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

002 - Task Resource Utilization Table

003 - Span Authorization Table

First glance shows lots of interesting fields. If you

have type 38's from Version 1.1, see member SENDDATA and

email me a few hours worth, and I can enhance MXG with

these three new datasets.
Change 15.385 Support for TME 10 Netview for OS/390 1.1 SMF type 37

VMAC37 record IS already included in MXG, as there were no

Feb 18, 1998 changes to the type 37 record, although the table numbers

were changed in documentation: "TME 10 Netview for OS/390

Application Progrmmer's Guide Version 1 Release 1, pub

number SC31-8223-00 (but don't look for that order number

on the publication itself; only on the Readers Comment

Form could I find the pub number!). IBM saves ink???

That publication documents SMF 37, 38, and 39 records.
======Changes thru 15.384 were in MXG 15.09 dated Feb 17, 1998======
Change 15.384 Support for RMF Monitor III VSAM file adds support for

ASMRMFV new record types of CPU, PGP, and ENC, but I have not had

ASMRMFV test data (ASMRMFV, when assembled, converts the RMF III

Feb 17, 1998 compressed files into flat records for TYPERMFV to read),

so output records from ASMRMFV with CPU, PGP, or ENC

enabled needs to be verified with TYPERMFV. In addition,

the support for the CSR records cannot be finished until

we can get a hex dump of the ASMRMFV program when it

tries to read a CSR record. We have created a special

version of ASMRMFV, named ASMRMFVX, that will Abend S0C3

when it tries to read a CSR record; if you have those

enabled and can generate the S0C3 Abend Dump, please send

it (tape or printed) to Merrill Consultants so we can

decode that RMF III record as well.

Change 15.383 MXG 15.09 only. Comma missing at end of INCODE= caused

TRNDRMFI error. This was caught in my QA test, but I failed to

Feb 17, 1998 copy the change from the QA library back into the master!
====Changes thru 15.382 were printed in MXG Newsletter THIRTY-THREE====
Change 15.382 NPM variables STARTIME, ENDTIME, and DURATM had missing

VMAC28 value in dataset NPMINRTM from SMF type 28 subtype '30'x.

Feb 16, 1998 Valid STARTIME and ENDTIME were INPUT in the STT section

processing, but the SAA section also INPUT STARTIME and

ENDTIME, and when there were no timestamps in the SAA

section, their missing value overrode the valid STT

values. To correct, the calculation of STARTIME and

ENDTIME in the AOFTYPE='SAA' section is performed by

IF STARTIME=. THEN DO; and IF ENDTIME=. THEN DO;

logic so that valid STARTIME/ENDTIME will be preserved.

Thanks to ???, Spar Handels AG, GERMANY.
Change 15.381 ASMTAPES Maintenance Level ML-16 now writes records for

ASMTAPES devices with four-digit UCBs. (We thought we had added

ZSMTAPES that support long ago, but testing uncovered our error.)

Feb 17, 1998 In addition, ML-16 corrects the "TEST" SYSPARM which did

not work as advertized. Note that "TEST" (which writes

to a flat file instead of SMF), does not support the WLM

(Workload Manager Fields) that ARE supported in non-TEST.

Thanks to Mark Schmitt, First Card, USA.

Change 15.380 Major enhancement to support for Boole & Babbage CMF

EXCMFVAL VSAM History File. MXG Previously decoded a dozen or

EXCMFVCD so subtypes, but this massive user contribution added

EXCMFVCF fourteen completely new datasets and their exits, and

EXCMFVCS creates variables in ten existing datasets that were only

EXCMFVDC a skeleton dataset before this change.

EXCMFVDV

EXCMFVGL The bulk of this change was written by

EXCMFVHD Pino Todesco, Boole and Babbage, Australia.

EXCMFVMP

EXCMFVMS This text will be updated with correct name, etc.

EXCMFVMV

EXCMFVRV As test data is received from interested sites for the

EXCMFVSO subtypes not yet decoded, MXG will be enhanced to also

EXCMFVST support those subtypes.

VMACCMFV

Feb 15, 1998 This is also called "Mainview" by Boole.

Thanks to Pino Todesco,Boole & Babbage, AUSTRALIA.

Change 15.379 Two new reports for "Batch Window" analysis, one based on

ANALBATW PDB.STEPS as input, and the second based on PDB.JOBS were

Feb 15, 1998 written by Grant and contributed by Ken.

Thanks to Grant Mackay, Standard Life, SCOTLAND

Thanks to Ken Williams, Southern Electric, UK.
Change 15.378 Some simple reports of Data Set activity based on the SMF

ANAL42 type 42 record's dataset TYPE42DS were contributed by

Feb 15, 1998 Ken from a previous lifetime.

Thanks to Ken Williams, Standard Life, SCOTLAND.

Change 15.377 Boole's CMF Cache Stats were split into two CMF datasets

VMACCMF CMF27CSC and CMF27C93, but there were no merge variables

Feb 15, 1998 with which to combine them, so the internal logic was

redesigned, so now the CMF27CSC dataset contains both

sets of variables. The CMF27C93 dataset is still built as

before (so your old reports won't fail), but you should

now use CMF27CSC for Cache reporting (and you can then

use your EXCMFC93 exit member to turn of creation of obs

in the now-archaic CMF27C93 dataset. Use CMF27CSC!

Thanks to Neil Ervin, Banc One Services Corporation, USA.

Change 15.376 Boole CMF subtype 16 record now creates two new datasets

EXCMF16M CMF16MAP (Virtual Storage Map) & CMF16LPA (LPA Modules)

EXDMF16L thanks to this significant user contribution.

VMACCMF

Feb 13, 1998

Thanks to Paul Hill, Midland Bank, UK.

Change 15.375 IMS Log records from Boole's IMF and from SAP and IBM's

TYPECIMS IMS log records can now be processed in one pass of the

TYPEIMFL IMS log. Member VMACIMS had internal variable MSGDRRN

VMACCIMS (a numeric variable) replaced by CMSGDRRN (a character)

VMACIMS to avoid conflict between CIMS and IMS member, and some

Feb 13, 1998 lines were relocated in both VMACCIMS and VMACIMS, and

macro _IMS in member VMACIMS now creates the variables in

the header that VMACCIMS expected, and new macro _CHAIN

is defined in member VMACCIMS, replacing the inline code

that handled Chained Transactions in member TYPECIMS, so

the end result is that new member TYPEIMFL contains:

%INCLUDE SOURCLIB(VMACCIMS,VMACIMS);

DATA _VARCIMS _VARIMS ;

_IMS ; _CDECIMS _CDEIMS ; _CHAIN ;

so TYPEIMFL creates all 25 datasets from IMF, IMS, and

SAP's IMS log records in one pass of the IMSLOG tape:

CIMSTRAN CIMSDBDS CIMSDB2 CIMSPROG

IMS01 IMS01M IMS03 IMS03P IMS07 IMS08

IMS31I IMS31O IMS35I IMS35M IMS35O IMS35P

IMS36 IMS36M IMS5901 IMS5903 IMS5936 IMS5937

IMS5938A

SAPIMSBA SAPIMSON SAPIMSSP

And if you don't want to create all 25 datasets, you can

use MXG's new &MACxxxx macro variable (Change 15.356) to

redefine the _Lxxxxx dataset macros to _NULL_ for the

unwanted data sets, and they won't be created:

%LET MACIMS=

MACRO _LIMS01 _NULL_ % MACRO _LIMS01M _NULL_ %

MACRO _LIMS03 _NULL_ % MACRO _LIMS03P _NULL_ %

MACRO _LIMS07 _NULL_ % MACRO _LIMS08 _NULL_ %

MACRO _LIMS31I _NULL_ % MACRO _LIMS31O _NULL_ %

MACRO _LIMS35I _NULL_ % MACRO _LIMS35M _NULL_ %

MACRO _LIMS35O _NULL_ % MACRO _LIMS35P _NULL_ %

MACRO _LIMS36 _NULL_ % MACRO _LIMS36M _NULL_ %

MACRO _LIMS591 _NULL_ % MACRO _LIMS593 _NULL_ %

MACRO _LIMS596 _NULL_ % MACRO _LIMS597 _NULL_ %

MACRO _LIMS598 _NULL_ %

;

%INCLUDE SOURCLIB(TYPEIMFL);

would only create the 4 IMF and the 3 SAP datasets.

Thanks to Reinhard Kelm, GRZ, GERMANY.

Thanks to Friedhelm Danne, GRZ, GERMANY.
Change 15.374 There was no KEEPs example that worked with ASUMUOW, but

IMACCICS comments implied there was. Now the three examples show

Feb 12, 1998 the KEEPs for CICSTRAN that are required for ASUMCICS,

ASUMCICX, and ASUMUOW. The actual KEEPs examples use

the DROP= statement in the _K macro to control keeping.

Thanks to Tom Weiland, Phoenix Home Life Mutual Insurance Co, USA.

Change 15.373 Support for OS/390 Version 2 Release 5 is contained in

none MXG 15.04 or later, as there were no changes to the SMF

Feb 12, 1998 records between OS/390 2.4 and OS/390 25., and MXG Change

15.270 added support for Version 2 Release 4 in 15.04.

Change 15.372 Support for Boole's IMF 3.2 (for IMS 6.1) INCOMPATIBLE

VMACCIMS because 400 bytes were inserted at the end of the fixed

Feb 12, 1998 portion of the record, causing DBD counts to be wrong

without this change. The 400 bytes added only four new

fields: GMT Adjustment, UTCA, Checkpoints/Syncpoints

CTCKPNTS, Maximum Locks Held MAXLOCKS and Total Locks

Held, TOTLOCKS.

Thanks to Reinhard Kelm, GRZ, GERMANY.

Thanks to Friedhelm Danne, GRZ, GERMANY.
Change 15.371 The new BLDNTPDB macro provides simple daily/weekly/

BLDNTPDB monthly dataset building and reporting from NTSMF data

Feb 12, 1998 records. This macro does everything for you, or you

can override its options. It even supports reruns!

Especially for new users who are not familiar with the

PDB architecture, this runs-only-on-ASCII-SAS example

should make life simple. More documentation on how to

use this facility will be forthcoming, and this is still

in development.

Thanks to Marti Henley, Matridigm, USA.

Thanks to Chuck Hopf, MBNA, USA.
Change 15.370 Concurrency analysis routine ANALCNCR has new MYTIME=

ANALCNCR argument for sites that want to specify a non-standard

Feb 12, 1998 time period for summarization. (Note: the MYTIME=

specification is not used to "sync" the intervals, and

SYNCINTV=YES is ignored if INTERVAL=MYTIME is used.

Thanks to Tom Parker, CSC FSG Banking Division, USA.

Change 15.369 Trending for the ASUMUOW using the PDB.ASUMUOW dataset

TRNDUOW daily as input, rather than building ASUMUOW weekly, is

Feb 12, 1998 recommended for sites that need to analyze CICS MRO with

or without DB2 by UOWID, i.e., by Unit of Work ID, due

to potential large volumes of data, CPU time, work space

and I/O operations if the trending were done from the

weekly PDB.

Thanks to Chuck Hopf, MBNA, USA.

Change 15.368 New argument DURATM=INTERVAL will cause variable DURATM

ASUMCICS to be created in VMXGSUM's output dataset, and DURATM

ASUMDB2A will contain the duration specified on INTERVAL (for the

ASUMDB2B "real" durations of QTRHOUR, HALFHOUR,HOUR, and DATE), to

ASUMHSM be consistent with MXG design intentions of having DURATM

ASUMJOBS in all interval datasets. The eight ASUMxxxx members

ASUMPRTR listed are the first to exploit this new DURATM=INTERVAL

ASUMTMNT argument for VMXGSUM. In addition, both VMXGSUM and

VMXGDUR VMXGDUR now contain new argument MXGWEEK= that allows

VMXGSUM you to change your "first day of the week" in Trending

Feb 12, 1998 datasets from MXG's default of MONday.

Thanks to Chris Weston, SAS ITSV Cary, USA.

Change 15.367 The EXCPNODD and EXCPTODD counts in step and job records

ANALMULT with MULTIDD='Y' (i.e., there were so many DD segments

Feb 12, 1998 that MVS created multiple type 30 records for a step)

are incorrect, because they are created for each step

record and are not summed across all of the multiple

records. Since these MULTIDD='Y' jobs are almost always

data center work (like SAR/RMDS/etc spoolers) the error

has not had any real effect, but this user has provided

a series of macros that will start with the raw SMF data

and correctly calculate EXCPNODD and EXCPTODD for these

MULTIDD='Y' tasks. Because it must sort lots of records,

and takes appreciable CPU and IO resources, I have no

plans to add the logic to BUILDPDB, but it is here if

you need it!

Thanks to Dr. Alexander Raeder, Karstadt, GERMANY.
Change 15.366 Member ANALUOW as announced in Change 15.205, but it did

ANALUOW not get copied into the MXG library until MXG 15.09, and

Feb 12, 1998 it was cleaned up and enhanced now as well.

Thanks to Chuck Hopf, MBNA, USA.

Change 15.365 CICS DFHSTUP "Shutdown" reports CICNQG, CICLGS, and

ANALCISH CICLGR have been added to member ANALCISH.

Feb 20, 1998

Thanks to Bruce Widlund, Merrill Consultants, USA.

Change 15.364 Support for StorageTek VSM SMF records adds 17 new data

EXSTCV09- sets for HSC subtypes 09 thru 26. The datasets are

EXSTCV25 named STCVSM09 thru STCVSM26 and they provide significant

VMACSTC measurements, including bytes read/written, duration of

Feb 12, 1998 mount, dismount, recall, migrate, etc., etc., etc!

Feb 20, 1998 The code has been syntax checked and subtypes 13, 14, 15

and 21 records have been processed.
Change 15.363 IDMS Journal Format Performance Records were apparently

TYPEIDMJ changed in IDMS Version 12, and the old Journal support

VMACIDMJ failed. This revision heuristically skips over the non

Feb 11, 1998 performance segments and has been tested with real data.

Unfortunately, the quick solution was to create VMACIDMJ

that knows about the idiosyncrasies of the Journal data,

but this means dual maintenance in VMACIDMS/VMACIDMJ with

each new version of IDMS!

Thanks to Alfred Mueller, ABN-AMRO Services Co., USA.
Change 15.362 NPM type 28 subtype 82 caused INPUT STATEMENT EXCEEDED

VMAC28 because 36 bytes were input when there were only 24. The

Feb 10, 1998 statement ELSE IF LENAOF GE 172 AND NPMSUBTYP NE 83X

THEN INPUT +36 @;

must be changed to ELSE .... THEN INPUT +24 @;

Thanks to Mel Hitchings, British Telecom, WALES.

Change 15.361 Creation of the WKnnPCPU, WKnnPPRV, and WKnnPUSR percent

NTINTRV values in the NTINTRV dataset was revised to convert the

Feb 10, 1998 percentages to durations, sum the durations, and recalc

the percentage values. If the DURATM of all PROCESS

intervals is the same, the new and old calculations are

identical, but the old calculations were wrong if there

were dissimilar values of DURATM in the PROCESS dataset.

Thanks to Wayne Holzback, Reynolds Metal Company, USA.

Change 15.360 Variable OTHRPUSR was not in the RETAIN list in member

NTINTRV NTINTRV, causing that variable to be missing from the

Feb 10, 1998 NTINTRV dataset.

Thanks to Marti Henley, Matridigm Corporation, USA.

Change 15.359 DCOLLECT variables DDCXSYS and DDCXREG were reversed in

VMACDCOL the input statement; DDCXSYS should have been first and

Feb 10, 1998 then DDCXREG second, as it was in DCOLLECT 1.2. In 1.3

the documentation reversed the order, but tests with

data confirm the real order is XSYS and then XREG.

Thanks to Mike Blocker, Fidelity Investments, USA.

Change 15.358 The TYPE42AU (Audit ACTIVATE, ENF, or VARY SMS events)

VMAC42 was not input correctly, but the record is also wrong!

Feb 10, 1998 MXG errors: insert +16 between INPUT and SMF42ETY.

change +72 to +88 before SMF42ESD.

IBM error: there are only 31 total bytes between SMF42ETY

and SMF42EVL, instead of the 32 documented.

I changed the +22 after SMF42ENM to +21 while

we sort out IBM's error.

Thanks to Michael E. Friske, Fidelity Investments, USA.
Change 15.357 Support for APAR OW29790 adds variables PGOALPCT and

VMAC99 SERVER01-SERVER05 and SERVPN01-SERVPN05 to the TYPE99_6

Feb 10, 1998 dataset to track service classes served.
Change 15.356 This enhancement adds a &MACxxxx macro variable in each

DOC VMACxxxx member so that you can dynamically override the

VMXGINIT IMACxxxx member without having to EDIT and save an IMAC!

Feb 8, 1998 In each VMACxxxx member, a new line with "&MACxxxx;" was

added immediately after the %INCLUDE SOURCLIB(IMACxxxx).

For example, in member VMACHSM, the statements are:

%INCLUDE SOURCLIB(IMACHSM);

&MACHSM;

Each of the &MACxxxx macro variables are initialized to

a null string, and GLOBALed, in member VMXGINIT.

So in your program, you can set the value of the &MACxxxx

macro variable equal to the old-style macro definition

syntax ("MACRO macro-name macro-text %" to thus change

the definition of the _L,_K dataset tailoring macros:

MACRO _Lxxxxxx - tailor Libref and Dataset Name

MACRO _Kxxxxxx - tailor KEEP/DROP list of vars

or to change the new _S PROC SORT macro added in 15.328:

MACRO _Sxxxxxx - PROC SORT and BY statements

or in a future version, change the planned _E Exit macro:

MACRO _Exxxxxx - will allow you to override the

%INC SOURCLIB(EXxxxxxx) exit.

and thus have complete dynamically tailoring for each

individual dataset. For example, for SMF 30, you could:
%LET MAC30=

MACRO _LTY30U1 _NULL_ %

MACRO _LTY30U4 MYDD.TYPE30_4 %

MACRO _LTY30U5 _NULL_ %

MACRO _LTY30UV _NULL_ %

MACRO _LTY30UD _NULL_ %

MACRO _LTY30U6 _NULL_ %

MACRO _LTY30MU _NULL_ %

MACRO _LTY30OM _NULL_ %

;

%INCLUDE SOURCLIB(TYPE30);

to create only the dataset TYPE30_4 from the subtype

4 record, and it would be written to the //MYDD library.

For a User-SMF record, where you must set the SMF record

ID in the MACRO _IDxxxx in the IMACxxxx for the product,

you can now redefine the _ID macro with the &MACxxxx.

For example, to process the HSM User SMF record and put

all HSM datasets in the //WORK library, you could use:
%LET MACHSM=

MACRO _IDHSMDS 240 %

;

%INCLUDE SOURCLIB(TYPEHSM);

For those not familiar with the syntax of the %LET, you

use %LET MACxxx= before the old-style macro statements

and then you end the %LET statement with a semi-colon.

If the text contains semi-colons, you can instead use

%LET MACxxx= %QUOTE( whatever-with-semicolons) ;
For those not familiar with the syntax of the old-style

MACRO statement, it is simply

MACRO macro-name macro-text

and then a single percent-sign ends the definition.

If the macro text has a percent-sign, then double it:

MACRO _EXXXXXX %%INCLUDE SOURCLIB(EXYYYY); %

This new &MACxxxx architecture is a significant piece of

the complete redesign to allow external control of all

MXG datasets. Because the &MACxxxx enhancement is safe

and can be used to override existing IMAC definitions, it

is added in MXG 15.09. The remaining pieces of this big

change will be implemented after MXG 15.15 has shipped!

Change 15.355 Variables ASICSAA, ASISQAA, ASIECSAA and ASIESQAA were

VMACRMFV horifically large, because they should have been read in

Feb 5, 1998 with &RB.4. informat instead of &PIB.4.

Thanks to Peggy Dunne, Depository Trust Company, USA.

Change 15.354 All VMACs that process SMF records were made consistent,

VMACSMF by locating the "IF ID = ..." statement immediately after

BUILD606 the MACRO _CDExxxx statement. All new VMACs have been

BUIL3606 written that way, but some old VMACs had LABEL statements

Feb 5, 1998 between the macro definition and the IF test. Now that

the _CDExxxx macros for SMF records all start with the IF

test for record IDs, the efficiency improvements of MXG

Change 15.329 are revised again. Instead of the logic

ELSE IF ID=xx THEN DO;

_CDExxxx

END;

now, the logic

ELSE _CDExxxx

can be used, eliminating the extra test imposed by the

external IF statement.

The VMACs whose IF statement was relocated are:

VMAC25 VMAC7072 VMAC71 VMAC73 VMAC74 VMAC75 VMAC76

VMAC77 VMAC78 VMACCTCP VMACEDGS VMACF127 VMACFTEK

VMACHSM VMACODS VMACSYNC VMACTSOM

Thanks to Chuck Hopf, MBNA, USA.

Change 15.353 Support for Landmark's TMON for CICS Version 2 (MXG1506)

VMACTMO2 conflicted with ITSV because some variables were changed

Feb 5, 1998 incorrectly by MXG, and some variables were not created.

Variable MONIMANT is now created, variable TATASKNN is

re-spelled TATASKNR, variable CICSLV is renamed to be

CICSLEV (Landmark changed it from PIB1 to $CHAR4 so I

had no choice), and variable FACTYP is changed back to

PIB1 and formatted as it was in prior versions.

Thanks to Rod Gaas, Eagle Star, UK.
Change 15.352 MXG 15.08. The label for variable S1ELEMLO was typed as

VMAC110 S1ELEMLO='LOWEST*FREE*ELEMENTS*='/ but SAS did not die

Feb 4, 1998 or produce an error message! The extraneous characters

are now removed. Also, macro _CICSTCMN was added to the

KEEP= list for dataset CICSSMED so the common variables

(in particular, SYSTEM and SMFTIME) will now be kept.

Thanks to Chris Weston, SAS ITSV, USA
Change 15.351 Cosmetic. Variable LCSLOV1 label should have been

VMAC28 'OVERFLOW*FLAG*BYTE'.

Feb 4, 1998

Thanks to Ley Teng, QANTAS, AUSTRALIA.

Change 15.350 The explicit dataset names TYPE72 and TYPE6 were replaced

ASUMPRTR with _LTY72 and _LTY6 macro names instead, so that if you

Jan 31, 1998 re-direct either dataset from the default of "WORK", MXG

will find it where you put it.

Thanks to Chris Weston, SAS ITSV, USA.
Change 15.349 The example "reptprod" file to control fields extracted

ADOCMWAI from HP's MeasureWare for AIX is now provided in the

Jan 28, 1998 ADOCMWAI member.

Thanks to Lorenzo Wright, NCCI, USA.

Thanks to Thierry Van Moer, Comparex, BELGIUM.
Change 15.348 Member VMACTRNS was a copy of VMACTRSN that had been

VMACTRSN saved under the incorrect VMACTRNS name, so the member

Jan 27, 1998 VMACTRNS has been deleted.

Thanks to Dr. Alexander Raeder, Karstadt AG, GERMANY.

Change 15.347 The VMXGFOR references in the commented out example that

RMFINTRV can be used to read SMF directly must all be %VMXGFOR.

Jan 27, 1998 Only if you remove the comment block to read SMF direct

would you see this error.

Thanks to Jim Chappell, Pacific Power and Light, USA.
Change 15.346 Support for Landmark for MVS Version 2 segments that were

VMACTMV2 not available for testing in MXG 15.08. New segments

Jan 26, 1998 that have now been data-validated are: JD, LM, and TR.

These record subtypes caused INPUT STATEMENT EXCEEDED.

Thanks to Jonathan McVey, Lowe's Company, USA.

Thanks to Mark Derreks, Hershey's, USA.

Change 15.345 MXG 15.08 only. The new MACRO _SCICEXC should have had

IMACCICS &PDB..CICSEXCE instead of &PDB.CICSEXEC. Fortunately,

Jan 20, 1998 the new macro is not (yet!) used in MXG code, so there

was no execution error.

Thanks to Freddie Arie, Lone Star Gas, USA.
Change 15.344 Variable NTVERSP was truncated to 12 positions, so it

VMACNTSM had only "Service Pack" instead of "Service Pack 3".

Jan 20, 1998 Move the NTVERSP in the line with $12 to the next line,

with $32, so it will be 32-bytes long.

Thanks to Marty Henley, Matridigm, USA.
Change 15.343 MXG 15.08 only, JES2 only. Change 15.329 reordered the