME,JINITIME,

Jun 1, 1987 SELAPSTM,JELAPSTM, and INTRVLTM are now created before

VMAC30 any MXG exits are taken. Formerly, they were created

just before the output exit of their respective data

set.

Thanks to Bill Mullen, BGS, USA.

Change 05.62 RACF Type 80 record with a zero length data segment

Jun 1, 1987 caused STOPOVER abend. Insert new lines 70.1, 70.2 and

VMAC80 72.1 after lines 70 and 72 in VMAC80:

@; 70.1

IF 0 LT RACFDLN LE 200 THEN INPUT 70.2

INPUT RACFDATA $CHAR200. @; 72.1

Thanks to Chuck Hopf, Dean Witter Reynolds, USA.
Change 05.61 Untested assembly code for the tape mount monitor. Do

May 21, 1987 not use, as it has not been tested.

ASMONTAP
Change 05.60 JCL errors were not kept in PDB.JOBS from JES2. Code

May 20, 1987 to delete the multiple JES2 purge records in an NJE

BUILDPDB environment also deleted the JCL error's purge record.

1.Add variable INREASON to the _PDB26 keep list in

line 37.

2.Replace line 222:

IF SYSEXEC GT ' ';

with these two lines:

IF INREASON='SR' OR

(INREASON='JR' AND SYSEXEC LT ' ' THEN DELETE;

Thanks to M. N. Woolley, Dept of Finance, AUSTRALIA

Thanks to Cathy Ellis, Dept of Finance, AUSTRALIA

Change 05.59 TYPE64 only contains data on the first 3 extents of

May 20, 1987 a VSAM entry. This change creates an additional data

EXTY64X set TYPE64X which contains the information on all of

VMAC64 the extents. (TYPE64 is unchanged and still will have

the data on the first three extents).

Thanks to Willie Antman, Federal Deposit Insurance Corp., USA.

Change 05.58 This change makes full use of the relocatable record

May 20, 1987 length information to protect MXG from any IBM changes

VMAC7072 to the performance group data. Insert two new lines

after line 1169:

SKIP=LENPGPS-56;

IF SKIP GT 0 THEN INPUT +SKIP @;

This change replaced by change 5.65, June 1, 1987.
Change 05.57 Support for NETVIEW changes to the type 39 record.

May 20, 1987 Without this change, some variables will be missing.

VMAC39 1.Change line 206 from:

IF VERSNLDM GE 000DX THEN INPUT

to read:

IF VERSNLDM GE 000DX OR PRODUCT='NETV' THEN INPUT

2.Change line 241 from:

IF VERSNLDM GE 000DX THEN INPUT

to read:

IF LENSCS GE 118 THEN INPUT

Thanks to John Fleig, Rochester Gas and Electric, USA.
Change 05.56 Devices offline at the end of the interval and any

May 20, 1987 device with CMBINVLD='Y' (indicating invalid data

VMAC74 from the hardware monitor in the channels) had none

of their utilizations calculated, but had some data

carried forward from the previous device. Delete

line 356 and line 379 so that utilizations will

always be calculated for all devices.

Thanks to MP Welch, Sisters of Charity of the Incarnate Word, USA.

Change 05.55 IBM's Report Management & Distribution System,

May 19, 1987 RMDS FDP creates a (default) type 217 SMF record,

EXTYRMDS which is supported by these new members. Data set

IMACRMDS TYPERMDS is created.

TYPERMDS

VMACRMDS

Thanks to Chuck Hopf, Dean Witter Reynolds, USA.
Change 05.54 RACF records with identical SMFTIME values could

May 18, 1987 not be ordered as they occurred. A new variable,

VMAC80 RECIDNO was added to the KEEP= list and created

ANALAUDT with one new line inserted after line 45 in VMAC80:

RECIDNO = _N_ ; 45.1

ANALAUDT was also modified to use RECIDNO in creating

the Auditor reports.

Thanks to David D. Anderson, U.S. Bancorp, USA.

Thanks to Jim Bentley, U.S. Bancorp, USA.
Change 05.53 1.IDMS Performance Monitor (formerly RTE) records

May 18, 1987 now decode the RT1 variable data area (ERUS Batch,

VMACRTE ERUS CICS, and DC information). Ten new variables

are created: RT1DCTC RT1CDLT RT1DCUI RT1DCPN

RT1EBJN RT1EBNAF RT1EBALN RT1EBAFN RT1EBPN and

RT1MSSEV, due to Rod. This member will be

renumbered in MXG Version 5.

2.IDMS-PM records were changed incompatibly in

IDMS Version 10.1, and many new data segments now

exist in the RTE record, based on test data sent

by Dennis. The following changes are required to

process IDMS Version 10.1 RTE records:

a. Insert one new line after line 141:

IF RT1HDRVN EQ 1 THEN DO; 141.1

b. Insert five new lines after line 216:

@; 216.1

IF RT1HDRTL EQ 252 THEN INPUT 216.2

RT1MSRS2 $CHAR4. 216.3

@; 216.4

ELSE INPUT 216.5

c. Insert five new lines after line 220:

END; 220.1

ELSE DO: 220.2

SKIP=RT1HDRTL-16; 220.3

INPUT +SKIP @; 220.4

END; 220.5

Once Cullinet provides the formats of the PMIM

record types, they will be supported by MXG.

See Change 5.95.

Thanks to Rodney L. Reisch, General Electric Silicon Division, USA.

Thanks to Dennis Pugh, Browning Ferris Industries, USA.
Change 05.52 Several minor errors in the VM/Account data sets are

May 15, 1987 corrected by this contribution.

TYPEVM a. Variable TERMADDR should be read in with $CHAR4. in

lines 237 247 and 256 (instead of PIB4.).

b. Variable MINIADDR should be $CHAR4. instead of PIB4.

in line 249.

c. Variable MINIADDR should be $CHAR3. instead of PIB4.

in line 259.

d. Add new line after line 224 (1077052512 is the value

when four blanks are read with a PIB4. format).

IF BLOCKS=1077952512 THEN BLOCKS=.; 224.1

Thanks to Jan Van Lent, Fokker BV, NETHERLANDS.

Change 05.51 A fine set of DB2 reports, similar to those produced

May 14, 1987 by the DB2PM product, were provided by Tony. This

ANALDB2R member generates the reports, with selection by DB2

system as well as reporting time selection as shown

in the comments at the beginning of the member.

Thanks to Tony Shaftel, Farmers Insurance, USA.

Change 05.50 TYPE6 data set variable CONTRIND could not decode the

May 14, 1987 multiple bit conditions which sometimes occur. This

VMAC6 variable has been replaced by nine one-byte variables,

CONTRIN0 through CONTRIN9, containing a "Y" if the

condition described by the variable's label did occur.

This also takes less storage (9 bytes versus 22). Most

sites do not actually used these flags which identify

certain operator actions to the print file.

Thanks to P. J. Lines, Centre-File Limited, ENGLAND.
Change 05.49 IBM documentation error caused variable DVQUQCNT in

May 14, 1987 VMON602 data set to be incorrect. The variable is

VMACVMON only one byte for all releases of VM. Delete lines

2487 thru 2490.

Thanks to Steve Glick, Southern Methodist University, USA.
Change 05.48 TYPE78PA (Private Area virtual storage of monitored

May 14, 1987 jobs) variables suffixed with 9 were not calculated

VMAC78 as averages. Replicate lines 1638 through 1645 and

then change both occurrences of 4 in each line to 9.

Thanks to Rodney L. Reisch, General Electric Silicon Division, USA.
Change 05.47 VM Account record type 07 (VCNA) caused a STOPOVER SAS

May 14, 1987 error condition. Line 157

TYPEVM INPUT

should be change to

INPUT @9 ACCOUNT $CHAR8.

Thanks to Jim Gilbert, Texas Utilities, USA.

Change 05.46 IMS transactions were not properly assembled when the

May 14, 1987 record counter for the log is reset. It used to be

IMACIMS reset with each new log tape, but it is not clear now

TYPEIMS exactly when the reset occurs. The variable IMSTAPNO

VMACIMS (which was in the original European code but removed

by mistake) has been restored to resolve the reset.

There are many lines changed.

June 3, 1987. Change 5.66 superceeds this change.

Thanks to Lee Adams, United Telephone System, USA.
Change 05.45 Reading type 78 records from the VSAM SMF file causes

Apr 25, 1987 a STOPOVER abend. (There is no problem with reading

VMAC78 the normal dumped BSAM SMF file.). To read the type 78

VSAM format data, add +OFFSMF to lines 1668, 1695, and

1731. (OFFSMF is set by the _SMF macro in VMACSMF and

is 4 if a VSAM file is being read, 0 otherwise.)

Thanks to Wing Louie, Metropolitan Life Insurance, USA.
Change 05.44 DB2 Release 1 fields QXQUERY and QXALTBF were changed

Apr 25, 1987 in Release 2 and are now QXSELECT and QXFETCH in MXG.

VMACDB2 (QXALTBF was reserved, and QXQUERY was split into the

two query types SELECT and FETCH.) Change all five

occurrences of QXQUERY to QXSELECT and change the

label to read "SELECT*STATEMENTS". Change all five

occurrences of QXALTBF to QXFETCH and change the

label to read "FETCH*STATEMENTS".

Thanks to A. Williams, Shell U.K, ENGLAND.

Thanks to Richard Whitnable, State of Wisconsin, USA.

Thanks to Martha Hall, Metropolitan, USA.
Change 05.43 QTXA... variables in DB2ACCT are missing because the

Apr 25, 1987 length of the lock segment was changed, causing MXG to

VMACDB2 not read them in. Line 1238 should be changed from

ELSE IF LENLOCK GE 28 THEN

to read

ELSE IF LENLOCK GE 24 THEN

Thanks to Tom Frankle, Chemical Bank, USA.
Change 05.42 More than 200 bytes of user data (counter, clocks, or

Apr 21, 1987 characters) added to the CICSTRAN data caused the MXG

VMAC110 error message "data format has changed". MXG will now

input the first 200 bytes and skip over any excess.

The member IMACICUS can be used to actually decode

any local counters. (The optional DL/I counters are

decoded in IMACICDL.)

Thanks to Shawn Weikel, CITIBANK, USA.

Change 05.41 Reserved Change Number.

Mar 19, 1987

Thanks to Jon Sahl, New York Life, USA.
Change 05.40 Durations and counts for CICSEXCE exceptions were not

Mar 16, 1987 being calculated for CICS 1.7 exceptions. MSWAITTM,

VMAC110 TSWAITTM, and VSWAITTM are set to ENDTIME-STRTTIME.

The existence of the exception is still the important

event. (VSAM waits are usually due to insufficient

LSR buffers).

Thanks to Tom Wiebe, NERCO, USA.
Change 05.39 Preliminary re-write of SYNCSORT support, will become

Mar 11, 1987 VMACSYNC when tested.

XMACSYNC

Thanks to Jill Raudabaugh, Whirlpool Corporation, USA.

Change 05.38 Record subtype, SUBTYPE, is now created in _SMF macro

Mar 11, 1987 and is available to exit IMACFILE, for those records

VMACSMF which contain a subtype. END=ENDOFSMF and an include

for IMACFILE were added to _JOURNAL macro (for type

110 records in CICS journal format) so that it is now

consistent with the _SMF macro.

Thanks to Mark Nelson, UCCEL, USA.
Change 05.37 Revised installation instructions for the IDMSSTAT

Feb 2, 1987 program (which writes SMF records from IDMS) were

IDMSEXIT added. No change to the ASM source code was made.

Thanks to Ron Szczepanski, Whirlpool Corporation, USA.

Thanks to Keith Stout, Municipality of Anchorage, USA.
Change 05.36 A set of ten reports for the EDP Auditor, from RACF

Feb 2, 1987 TYPE80 data, was provided by its author, and with

ANALAUDT minor revisions was added as member ANALAUDT.

Thanks to David D. Anderson, U.S. Bancorp, USA.

Change 05.35 Variable VTAMCPUI, the CPUID*FROM*SYSTEM, at the end

Jan 29, 1987 of this VTAM SMF Accounting Record is now created by

VMAC128 MXG.
Change 05.34 To protect users from accidentally truncating the WORK

Jan 28, 1987 variable (which defines worktype and is expected by

IMACWORK RMFINTRV to be $CHAR5.), INFORMAT WORK $CHAR5. was

added. This is a minor change.

Thanks to Brian Cooper, Franklin Mint, USA.
Change 05.33 Announcement of the 3090-600 suggests there really is

Jan 27, 1987 going to be a 3090-800. TYPE70 data set was modified

VMAC7072 to support nine engines (numbered 0 thru 8 inclusive).

The eight CPU-specific variables (CAIn, CPUSERn, VFONn

CPUWAITn, IORATEn, PCTCPBYn, PCTTPIn, VFAFFTMn) now

exist for n=0 thru 8. This change supercedes change

5.1. These variables are not typically important; most

sites won't need to know these CPU-specific values.

The overall system measures related to these values

(CPUWAITM, IORATE, PCTCPUBY, PCTTPI) are correct with

or without changes 5.32 and/or 5.1.
Change 05.32 Support for the Online Business System WYLBUR Session

Jan 26, 1987 SMF record was added. One record per WYLBUR session

TYPEWYLB is written containing CPU, I/O and counts of some

VMACWYLB activities.

IMACWYLB

EXTYWYLB
Change 05.31 Support for the Cache RMF Reporter (IBM Program Number

Jan 25, 1987 5798-DQD) SMF records (default 244, 245) which invokes

TYPECACH AMS LISTDATA to extract counters for Cache DASD 3380

VMACACHE control units (models 11,21,13,and 23) was added. This

IMACACHE code was originally written in Germany and contributed

EXCAPAG to MXG. The code was revised and consolidated into the

EXCACHE members indicated, and has been tested with type 245

SMF records.

Thanks to ???, DZ/SP-CM Gotaher Verischerung VVAG, GERMANY.

for the original code,

Thanks to Kirby Kern, Commercial Federal S & L Omaha, USA.

for test data.
Change 05.30 Format MG078CV., which is used in many MXG data sets

Jan 24, 1987 to convert bytes to nnnK or nnnM printout was refined

FORMATS so that 16777216 bytes prints as 16MB (before this

change, the value had to be 16777222 before the format

changed from 15M to 16M). Correct this minor error by

changing .0009766 to .0009765625 in line 163, and by

changing .000000953674 to .00000095367432 in line 164.
Change 05.29 Support for TSO/MON Version 5.0 coded. Accounting

Jan 23, 1987 variables added to TSOMCALL and TSOMSYST. A new data

VMACTSOM set, TSOMDSNS captures data set names and members

which were accessed by commands captured in TSOMCALL,

if new TSO/MON option to do so was enabled. MXG 4.4

supports TSO/MON Version 5.0 without error, but if you

want the data new in TSO/MON Version 5.0 before we

ship MXG Version 5, call us. Too many lines changed

to provide here.
Change 05.28 The equation for PCTCUBSY (percent control unit busy

Jan 14, 1987 in TYPE78CF for 3090s) is incorrect in the RMF manual

VMAC78 on page 374. The minus is actually a plus.

The value of AVGCUHQL (CU-HDR queue in TYPE78CU) was

incorrectly calculated. The equation was corrected.
Source statement Line number
PCTCUBSY=100*PCTCUBSY/CHPIDTKN+PCTCUBSY); 1724

IF NRCUHREQ GT 0 THEN AVGCUHQL=(NRCUHENQ-NRCUHREQ)/NRCUHREQ; 1748

ELSE DO; 1749

NRCUHENQ=.; 1749.1

AVGCUHQL=.; 1749.2

END; 1749.3

Change 05.27 After each statement in which INVALID,VARIED,ONLINE,

Jan 14, 1987 LCI, and IDWRONG is set equal to 'Y', add a statement

VMAC73 ELSE XXXXXXXX=' '; where XXXXXXXX is the name of

the variable being set to 'Y'. Without these new ELSE

statements, once any of these variables were set to a

value 'Y', the value remained 'Y' for the rest of the

observations created from that type 73 record.

Thanks to Paul Larson, Marine Bank Services, USA.

Change 05.26 New variables SWAP EXTAUX LOGAUX LOGEXT PHYAUX and

Jan 14, 1987 PHYEXT are created, each containing the total swap

VMAC71 rate for tha swap transition. This total is created by

summing all eleven swap reason codes for a particular

swap transition. Insert after line 742, these lines: ,
SWAP =SUM(SWAPAS, SWAPDW, SWAPEX, SWAPNQ, SWAPNS, SWAPRS,

SWAPTI, SWAPTO, SWAPUS, SWAPVR, SWAPWT);

EXTAUX=SUM(EXTAUXAS,EXTAUXDW,EXTAUXEX,EXTAUXNQ,EXTAUXNS,EXTAUXRS,

EXTAUXTI,EXTAUXTO,EXTAUXUS,EXTAUXVR,EXTAUXWT);

LOGAUX=SUM(LOGAUXAS,LOGAUXDW,LOGAUXEX,LOGAUXNQ,LOGAUXNS,LOGAUXRS,

LOGAUXTI,LOGAUXTO,LOGAUXUS,LOGAUXVR,LOGAUXWT);

LOGEXT=SUM(LOGEXTAS,LOGEXTDW,LOGEXTEX,LOGEXTNQ,LOGEXTNS,LOGEXTRS,

LOGEXTTI,LOGEXTTO,LOGEXTUS,LOGEXTVR,LOGEXTWT);

PHYAUX=SUM(PHYAUXAS,PHYAUXDW,PHYAUXEX,PHYAUXNQ,PHYAUXNS,PHYAUXRS,

PHYAUXTI,PHYAUXTO,PHYAUXUS,PHYAUXVR,PHYAUXWT);

PHYEXT=SUM(PHYEXTAS,PHYEXTDW,PHYEXTEX,PHYEXTNQ,PHYEXTNS,PHYEXTRS,

PHYEXTTI,PHYEXTTO,PHYEXTUS,PHYEXTVR,PHYEXTWT);

MXG Version 5 goes further and sets all seventy two of

swap rate variables to missing if there are no swap

transitions for that swap reason code. This will cause

PROC PRINTs and MEANS to highlight only the transition

and reason codes which actually occurred.
Change 05.25 Variables VECTORON and VECONSAM were replaced with

Jan 14, 1987 VFON0 thru VFON3 and VFAFFTM0 thru VFAFFTM3 to track

VMAC7072 Vector Facility Online and Affinity by processor.
Change 05.24 Variable ABNDRSNC (abend reason code) was added to

Jan 7, 1987 TYPE30_4 and TYPE30_5 KEEPs, and new lines inserted

VMAC30 after lines 147, 326, and 487:
Source statement Line number
ABNDRSNC='ABEND*REASON*CODE' 145.1

ABNDRSNC $HEX8. 323.1

IF LENCOMP GE 8 THEN INPUT ABNDRSNC $CHAR8. @; 483.1
Change 05.23 Class 1 record with LENDATA=0 causes STOPOVER ABEND

Jan 7, 1987 because $VARYING is not smart enough to handle zero.

VMACVMON Lines 2087-2088 should be changed to:
Source statement Line number
INPUT @19 LENDATA PIB1. @; /*MN10YCNT*/ 2087

IF LENDATA GE 1 THEN 2087.1

INPUT @20 DATALINE $VARYING128. LENDATA /*MN10YI0*/ 2088

Thanks to Bill Cohen, Drexel Burnham, USA.

Change 05.22 OBJVALUE was ten times too large. Line 288 should be

Jan 7, 1987 OBJVALUE PIB4.1 instead of PIB4.

VMAC39
Change 05.21 The 3480 data fields were never read in, due to wrong

Jan 4, 1987 test value in line 68. New variables DEVICE and OPEN

Jan 26, 1987 are added. TAPUNSER was incorrectly decoded as numeric

ANALESV but is actually a character string in hex format. To

VMAC21 avoid conflict, new character variable TAPCUSER now

replaces TAPUNSER. For 3480s TAPCUSER contains the

last four digits of control unit serial number and the