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

Change 21.100 CICSTRAN variable RTYPE has two new values, 'F' and 'S',

FORMATS that are now decoded by the $MG110RT format:

Jun 9, 2003 'C'='C:TERMINAL CONVERSE'

'D'='D:USER EMP DELIVER REQUEST'

'F'='F:FREQUENCY REQUEST'

'M'='M:SEMI-PERMANENT MIRROR SUSPEND'

'S'='S:SYNCPOINT'

'T'='T:TASK TERMINATION'

Thanks to Diane Eppestine, SBC, USA.
Change 21.099 Support for ten new NDM subtypes; this is incomplete;

VMACNDM only the skeleton set of variables is created for each

Jun 8, 2003 subtype, but all of the new members and VMXGINIT macros

exist, so only VMXGNDM will likely need to be updated;

just got DSECTs and no time to write the code today.

Thanks to Peter Lines, Royal Bank of Scotland, SCOTLAND.

Change 21.098 These TYPE7204 variables that contained 'SUM OF ALL'

VMAC7072 R724ACTF /*SUM OF ALL ACTIVE FRAMES*/

Jun 8, 2003 R724IDLE /*SUM OF ALL IDLE FRAMES*/

R724SLOT /*SUM OF ALL SLOTS USED*/

R724DIV /*SUM OF ALL DIV FRAMES*/

R724FIX /*SUM OF ALL FIXED FRAMES*/

in their label are now divided by NRSAMPLE to change

their value to the more useful and expected average

value for the interval, and 'SUM OF ALL' was removed from

their labels.

These variables had correct values because MXG did divide

by NRSAMPLES, but still had 'SUM OF ALL' in their labels:

R724TSV /*SHARED PAGE VIEWS*/

R724VIN /*VALID SHARED PAGES IN CSTOR*/

R724VLC /*SHARED PAGE VALIDATIONS*/

R724GPI /*SHARED PAGEINS FROM AUX*/

which is now removed.
And these variables were correct by accident, as they had

the average value already, but MXG had 'SUM OF ALL':

R724USER &PIB.4. /*USERS*FOUND*/

R724ACTV &PIB.4. /*ACTIVE USERS*FOUND*/

so only the label was changed and no division required.
The IBM documentation in both the SMF Manual and DSECTS

are incorrect for USER and ACTV variables; it took real

RMF data to revise the code to decode the data.

Thanks to Lawrence Jermyn, Fidelity, USA.

Change 21.097 -Support for CA/TMS PDC # QI30130 which adds 4 variables:

TYPETMS5 to the TYPETMS (Volume) dataset:

VMACTMS5 COMPRES ='PCT SPACE*SAVED*DUE TO*IDRC COMPRESS'

Jun 8, 2003 CTLGCNT ='NR TIMES*DATASET WAS*UNCATALOGED'

FILPERC ='PCT OF*PHYSICAL SPACE*USED BY*FILE'

VOLPERC ='PCT OF*PHYSICAL SPACE*ON VOL*IN USE'

and adds the first 3 variables to the TYPEDSNB (file)

dataset.

-Support for TRTCH='EC'x,'ED'x decode for 3590-384T

(standard) and 3590-384X (extended length cart) 3590

devices with 384 tracks.

-Variables B1INT and B1DIS were replaced by the new fields

and no longer exist.

Thanks to Len Rugen, University of Missouri, USA.

Change 21.096 Support for GEPARMKY=001B (PRTQUEUE) creates new variable

IMAC6ESS ESSPRQUE in TYPE6 if IMAC6ESS is used to decode ESS data.

VMAC6ESS The 001B also caused INPUT STATEMENT EXCEEDED error if

Jun 6, 2003 had a tailored, earlier IMAC6ESS in your USERID.SOURCLIB.

Thanks to Reinhard Nitsch, Provinzial Versicherungen, GERMANY.
Change 21.095 -Support for six new TNG objects from NT:

EXTNT036 ACTIVE SERVER PAGES, DISTRIBUTED TRANSACTION,

EXTNT037 HTTP INDEXING SERVICE, INTERNET INFORMATION,

EXTNT038 PRINT QUEUE, and WEB SERVICE,

EXTNT039 that create NT036 thru NT041 datasets.

EXTNT040 -Support for new fields in LOGICALDISK, PHYSICALDISK,

EXTNT041 MEMORY, SYSTEM, and UDP objects.

FORMATS -Support for 1-Header record, found only in a "Monthly"

IMACTNG TNG Performance Cube.

VMACTNG

VMXGINIT

Jun 8, 2003

Thanks to Peter Krijger, National Bank of New Zealand, NEW ZEALAND.
====== Changes thru 21.094 were in MXG 21.02 dated Jun 5, 2003=========
Change 21.094 Calculation of MSU4HRAV in PDB.RMFINTRV was revised. It

VMXGRMFI was incorrect in several instances: after a change in RMF

Jun 5, 2003 interval value, or after an IPL, and it was calculated

when it should not have been, notably, if there was a gap

in the data. It can still be missing when SMF70LAC is

not, and it can be different from SMF70LAC when there are

not four hours of data available to MXG.

Thanks to Frank De Bree, DEXIA, BELGIUM.

Thanks to Mark Nouwen, DEXIA, BELGIUM.
Change 21.093 PDB.ASUMUOW corrections (SYSTEM could be blank) and logic

VMXGUOW revisions more robustly handle CICS and DB2 data merge.

Jun 5, 2003 -Several variables, including QWACAJST added to SUM= list.

-Added HOLDSTRT/HOLDEND to RETAIN, and a DATA step to get

the FIRST.UOWIDCHR values, to correct an exposure that

could have caused missing STRTTIME and wrong ELAPSTM.

-Corrected LAST.UOWIDCHR logic that was assigning blanks

for the SYSTEM variable for CICS-only work.

-Moved &OUTCODE to the end of the DATA step, for safety.

-Formatted WAITTOTM as TIME12.2.

-If an OBS was SPUN, when it came back in, if any MIN/MAX

variables had been specified, the values from SPIN were

not used in the MIN/MAX calculation, but now are.

These changes were suggested with implemenation examples!

Thanks to Scott Barry, SBBWorks, Inc, USA.
Change 21.092 User records RMDS,SYNC,WYLA,WYLB,HSM,TSMO,COM,M204 might

UTILBLDP have zero obs; those products have non-standard _aaaaID

Jun 4, 2003 macro names for their SMF IDs, or they create multiple

multiple SMF records and they all require separate tests,

but only 2 of 3 places had been fully updated to protect

all of these special user records. The error only

occurred whenf %LET MACKEEP= was used.

Thanks to Scott Barry, SBBWorks, Inc, USA.

====== Changes thru 21.091 were in MXG 21.02 dated Jun 3, 2003=========
Change 21.091 A major revision of support for RMF III VSAM files.

ASMRMFV Changes to ASMRMFV in particular are to position the code

CLRMFV for future upgrades to handle more RMF III tables.

DOCLRMFV -When ENDTIME of a MINTIME interval was the same as the

Jun 3, 2003 STARTIME of the next MINTIME interval, it was incorrectly

selected. RMF III almost always has the same time for

the end of one MINTIME interval and the start of the

next. Only noticed when testing for a single minute to

be selected, but two were output.

-Parameter processing completely rewritten to be table

driven, allowing new keywords and keyword aliases to be

easily added for new table support.

-Parm keywords have shorter aliases to fit in the JCL

limit of 100 characters for a parameter. The full list

of all alias names are given in the DOCLRMFV member.

NOxxx names added to assist.

-Max LRECL=32760 (increased from incorrect LRECL=32756).

-Date selection formats expanded YYYDDD YYYDDD YYDDD YDDD

DDD DD or D.

-Time-based selection added, with FROMTIME/TOTIME= parm

and HHMM, HMM, MM, or M formats, plus FROM/TO= nn, for

relative number of days before or after today.

-ASMRMFV output reports MXG Last Change Number in output,

includes DSNAME and filtering counts, and a summary if

multiple systems are processed, min/max times, etc.

NODETAIL keyword will suppress the detail reports.

-Debugging enhancement in RMFSKIP option.

-ENC record outputs only one per enclave, correcting case

where extensive use of Enclaves caused the RMF output

record to exceed the output LRECL limit.

-An (unexpected) Table ID mismatch will cause ASMRMFV to

issue a message and then ABEND.

Changes to the CLRMFV CLIST that executes ASMRMFV:

-The changes in CLRMFV are not backwards compatible; this

version must be used with the revised ASMRMFV program.

-PARM keyword default is no PARM('-D'), equivalent to the

old default of PARM('ALL,NODVT')

-New keyword RMFSKIP(NO) to control optional debugging.

-New keyword ONECALL(YES). This allocates all RMF III

Monitor Files for an LPAR first, and then calls ASMRMFV

once to process all records from that LPAR.

(Code ONECALL(NO) to invoke ASMRMFV once for each file,

which is how it was in prior versions).

The default YES should be best for typical use; ASMRMFV

specifies FREE=CLOSE internally so that each RMF III file

is freed after it was read, and the SHR enqueue released

as soon as possible. This produced a 9-11% reduction in

elapsed and CPU time with ONECALL(YES) default.

-Allocation messages show DDNAME

-New keywords FROMTIME/TOTIME added to support time based

data selection in ASMRMFV.
A special thanks to Acxiom management for allowing this

MXG user to contribute these enhancements for others!

Thanks to Jerry Urbaniak, Acxiom, USA.
Change 21.090 Multiple enhancements to CICS summarization:

ASUMCICS -Macro _BSUCICS is used in the BY list instead of the

ASUMCICX list of specific variables, to be consistent and in case

TRNDCICS of future changes. Aug 7: this change caused variable

TRNDCICX DATETIME to not be created in PDB.CICS in MXG 21.02-03.

Jun 3, 2003 STRTTIME, which has always been the correct variable to

Jun 4, 2003 use, was not affected. But Change 21.145 reinstated the

Aug 7, 2003 creation of variable DATETIME in PDB.CICS, so you would

not have to cange your reports to use STRTTIME.

-SHIFT and SYSTEM added to the BY list.

-Modified to carry the values used to set the buckets in

variables RESPVAL1-RESPVAL7, and comments placed around

the place to reset the values for "User Change".

-The FLOOR(IRESPTM) was removed, so that fractional values

can be used for the RESPVALn bucket values.

-ESUCICS is used to control the output of CICS dataset.

-Macro variables SUCIVAL1-SUCIVAL7 are created so that the

response buckets values can be set externally, and macro

variable SUCIINTV externalizes the summarization interval

(with HOUR as the default). The chosen interval value is

imbedded in the dataset label, while the response bucket

values are imbedded in the variable labels for RESPBKTx.

-For Trending, macro variable TRCIINTV sets the interval

default of WEEK.

Note: We generally do NOT recommend the use of ASUMCICS

and TRNDCICS, based on CICSTRAN dataset, because

CICSTRAN has an obs for each segment of an MRO

unit of work; CICSTRAN observation are incomplete:

- the TOR obs has no CPU time, but has the real

TRANNAME and LUNAME

- the AOR obs has all of the CPU time, but its

TRANNAME is 'CSMI' (the mirror transaction).

- observation counts are not transaction counts.

Instead, we recommend you include ASUMUOW first,

to create the PDB.ASUMUOW "unit of work" dataset,

which has one observation for each UOW with the

correct TRANNAME, LUNAME, CPUTM, IRESPTM, etc,

and then include ASUMCICX, which reads ASUMUOW to

create a PDB.CICS dataset that summarizes UOWs.

Thanks to Scott Barry, SBBWorks, Inc, USA.

Thanks to Pat Curren, SuperValu Inc, USA.
Change 21.089 New MICSMXG member contains notes and comparisons from

MICSMXG users who have replaced CA's MICS with MXG Software.

Jun 2, 2003
Change 21.088 Execution under ASCII only.

ANALCISH -VMACDB2H. NETSNAME was incorrectly TRANSLATEed; the

VMACCRAY statement TRANSLATE(NETSNAME,'00'X,'40'X); needs to be

VMACDB2H TRANSLATE(NETSNAME,'00'X,' '); to execute under ASCII,

VMACEREP so that blanks (which are '20'x on ASCII) are converted

VMACSFS to hex zeroes.

VMACTMDB -All TRANSLATE() statements were examined, and these

VMACTMS5 members had similar incorrect-under-ASCII-only-syntax:

VMACTMVS VMACCRAY, VMACEREP, ANALCISH, VMACSFS, VMACTMDB,

VMACTSOM VMACTMS5, VMACTMVS, and VMACTSOM

Jun 2, 2003

Thanks to Matthew T. Chappell, Queensland Transport, AUSTRALIA.

Change 21.087 New variable MPL74 is created in the TYPE74 dataset, as

ANALFIOE MPL74=SUM(DEVCONTM,DEVDISTM)/DURATM;

VMAC74 LABEL MPL74='IO*MULTI*PROGAMMING*LEVEL'

Jun 3, 2003 an estimate of the concurrent I/Os to this device.

Jun 4, 2003

New ANALFIOE member reads TYPE73, TYPE74 and TYPE78CF to

create two new datasets:

TYPE74OE - Sums the MPL74 for each LCU from TYPE74.

TYPE73OE - The LCUs from FICON channels are selected,

and their I/O is uniformly distributed to

each CHPID in that LCU, and each CHPID's

concurrent "Open Exchange" variables are

estimated in these variables:

MPL73 ='ESTIMATED*FICON*OPEN*EXCHANGES'

IORATE ='ESTIMATED*FICON*CHANNEL*IORATE'

OEDURATM='EST*OPEN*EXCHANGE*msec per io'

The TYPE73OE is just the original TYPE73

with these variables added for FICON CHPID.

The MPL73, or "Open Exchanges" appears to be a useful

new metric for detecting saturation in FICON channels;

See http://www.perfassoc.com/publishedpapers for Pat's

"Understanding FICON Channel Path Metrics" article.

Thanks to Dr. H. Pat Artis, Performance Associates, USA.
Change 21.086 Support for IDMS Performance Monitor type 30 subtype 3.

IMACACCT In addition to the user SMF record created by PERFMON,

VGETJESN that CA product also creates SMF type 30 subtype 3

VMAC30 records at transaction end. These are very-old-format

May 30, 2003 subtype 3 that do not contain JCTJOBID nor SUBSYS (fields

SMF30JNM/SMF30WID) causing TYPETASK NOT DECODED messages.

The record has a (very old length=14) Product section,

a (very old length=124) ID Section (causing INTBTIME

to be, and which caused ITSV to error these records as

it uses INTBTIME for its DATETIME validation!), a

Completion section, and a Storage Section, but it does

not have the UREC, CPU, PERF, OPER or EXCP sections.

The record is identifiable because PRODUCT (SMF30PNM)

is 'PERFMON'.

It does have useful information about each transaction,

in the unique accounting fields, with different fields

for batch, DC/UCF, and CICS access to IDMS.

-VGETJESN recognizes PERFMON records, sets TYPETASK='STC'

and SUBSYS='PERFMON'.

-IMACACCT contains a comment block around the PERFMON code

that creates these IDMS variables from the Account field

IDMSCPTM='IDMS*TRANSACTION*CPU*TIME'

IDMSTID ='IDMS*TASK*ID*NUMBER'

IDMSACT1='IDMS*BATCH*ACCOUNT*ONE'

IDMSACT2='IDMS*BATCH*ACCOUNT*TWO'

IDMSTRN ='IDMS*ERUS*TRANSACTION*ID'

IDMSTRM ='IDMS*ERUS*TERMINAL*ID'

IDMSOPR ='IDMS*ERUS*OPERATOR*ID'

IDMSTSK ='IDMS*DC/UCF*TASK*CODE'

IDMSLTE ='IDMS*DC/UCF*LTERM'

IDMSBLG ='IDMS*DC/UCF*BILLING*GROUP'

and when you remove the comment block, those variables

will then be output in TYPE30_V, PDB.SMFINTRV, XSMFINT

datasets. IMACACCT also populates INTBTIME if it is

still missing in these PERFMON type 30 records.

-VMAC30 was updated to output the IDMS variables if they

have been created; unrelatedly, those LENACCT1-LENACCT9

variables that exist will now be output in TYPE30_V; they

should have been there all along!

Thanks to Martin Legendre, Regie des Rentes du Quebec, CANADA.

Change 21.085 Protection for 0 length records, which are not created by

VMACNTSM NTSMF product, but have been unintentionally created by

May 27, 2003 user's SAS programs that copied raw data filed. On MVS,

the records actually have length 1.

Thanks to Uriel Carrasquilla, NCCI, USA.
Change 21.084 Type 42 Subtype 8 (which is not yet documented in the SMF

VMAC42 manual) caused ERROR.VMAC42.CLLEN and hex dump of the SMF

May 23, 2003 record. The record did not contain a CL segment, but MXG

did not test NRCL to see there was no segment before it

printed the error message. However, the record was then

deleted, whereas now it is output.

Thanks to Rita Bertolo, Canadian Pacific Railway, CANADA.
Change 21.083 With either RUNWEEK=WEEK or RUNWEEK=WTD, the rolling of

BLDNTPDB weekly datasets was occurring at the wrong time, causing

May 23, 2003 weeklies with no data.

Thanks to Terry Heim, ECOLAB, USA.

Change 21.082 GPARMKY=002F segment in SMF 6 Extended ESS data is now

IMAC6ESS recognized and decoded into new variable ESSNOTFY that

VMAC6 will be kept in TYPE6 dataset (only if IMAC6ESS has been

May 19, 2003 tailored and its commment block un-commented). Some

May 23, 2003 misspellings were discovered and corrected, and all of

May 26, 2003 the $VARYING fields are now translated to $EBCDIC; a few

had been overlooked.

-GEPARMKY=002E segment decoded, ESSULIB1-ESSULIB4 for up

to four DSNAMEs of USERLIBs.

-GEPARMKY=002B, 002C decode OUTDISP NORMAL/ABNORMAL in the

ESSOUTDB and ESSOUTDC flag variables.

-GEPARMKY=001B is now correctly decoded with 001B instead

of 000B in the IF test; variable ESSUCS is expanded to

four bytes to match the size of the UCS field.

-GEPARMKY=0021 is supported in ESSPMISG flag variable.

-A 002C segment with unexpected GEPARMLN=0 was found, so

all segments are protected for a zero length, and that

segment's variables are not created.

Thanks to Reinhard Nitsch, Provinzial, GERMANY.

Thanks to Denny Wong, New York Life, USA.

Change 21.081 The Undocumented +2 bytes for alignment are after SMSMC,

VMACCTLT not after DDSPOS, which caused wrong values for BLOCKCT

May 19, 2003 and DDSCSIZE thru DDSPOS. The +2 statement was moved.

Jul 8, 2003 -The BLKSIZE field had been relocated to the end of the

Jul 25, 2003 record as a 4-byte field, but that was overlooked in the

Jul 28, 2003 MXG support - 8Jul03.

-MXG 21.02 only: CTLT Version 5 created zero observations

because it has LRECL=460; MXG logic changed to test for

... 400 LE LENGTH LE 460 ....

in two places to read V4, V5, and V6 records. 25Jul03

Thanks to Francesco Bragadin, Phoneix Spa, ITALY

Thanks to Rob Gibson, CPR, AUSTRALIA.

Change 21.080 The example JCL procedure name of MXGSASV8 is now used in

JCLMNTH place of the old MXGSAS example name, to point new users

JCLMONTH to the correct, current JCL procedure example for SAS V8.

JCLWEEK

JCLWEEKT

May 19, 2003

Thanks to Matthew Collier, Oklahoma State University, USA.
Change 21.079 SMF 80 with APPLNAME less than 8 caused INPUT STATEMENT

VMAC80A EXCEEDED in RACFTYPE=20 segment, due to hard-coded input

May 19, 2003 with $EBCDIC8. Now, the RACFDLN=7, etc is honored.

Thanks to John Grasing, MetLife, USA.

Change 21.078 The optional CA-Dispatch variables are now automatically

IMACCADI added to PDB.PRINT dataset, when you open the comment

May 14, 2003 block in your IMACCADI member. By adding a %LET ADD6= ..

statement, the BUILDPDB &ADD6 macro will be added to the

KEEP= list for PDB.PRINT dataset.

Thanks to Dan Adams, NITC, USA.

Change 21.077 Variables EXCPDASZ EXCPTAPZ IOTMDASZ IOTMTAPZ RACFUSRZ

ANALDSET SELAPSTZ STEPTIME were not initialized, sometimes causing

May 14, 2003 their counterpart variables to be incorrect.

Thanks to Chuck Hopf, MBNA, USA.

Change 21.076 Variables QWACSPEA QWACSPEB QWACSPNE QWACSPTT for Stored

VMXGUOW Procedure activity are now included in the PDB.ASUMUOW

May 5, 2003 dataset.

Thanks to Chuck Hopf, MBNA, USA.

Change 21.075 This TCP/IP report example was truncating leading chars

ANALTCP of Host Name, and had additional revisions.

May 2, 2003

Thanks to R. Wells, American General Finance, USA.

Change 21.074 IBM documentation for the raw units of time in RMF III's

VMACRMFV DVT table was incorrect; instead of 128 microsecond units

May 2, 2003 the fields used to calculate CONNTM, PENDTM, DISCTM,

DVBUSYTM, CUBUSYTM and SWPODLTM are in 2048 microsecond

units, so those variables are now corrected in the ZRBDVT

dataset.

Thanks to Betty Wong, Bank of America, USA.
Change 21.073 Format MGPDSTY is updated for PDSMAN Version 7.41 which

FORMATS has new value "E:EXEC', used in TYPEPDSM support.

May 2, 2003

Thanks to Rob d'Andrea, Royal Bank of Scotland, SCOTLAND.

Change 21.072 ESAMAP v2.2 caused SHORT SEGMENT errors. MXG code had

VMACXAM not been tested with that earlier release. MXG code in

May 2, 2003 XAMSYS for MTRMEM, SYTSYG, and STOASI, in XAMCPU for

SYTRSP, and in XAMDEV for CONFIG and IODDEV was revised

to read the shorter segments.

Thanks to Tony Lobley, EDS, ENGLAND.

Change 21.071 If you used %VMXGTIME/TIMEBILD to shift time zones, you

VMAC30 could have negative RDRTM, and it was possible that

May 1, 2003 ALOCTIME and/or LOADTIME could have had the wrong date.

The calculation of RDRTM was moved above the %VMXGTIME

for REND, and the logic to set dates for ALOC/LOAD times

was precedeed by %VMXGTIME with REVERSE=YES, and then all

four datetimes (INIT,TERM,ALOC,LOAD) were un-REVERSED.

Thanks to Chuck Hopf, MBNA, USA.

Change 21.070 Variable RDRETDAT should have been FORMATed DATE9., and

VMACEDGR now it is.

Apr 30, 2003

Thanks to Barry McQueen, Department of Defence, AUSTRALIA.

Change 21.069 Exit member EXUTILEX allows you to process a non-standard

EXUTILEX "User" segment (i.e., your CICS guru created a segment

UTILEXCL but did not use "USER" for the CMODNAME/CMODHEAD entries.

Apr 25, 2003 Comments in EXUTILEX show how to reset the CMODNAME/HEAD

Nov 3, 2003 values to "USER" so that MXG will generate IMACICUS call.

You will also, then, have to update IMACICUS to tell MXG

the length of data to be kept in the USERCHAR variable.
Nov 3, 2003 documentation update:

Using this EXUTILEX member solved USERCHAR always blank.

One site's Hogan data segment was named "HOGAN", and so

the IMACEXCL member created by UTILEXCL did not contain

the call to IMACICUS, where the site's code to decode the

Hogan data was located, but the UTILEXCL program didn't

see a "USER" segment, so it did not generate an include

for the IMACICUS code, so USERCHAR was always blank.

By using this (new) EXUTILEX exit, you can tell UTILEXCL

to treat your "HOGAN" segment as if it were named "USER".

An error "BY VARIABLES MUST BE IN FIRST 4096 BYTES" was

seen with SAS Version 6, but not with SAS Version 8+, so

run UTILEXCL under V8+ to be error-free.
Thanks to Michael Creech, Fidelity Information Services, USA.

Thanks to Herb Strozinsky,US Bank, USA.

Change 21.068 MXG variable VELOCIOD when I/O Priority Management is NOT

VMAC7072 enabled (i.e., R723VELI=' ') did not match IBM RMF report

Apr 24, 2003 value: MXG was including PCTDLTDQ (R723CTDQ) in the denom

but IBM's calculation does not include that value. MXG

was changed to match the IBM reports, assuming they are

correct.

Thanks to Russell Dewar, National Australia Bank, AUSTRALIA.
Change 21.067 -SYSPLEX was not in the BY list for TRNDRMFI, but now is.

VMXGRMFI -Variable PCTMVSBY is added to PDB.RMFINTRV and TRNDRMFI.