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

Thanks to Laurie Bigelow, UCCEL, USA.

Change 05.114 Variable SD2DEVNR was mis-spelled in the KEEP list for

Sep 3, 1987 CACHETY data set, and was thus not kept. Correct the

VMACACHE obvious mis-spelling.

Thanks to Ray Dickensheets, Yellow Freight System, USA.

Change 05.113 PGBLPGS, STGUTIL, and DRUMUTIL variables in VMONPERF

Aug 31, 1987 were not calculated in non-HPO data. COPY lines

VMACVMON 1545-1546 to become 1113.1 and 1113.2. MOVE line 1547

to 1113.3. Change 1113.1 to be: PGBLPGS=MAXSIZEP;

Thanks to ???, ???.
Change 05.112 This simple example of reading SYSLOG to track IEF244I

Aug 21, 1987 (job delayed due to tape allocation) messages from the

XSYSLOG JES2 SYSLOG file to create MXG dataset IEF244I will be

expanded in future versions of MXG to provide extended

analysis of important SYSLOG messages. Suggestions for

other events which are only available from SYSLOG are

solicited. This will not be expanded until Version 6.

Thanks to John Maher, American Honda Motor Company, USA.

Change 05.111 The bit testing to create TYPE25 (JES3 only) variables

Aug 21, 1987 ALOCBYDD and ALOCAUTO were reversed. Reverse the 0 and

VMAC25 1 values in lines 57-58 and lines 61-62.

Thanks to Miss C. Keelan, Commonwealth Bank of Australia, AUSTRALIA.

Thanks to Stephen Geard, Commonwealth Bank of Australia, AUSTRALIA.
Change 05.110 The decode of RECFM in both VTOC and TYPE1415 logic is

Aug 19, 1987 moved to new member VMACRCFM, and additional record

VMAC1415 formats (formerly OTHR, like FBA, FBS, etc.) have been

VMACRCFM added to (hopefully) completely decode RECFM and to

VMXGVTOC eliminate blank or OTHR values.

Thanks to John Potter, Mitsubishi Electronics, USA.

Thanks to Chuck Hopf, Dean Witter Reynolds, USA.
Change 05.109 This change fixes an error which only existed in the

Aug 19, 1987 pre-release 5.1. For jobs which went through the SPIN

BUILDPDB data set, the variable SYSTEM was usually blank. The

BUILDPD3 change affected the five data sets built at line 284

(JES2, line 294 JES3). IN= logic was added to TYPE....

in the SET statement and the storing of SYSTEM into

the five SYSxxx variables was made conditional.

Thanks to Chuck Hopf, Dean Witter Reynolds, USA.

Change 05.108 Some MXG modules now use the "new macro, or %macro"

Sep 3, 1987 facility of the SAS System. (MXG has always used many

JCLUXREF "substitution" or old-style SAS macros). The use of

VMXGVTOC the "new macros" requires the SAS system option MACRO

to be enabled (SAS 5.16 defaults to NOMACRO). This is

done either by adding MACRO to the OPTION list on the

EXEC SAS statement for a step, or by changing the SAS

system default (PROC SETINIT by the SAS installer.)

The listed members used the "new macro" facility and

their JCL example now specifies MACRO. Hereafter, all

MXG members that define "new macros" will be so noted.

Thanks to Ray Dickensheets, Yellow Freight System, USA.

Thanks to John Potter, Mitsubishi Electronics, USA.

Thanks to Rob Ray, EUA Service Corporation, USA.

Change 05.107 The variables TAPE3420 and TAPE3480 are now included

Aug 14, 1987 in the PDB.STEPS and PDB.JOBS data sets. Both members

BUILDPDB were re-numbered

BUILDPD3

Thanks to Dennis Dwyer, CITICORP, USA.
Change 05.106 The type 128 record which can be created by NPM is

Aug 14, 1987 actually supplied by IBM as source code for the exit.

IMACNPM Since the record id is not fixed at 128, this change

VMAC128 adds member IMACNPM in which you tell MXG the actual

record ID of the NPM VTAM Session records.

Thanks to Joseph J. Faska, Chemical Bank, USA.

Change 05.105 Support for SYNCSORT user SMF record was added. The

Aug 14, 1987 SYNCSORT data set is created. These members replace

EXSYNSOR XMACSYNC (Change 5.39).

IMACSYNC

TYPESYNC

VMACSYNC
Change 05.104 Support for CINCOM SUPRA 1.1 System log file produces

Aug 14, 1987 three data sets, SUPRDBMX, SUPRFILE, and SUPRINIT.

ANALSUPR Sample reports are in member ANALSUPR.

EXSUPDBM

EXSUPFIL

EXSUPINT

TYPESUPR

Thanks to Mark Degner, Navy Finance Center, USA.
Change 05.103 VM/Monitor Seek Analysis had one too few percent signs

Aug 12, 1987 which caused syntax errors. In line 459, change the

ANALVMOS '%' to '%%'.

Thanks to Steve Smith, Hammermill Paper, USA.

Change 05.102 Default SMF record id numbers for ACF2, MODEL204,

Aug 10, 1987 ROSCOE, and TSO/MON were changed to the inpossible

IMACACF2 values of 512 or higher, to prevent STOPOVER ABEND in

IMACM204 new sites testing JCLTEST.

IMACROSC

IMACTSOM

Thanks to Joseph J. Faska, Chemical Bank, USA.
Change 05.101 Format for CPUWAIT4-8 and PCTCPBY4-8 was not provided.

Aug 10, 1987 This only affected the 5.1 tapes. Change the CPUWAIT3

VMAC7072 in line 298 to CPUWAIT8. Change the PCTCPBY3 in line

316 to PCTCPBU8. Change CPUWAIT3 in line 479 to read

CPUWAIT8, only for looks since 370 can't run on 3090.

Thanks to Ruth Heidel, UCCEL, USA.

Change 05.100 CICS 1.6.1 FCGETCN field is increased from two to four

Jul 31, 1987 bytes by PTF UL10276.

IMACPTF IMACPTF - Replicate the logic for AP40463, and change

UTILCICS AP40463 to UL10276.

VMAC110 UTILCICS- Replicate lines 354-357 (for AP40463) and

change the test to 'FC-GET-COUNT' and set

PTF='UL10276';

VMAC110 - Replace lines 546-547:

FCGETCN PIB2.

FCPUTCN PIB2.

With these four lines:

@;

IF _UL10276 THEN INPUT FCGETCN PIB4. @;

ELSE INPUT FCGETCN PIB2. @;

INPUT FCPUTCN PIB2.

Thanks to Malcolm Morgan, Wachovia Bank, USA.

Thanks to Dave Feigenbaum, UPS, USA.

******Changes through 5.99 were on the first pre-release of MXG 5.1 ****
Change 05.99 The two new SMF records created with MVS 2.2 are now

Jul 31, 1987 available. A type 36 is written for each import/export

TYPE36 of the ICF catalog. The type 41 is written for each

VMAC36 access to a Data-In-Virtual object. The type 41 is

EXTY36 currently incomplete; the job name, readtime, etc of

TYPE41 the task using the DIV object was inadvertently left

VMAC41 out of the SMF record in ESP. A PTF is expected that

EXTY41 will add the data and probably break the VMAC31 code,

but I don't have the PTF number or change info yet.

With this change, MXG Version 5.1 Supports MVS 2.2.

Change 05.98 Minor, cosmetic change in creating SMF71IS1 and IS2

Jul 31, 1987 field names from the SMF71AVA field, none of which are

VMAC71 kept.
Change 05.97 This standalone JCL and program will decode the SAR

Jul 26, 1987 account records from its non-SMF journal file.

XMACSAR

Thanks to Chuck Hopf, Dean Witter Reynolds, USA.

Change 05.96 MXG DASD Storage Measurement is now supported by these

Jul 26, 1987 new members. The earlier XMAPVTOC program has been up

FMXGUCBL dated and is now an executable (new style) SAS macro

VMXGVTOC %VMXGVTOC, which produces the VTOC and FREE data sets

TYPEVTOC from the DDNAME of DISK. These two data sets describe

the contents of that volume for billing DASD storage

and/or DASD capacity measurement. To measure all DASD

volumes in one execution, and to avoid JCL errors due

to offline volumes, the MXG-provided SAS function

FMXGUCBL dynamically allocates all online DASD volumes

to DDNAMES of DDnnnnnnn, and returns the number of

volumes found online. The example program in VMXGVTOC

can be executed to create the MXGVTOC and MXGFREE data

sets from all online DASD volumes.

Change 05.95 Support for IDMS 10.1 (pre-release) completed. The new

Jul 26, 1987 new "Application Monitor" record subtypes create ten

VMACRTE new IDMS.... data sets. The original TYPERTE data set

EXRTEARA is now created from the subtype 1 record. See comments

EXRTEBUF in VMACRTE for documentation and comments about data.

EXRTECDM MXG variable names are the same as the Culprit names

EXRTEDBK used in Cullinet's PMIRPT00 example report program for

EXRTEERU this IDMS Performance Monitor (formerly the Run Time

EXRTEINT Evaluator product, hence the RTE acronym). See Change

EXRTEJRN 5.120.

EXRTEPGM

EXRTESTG

EXRTETPI

Thanks to Terry Layman, Cullinet, USA.

Change 05.94 Validation of ACF2 record support has been completed.

Jul 22, 1987 See updated description of change 5.68.

Change 05.93 PDB.JOBS variables RACFGRUP RACFTERM and RACFUSER were

Jul 22, 1987 kept in _PDB30_4 instead of _PDB30_5. Thus these three

BUILDPDB variables in PDB.JOBS came only from the initiate 30_1

BUILDPD3 data set. A value placed in RACFTERM by an SMF exit at

terminate was not picked up by MXG. With this change,

the three RACF variables in PDB.JOBS will have their

value from the 30_5 if it exists, or from the 30_1 if

no job termination record was found. It was supposed

to have been this way all along!

Change: blank the three variables in line 84 and add

the three variables in lines 102-103.

Thanks to Kenneth D. Jones, Maritime Telegraph and Telephone, CANADA.

Change 05.92 RMFINTRV standalone execution had an extra semi-colon

Jul 21, 1987 in an include in the commented out code. Fixed.

RMFINTRV

Thanks to Steve Glick, Southern Methodist University, USA.

Change 05.91 If SRBCOEFF is zero, a division exception occurred in

Jul 21, 1987 MVS/XA. The division is now protected.

VMAC7072

Thanks to John Mueller, ARMCO Steel, USA.

Change 05.90 The CICS 1.6 variables ICV and ICVTSD were incorrectly

Jul 21, 1987 calculated. ICV=ICV/76.8 and ICVTSD=10*ICVTSD/3 in the

VMAC110 1.6 code only.

Thanks to Chuck Hopf, Dean Witter Reynolds, USA.

Change 05.89 The variables QISEDBD and QISEPAGE in DB2STAT1 dataset

Jul 14, 1987 were wrong. Delete the two lines in ANALDB2:

ANALDB2 QISEDBD =DIF(QISEDBD);

QISEPAGE=DIF(QISEPAGE);

because these two variables are not accumulated.

Several other discrepancies between MXG and DB2PM are:

a. QISEFREE "FREE PG IN FREE CHAIN" reported in DB2PM

is (incorrectly) the value from the previous record.

b. DB2PM "Record Time" is QWHSSTCK-DURATM, the start

time of the interval, not the SMF record timestamp.

c. The DB2PM calculations for EDM Pool percentages are

frequently total several percent more than 100. This

is because QISEFREE is from the previous interval, as

noted above.

d. DB2PM reported "datasets open - current" is always

greater than the "maximum" value reported, and the

"current" number equals both QTDSOPN and QDDSMAX in

the record (which seems in error itself). The value

under "maximum" seems close to current; in most of

intervals it is equal to QB1TDSO (and there were no

higher buffer pools used). In other intervals the

printed "current" value is 2, QB1TDSO is 1 and no

other buffer pools were used.

e. DB2PM is flat wrong on the report page which has

both "AGENTS SERVICES - EXECUTION UNIT SWITCH" (for

QVASX... fields) and "STORAGE MANAGER (QSST.. fields).

The first six lines of the report are repeated in

lines 7-12 (which should be the first six S.M. lines).

Lines 13-19 contain what should be in lines 7-13.

What should be in lines 14-19 does not exist in DB2PM!

f. Other than that, DB2PM agrees with MXG.

Thanks to Martha Hall, Metropolitan Life, USA.

Change 05.88 The new IMS log processing logic (MXG Change 5.66) was

Jul 14, 1987 validated in detail by three sequential transactions

TYPEIMS under IMS 2.1. The new logic stood up to the test. The

only data we seem to loose now is the data from the

IMS type 36X log record, DEQTIME. The MSGDRRN value of

the 36 record is different than the MSGDRRN in the

0103 record; this prevents the 36 from matching. This

matter is under discussion with IBM. For now, IMS 2.1

is supported by MXG, but the three DEQ.... variables

are missing in the IMSTRAN data set from IMS log data.

Thanks to John D. Pike, American Airlines, USA.
Change 05.87 Support for the SMF record written by the STOPX37

Jul 12, 1987 product.

EXTYX37

IMACX37

TYPEX37

VMACX37

Thanks to Chuck Hopf, Dean Witter Reynolds, USA.
Change 05.86 The labels for variables READY11 thru READY15 were all

Jul 11, 1987 for READY11 in the TYPE70 data set.

VMAC7072

Thanks to Ruth Heidel, UCCEL, USA.

Change 05.85 The RMF CACHE Report SMF support added by Change 5.31

Jul 07, 1987 provides correct device segment data, but only the 1st

VMACACHE control unit segment was read. Major logic changes are

in the logic to associate the control unit segments

properly with their device segments. MXG has now been

tested with REPORLVL data values of 17 and 18.

Thanks to Jon Sahl, New York Life, USA.
Change 05.84 For CICS 1.7, the CICSYSTM (global performance data)

Jul 01, 1987 variables CPUTCBTM and OWSTCPTM were wrong. This had

VMAC110 no effect on the CICSTRAN data set variables. A minor

error in calculation of TASKCPTM in CICSYSTM was also

corrected by this change. The figure on page 66 of

the MXG Supplement is accurate for CICS 1.6; the below

figure is correct for CICS 1.7 and uses the CICSPARS

descriptions of maintask, subtask, etc., values:

CPUTCBTM SRB

|-----------------------------------------------| |--|

MAINCPTM SUBTCPTM

|--------------------------------------|--------|

KCCPUTM USRCPUTM OSWTCPTM

|-------|------------------------------|--------|

| JCCPUTM TCCPUTM TASKCPTM |

|---------|---------|----------|

"CICS" "APPL"

|---------------------------|----------|

1. Change both occurrances (lines 821 and 884),

TASKCPTM=USRCPUTM-TCCPUTM;

to read

TASKCPTM=USRCPUTM-TCCPUTM-JCCPUTM;

2. Change line 883, which now reads

CPUTCBTM=SUM(KCCPUTM,OSWTCPTM,USRCPUTM);

to read:

CPUTCBTM=OSWTCPTM;

3. Insert after line 884 these five lines:

CICSCPTM=KCCPUTM+TCCPUTM+JCCPUTM;

MAINCPTM=TASKCPTM+CICSCPTM;

SUBTCPTM=OSWTCPTM-MAINCPTM;

OSWTCPTM=SUBTCPTM;

TOTLCPTM=MAINCPTM+SUBTCPTM+CPUSRBTM;

4. Add these five new variable names to the KEEP=

list (lines 66-81) for _CICOTHR.CICSYSTM.

Thanks to Joseph J. Faska, Chemical Bank, USA.
Change 05.83 CICS 1.7 sites with a large number of user clocks or

Jun 29, 1987 counters added to their type 110 SMF record received

VMAC110 the "Invalid task number" MXG error message because of

UTILCICS an MXG error. The string of connectors can be over 200

bytes, but MXG had only a 200-byte character variable.

The confirming symptom of this error is MCTSSDCN, the

number of "connectors", will have a value over 100 in

the MXG error message. Additional lines were changed

UTILCICS and VMAC110; only the execution-critical code

changes are printed here:

After line 493 (END;) and before line 494 (ELSE INPUT)

insert the following five lines:

ELSE IF CONNBYTE GT 200 THEN DO; 493.1

INPUT CICTRANV $CHAR200. @; 493.2

SKIP=CONNBYTE-200; 493.3

INPUT +SKIP @; 493.4

END; 493.5

Thanks to Shawn Weikel, CITIBANK, USA.

Change 05.82 Support for the modified type 6 SMF record written by

Jun 29, 1987 the SAR (SYSOUT Archival and Retreival) product from

VMAC6 Central Software and the VPS (Virtual Printer Support)

product from 1:

New line 111.1:

ELSE IF SUBS=50658 THEN SUBSYS='SAR';

Change line 137:

ELSE IF SUBSYS='JES3' THEN DO;

To read:

ELSE IF SUBSYS='JES3' OR SUBSYS='SAR' THEN

Thanks to Frank Bolan, Bergan Brunswig, USA.
Change 05.81 The CICS 1.7 variable UOWID (which ties a transaction

Jun 24, 1987 in an MRO Terminal Owning Region to the same logical

VMAC110 transaction in other Application Owning Regions) is

now formatted as a $HEX16 so it prints "better". The

first six bytes are the timestamp of attach and they

are now decoded into new variable UOWTIME. (For DL/I

batch jobs, only the time, HHMMSS, is actually in the

data record. The SMF record date is used for DL/I.)

Change 05.80 The interval of aggregation of the RMFINTRV data set

Jun 16, 1987 (set by macro _DURSET) is now defined in IMACRMFI so

IMACRMFI that inline changes do not have to be made. This is

RMFINTRV consistent with the description in the MXG Supplement.

Change 05.79 XMACVTOC now supports 3380 devices. This "extra" MXG

Jun 14, 1987 program reads the VTOC of a device pointed to by the

XMACVTOC DISK ddname, and provides the mapping of the disk

size and physical location of all data sets and free

spaces on a volume. You should compare this program

with MAPDISK on the SAS Sample library. This member

was replaced July 31, 1987 by Change 5.96.
Change 05.78 Support for DFSORT Release 9 which added new variables

Jun 14, 1987 to the type 16 SMF record. Access methods used, start

VMAC16 and end of sort timestamps, SRB CPU time, work data

set tracks and extents, I/O counts (SORTIN, SORTOUT

and total work), and sort return/reason codes are now

available for each sort execution. This member was

renumbered.
Change 05.77 1.The SYSTEM variable in PDB.PRINT will frequently be

Jun 10, 1987 wrong; usually the value of SYSTEM in PDB.PRINT was

BUILDPDB from the Step (30_4) record. This change corrects this

BUILDPD3 error and creates variables SYSINT, SYSSTP, SYSJOB,

SYSPUR and SYSPRN in PDB.JOBS to identify the system

from which each of the five records was found. As a

result of this redesign, SYSTEM in PDB.JOBS will come

from the first non-blank value in this ordered list:

SYSSTP, SYSJOB, SYSINT, SYSPUR, or SYSPRN to ensure

SYSTEM is non-blank and contains execution system if

execution records were found, otherwise it contains

the purge or print system id.

2.JOBCLASS was added to the PDB.PRINT and PDB.STEPS

data sets (from the PDB.JOBS) at the request of many.

3.PROTECT=ZZZZZ was added to the three CICS data sets

built by BUILDPD3 so that it would be consistent with

the JES2 version BUILDPDB, which has always used the

PROTECT= keyword.

Thanks to Georg Simon, Prudential, USA.
Change 05.76 The value of IRESPTM in CICSTRAN was set to missing if

Jun 10, 1987 any bit was on in TASERRFG, as these bits indicated an

VMAC110 error in timings. The bit meanings have been changed

FORMATS in CICS 1.6.1 and CICS 1.7 and CMF no longer corrupts

its timings. Therefore, IRESPTM is no longer dependent

on TASERRFG, and will always be calculated as ELAPSTM

minus WTTCIOTM. Format MG110ER has been changed to now

properly decode the new bit meanings of TASERRFG.

Thanks to Jo Engles, Virgina Light and Power, USA.
Change 05.75 Labels in VMAC110 were alphabetized. JCLUXREF needed a

Jun 10, 1987 DCLOG DD, and /* in JCL was changed to //* comments.

VMAC110 Option ERRORABEND was added to JCLPDB. All MXG

JCLUXREF production jobs must specify OPTION ERRORABEND to

JCLPDB protect you from yourself. ERRORABEND causes any SAS

error condition (but not warnings) to immediately

terminate the execution with a USER 999 abend. The

actual error condition will be printed on the log.

Thanks to Chuck Hopf, Dean Witter Reynolds, USA.

for alpha testing.

==========Changes thru 5.74 were published in Newsletter TEN===========
Change 05.74 Summarization and accumulation of VM/Monitor data a la

Jun 9, 1987 RMFINTRV into hourly (default) performance data with a

ANALVMOS small number of critical variables kept allows a years

DOCVMOS data in a small file for trending. Additionally, seek

EXECDALY histograms for VM disks is provided.

Member names have changed since Newsletter TEN.

Thanks to Steve Glick, Southern Methodist University, USA.
Change 05.73 Several minor cleanups detected in alpha testing of a

Jun 9, 1987 pre-release of Version 5:

VMAC128 a. VMAC128, two DOs removed and an END; removed.

TESTUSER b. TESTUSER, changed RMDSDATA to TYPERMDS (all).

TYPEDISO c. TYPEDISO needed quotes around SPLIT='*'.

Thanks to Chuck Hopf, Dean Witter Reynolds, USA.

Change 05.72 The VM/Monitor TOD time stamp is written in GMT time,

Jun 9, 1987 but the Start Collection time stamp is in local time.

VMACVMON Local time is offset from GMT by the SYSTIME parameter

ZONE in DMKSYS. Some sites (my test site included) set

ZONE=0 and the operators enter local time. This causes

the TOD time stamp to be local time. However, if your

site uses a nonzero value for ZONE, the STARTIME value

produced by MXG for each record was in GMT zone. This

change corrects this MXG oversite by using the Start

Collection (Class 0 Code 97) record to calculate the

time zone difference and set MXG times to local time.

The following lines inserted/changed will correct the

MXG time stamps. (SAS 5.16 under CMS will not accept

nulls in the DHMS function, while MVS SAS will. Line

2025 replaces the nulls with zeros for this reason.)

TIMEZONE=0; 276.1

STARTIME=16*MNHTOD+BASETIME-TIMEZONE; 279

RETAIN BASETIME INTERVAL STARTIME TIMEZONE ZDATE; 284

COLLTIME=DHMS(DATTODCL,0,0,TIMTODCL); 2025

DELTATOD=(16*MNHTOD+BASETIME)-COLLTIME; 2025.1

TIMEZONE=3600*FLOOR((ABS(DELTATOD)+99)/3600); 2025.2

IF TIMEZONE GT 0 AND DELTATOD LT 0 THEN TIMEZONE=-TIMEZONE; 2025.3

STARTIME=16*MNHTOD+BASETIME-TIMEZONE; 2025.4

The actual change in Version 5 is more extensive, and

warns you if a 0/97 record was not found. It also will

process VM/Monitor data from any time period, as long

as the monitor data includes a 0/97 (formerly, only

data within the last 203 days, the wrap time of the 5

byte TOD field, could be accurately processed). If the

first record is not a 0/97, MXG defaults the time zone

to zero and what you get could be either local or GMT.

Thanks to Terry Magill, NAS, USA.

Thanks to Steve Glick, Southern Methodist University, USA.
Change 05.71 Variables VECTUTM, VECTU0TM and VECTU1TM, VM/Monitor

Jun 9, 1987 reported Vector Facility Usage time (total, CPU, APU)

VMACVMON were added to VMONPERF data set. Note that only the

usage time is reported, and only globally; there is

no vector usage time reported by user in VM. The MXG

supplement discusses the vector Facility measurements.

Change 05.70 Variable INTERVAL was added to all of the VM/Monitor

Jun 9, 1987 data sets.

VMACVMON

Thanks to Steve Glick, Southern Methodist University, USA.