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

Thanks to Jon Caldwell, U.S. Department of Veterans Affairs, USA.

Thanks to Mike Hampton, First Nationwide Bank, USA.

Change 13.285 Cosmetic documentation change. References to ANALDB2 were

ADOCDB2 changed to DIFFDB2, the "four datasets" note was changed

DIFFDB2 to "three datasets", and change 12.033 is referenced

Dec 12, 1995 instead of change 12.034.

Thanks to Nico Lenaerts, SAS BELGIUM, BELGIUM.
Change 13.284 REXX program to convert GTF trace records from DB2 into a

REXXDB2 legitimate (un-segmented) records had typographic errors.

Dec 12, 1995 -All C2K should have been C2X instead.

-The NE should have been <> instead.

-The statement I=REC must be changed to F=REC.

-The concatenation symbol '6A'x needs to be '4F'x for MVS.

That character is mis-translated between EBCDIC/ASCII by

many upload/download packages, so the actual change was

to replace F=F||G with F=(F)(G)so that the

REXX program is impervious to upload/download.

Thanks to Eric Thornton, D&B, USA.

Thanks to Chuck Hopf, MBNA, USA.

======= Attended CMG 95 Conference in Nashville, Tennessee ============
Change 13.283 Support for TANDEM D20, D30, and D40 releases is added

VMACTAND compatibly. However, I found I cannot trust the TANDEM

Dec 2, 1995 MEASURE documentation; its DLLs show changes where there

Dec 12, 1995 were none! (Fortunately, CMG came to the rescue as there

Jan 3, 1996 I met a TANDEM employee who put me in touch with the real

programmer who wrote the code!). Two variables were added

compatibly by D30 (BEGTRANS,ABRTRANS) by using reserved

space in the PROCESS record. Several measurement fields

(lock-pages-qtime/count and UCL-lock-qtime/count in the

PROCESS record, and the four pairs of START/END variables

for UDS-LOCK, SDS-LOCK,UCL- LOCK, and SCL-LOCK in the

DISC record) were made reserved fields in D40 (because

they were too expensive to capture!). The DDL for D40 are

wrong, as they show BEGTRANS/ABRTRANS in the wrong place,

and the now-reserved fields were deleted from the DDL,

but they were not deleted from the physical record.

Thanks to Joe Fleischmann, US Bancorp, USA.

Thanks to Todd Tomita, US Bancorp, USA.

Thanks to Steve Smith, BGS Systems, USA.
Change 13.282 MAINTLEV 7 of the MXG Tape Mount and Allocation Monitor

ASMTAPES corrects the JSCB access problem, the CA-11 restart case,

Nov 30, 1995 and supresses the SRB dump messages (unless we ask you to

enable DEBUGGING!). This iteration has been running in

two sites for several weeks with no failure. The previous

monitor code was copied into ZSMTAPES for backup.

Change 13.281 This replacement for member CICINTRV is temporarily put

CICINTRZ in this member for extensive testing, but it will become

Nov 30, 1995 CICINTRV in the near future. The present CICINTRV logic

is incorrect, and this new logic correctly creates the

CICS interval datasets from the statistics datasets.

This version first summarizes the individual datasets at

the lowest level, and performs deaccumulation with DIF()

function for the REQ and USS records so that all four

types of CICS stat records are correctly summarized into

the CICINTRV dataset. Note that this can be resource

intensive if you have lots of CICS activity, but you can

construct the CICS activity from these statistics records

even if you have turned off CICSTRAN creation!

Revised Jun 13, 1996: Resources are not a problem; see

resource measurements in text of Change 14.132, which

implemented these enhancements starting with MXG 14.04.

Thanks to Chuck Hopf, MBNA, USA.
Change 13.280 Correction. If no summary dataset was created, but

ANALCNCR summary reports were requested, the summary reports were

Nov 30, 1995 not produced.
Change 13.279 New parameters SMFBEGIN and SMFEND were added to allow

READDB2 selection while the raw SMF records are read. These new

Nov 30, 1995 parameters are now used by ANALDB2R (Change 13.278).
Change 13.278 Several enhancements to DB2 reporting.

ANALDB2R -Reports can now be produced from MNTHxxxx datasets, if

Nov 30, 1995 you have used the MNTHxxxx members to trend monthly.

-When reading SMF, the BEGTIME and ENDTIME values are now

passed to READDB2 (as SMFBEGIN/SMFEND) so that selection

applies to the raw data as it is read, which will reduce

DASD space and run time, especially with big traces.
Change 13.277 This utility (used only in JCLTEST6, to select ten SMF

VMXGGETM records of each type) has new INCODE= operand added to

Nov 30, 1995 enhance selection criteria, for those of you who have

found this utility useful! You could now code

INCODE=IF (ID=30 AND 4 LE SUBTYPE LE 5) OR ID=72; ,

to select only those records and subtypes.

Thanks to Chuck Hopf, MBNA, USA.
Change 13.276 The revised VMXGSUM logic has been moved from XMXGSUM to

VMXGSUM VMXGSUM, and member XMXGSUM has been deleted. (Just in

XMXGSUM case, the old VMXGSUM was copied into ZMXGSUM for backup,

ZMXGSUM but that member too will go away in time). The new logic

Nov 30, 1995 in VMXGSUM will significantly reduce the DASD space, CPU

time and run time, as it keeps only the variables that

are actually needed by the summarization, and (unlike the

old VMXGSUM), it does not create dummy variables in

the output dataset. It also supports variable lists with

hyphenated syntax. Many sites with large data volumes

have been using the XMXGSUM logic, so I believe it is now

safe to make the MXG default to be the new logic.

Change 13.275 New parameters INTERVAL and MYTIME are defined for report

ANALRMFR summarization, but they are only implemented in MXGCHAN

Nov 30, 1995 report at this time.
Change 13.274 CICS shutdown reports CICCONSR or CICCONMR can cause many

ANALCISH blank pages with only the heading and no content; several

Nov 30, 1995 line changes were required, too complicated to show here.

Also, END; statement was missing after IF INOBS EQ CXMC.

Thanks to Richard S. Ralston, Whirlpool. USA.
Change 13.273 Support for DB2 4.1 type 102 trace records has tested the

VMAC102 new IFCIDs 221, 222, and 231 for parallel group tracing,

Nov 30, 1995 and adds new fields in existing trace datasets for IFCIDs

8, 10, 20, 21, 22, and 28. There are still other fields

added to other IFCIDs by 4.1 that are not yet decoded by

MXG due to absence of test data records; those will be

added when user demand and test data arrive together.

Thanks to Ted Blank, IBM, USA.

Change 13.272 Corrections to several variables in HP PCS records:

VMACHPAI -VMACHPUX, comment now has HPUX as correct INFILE name.

VMACHPSU -VMACHPUX, PIN variable removed from LENGTH statement so

VMACHPUX that it will be numeric rather than character (as it was

Nov 29, 1995 in HPAI and HPSU members, like all other PINs).

-INTEREST now input as INTEREST $CHAR12. (instead of with

no INFORMAT) as the file may contain leading blanks.

First test for INTEREST that sets IMPWTHI was deleted.

Last four tests for INTEREST must test for lower case

letters c,d,m,i rather than upper case values.

Thanks to Thierry Van Moer, Procter & Gamble Europe, Belgium.
Change 13.271 Variables SYSNAME and SYSPLEX were not kept in BUILDPDB

IMACPDB datasets PDB.JOBS, PDB.STEPS, and PDB.PRINT, but they are

Nov 29, 1995 now added to _PDB30_1, _PDB30_4, and _PDB30_5 macros in

member IMACPDB so they will be kept in the PDB datasets.

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

Thanks to Harmuth Beckmann, Karstadt, AG, Germany

Change 13.270 IBM writes truncated EREP records, but MXG did not catch

VMACEREP the truncation, causing INPUT STATEMENT EXCEEDED RECORD

Nov 29, 1995 for a hardware detected VLF software record. Protection

was added for dataset EREPSDW. Additionally, the ERRORID

field at the end of the SDW record was not input from the

correct location, and variable CLASRC is now kept in the

EREPSDW, EREPEOD, EREPMDR, and EREPOBL datasets.

Thanks to Solomon Baker, The Prudential Service Company, USA.

Change 13.269 Variables QBSTHPL and QBSTVPL were removed from DIF()'ing

DIFFDB2 because they are not accumulated values, but rather are

Nov 28, 1995 the number of hiperspace and virtual pool buffers.

Thanks to Alan Fendler, Pershing Info Management Services, USA.

Change 13.268 Variable USER was added to the default summarization of

ASUMCICS CICSTRAN dataset into PDB.CICS, because OPERATOR is now

Nov 28, 1995 usually blank, while USER contains the wanted user-id.

Dec 18, 1995 The summarization default is now by

APPLID OPERATOR USER TERMINAL TRANSACT (for each hour).

Thanks to Clark Jennings, Reynolds Metal, USA.

Change 13.267 Another IDMS 12.01 error, INVALID DATA FOR PMHSDATE for

VMACIDMS the PMHRTYPE=6 (Journal Wait) subtype. The statement

Nov 20, 1995 SKIP=SKIP-108; that is two lines prior to the statement

%%INCLUDE SOURCLIB(EXIDMJRL); should been

SKIP=SKIP-104;. (My test data stream did not have any

journal wait data, but this was an MXG coding error.)

Thanks to Dan Gilbert, Bergen Brunswig Corporation, USA.
Change 13.266 Variable STARTIME was missing TSOMCMND if the TSO/MON

VMACTSOM SMF record was written in segments (because there were

Nov 20, 1995 more logged on users that would fit in one SMF record).

STARTIME was added by Change 13.089 for consistency, but

the pre-existing, same value variable STRTTIME was never

wrong! Immediately following STRTTIME=TSOMSTAR; insert

STARTIME=STRTTIME;

Thanks to Neil Ervin, Huntington Services Co, USA.

Change 13.265 Support for IMS 5.1 records (INCOMPATIBLE) was reported

ASMIMSLG with these changes.

VMACIMS -ASMIMSLG - Replace these three non-contiguous lines

Nov 17, 1995 TM MSGCFLG1,MSGC1RAC TM MSGCFLG1,MSG3RACF

---17 lines ----

USING MSGRACF,R4 USING MSGSEC,R4

--- 8 lines ----

MVC ORGENT(8),MSGRACGP MVC ORGENT(8),MSGSAFNM

(with this change to ASMIMSLG, it can ONLY process 5.1

records, so you will need to maintain two separate load

libraries and separate job streams).

VMACIMS

18 new 4-byte fields were inserted in the 07 log record

between MSGGCMD and PDATE.

This is a documentation only change at this time, as I am

still awaiting data and documentation so that I can

validate this report and then change the MXG coding.

Now, see Change 14.030.

Thanks to Mr. Hellmann, Sudwestdeutsch Landesbank, GERMANY.
Change 13.264 TANDEM disk type format MGTANDS values are decimal, not

VMACTAND hex, so the "X"'s were removed, and the 35:3GB value has

Nov 16, 1995 replaced the 35:MGB spelling.

Thanks to Steve Smith, BGS Systems, USA.

Change 13.263 IBM lied, and JESNR may show only four digits in TYPE26J2

VMAC26J2 dataset (and if BUILDPDB finds only a purge record for a

Nov 13, 1995 job, its PDB.JOBS observation will have JESNR=1179 where

Feb 26, 1996 it should be JESNR=11179. IBM documentation of SMF26JNM

(the old, 4-position EBCDIC JESNR) says it will be zero

if the job number is 10,000 or greater, causing MXG to

get JESNR from SMF26JID, and this was true until now, but

it appears MVS/ESA 5.2 with JES x.y are now putting the

truncated JESNR back into SMF26JNM! While I chase after

the IBM INCOMPATIBLE change to type 26 record, I can fix

the MXG logic to work no matter what IBM does. Change:

ELSE DO;

IF JESNR GT 0 THEN

INPUT @57+OFFSMF TYPETASK $EBCDIC3.

+5

@;

ELSE

INPUT @57+OFFSMF TYPETASK $EBCDIC3.

@60+OFFSMF JESNR &NUM.5.

to read:

ELSE DO;

INPUT @57+OFFSMF TYPETASK $EBCDIC3.

@60+OFFSMF JESNR &NUM.5.

Feb 26, 1996 update: IBM APAR OW18822 acknowledges the

error and should correct the non-zero value back to zero,

but the MXG correction in MXG 13.13 fixes it anyhow!

Thanks to Tim Terbieten, Newell Company, USA.

Change 13.262 Variable DEVPLX, the device address of the duplex volume,

VMACACHE is an offset from the first device instead of the real

Nov 8, 1995 device number; now, DEVPLX will contain the true device

Dec 2, 1995 number by inserting these lines:

IF NDVCNT=1 THEN BASDEVN=DEVN;

IF DEVS1='....1...'B THEN DEVPLX=BASDEVN+MOD(DEVS2,64);

before the %%INCLUDE SOURCLIB(EXCAC90); statement.
I note that the Cache record does not contain a segment

for the duplex device; DEVPLX=05x, BASDEVN=2C0x, so the

duplex device address is now DEVPLX=2C5x, but there will

be no observation in CACHE90 with DEVN=2C5x.

The above correction worked until MVS/ESA 5.2, which has

caused an unexpected (at least by the CRR-folks) change

in the CRR record. The BASE device number used to be the

first device segment returned by the 3990 controller, and

that address is copied into the statistics segment. But

in 5.2, the order in which devices are varied online at

IPL is different, and the 3990 returns devices in order

they came online, so the base device is no longer going

to be in the first segment. IBM CRR Level II has this;

when they decide what they are going to do, so will I!

Thanks to Kurt Koch, West Publishing Corporation, USA.
Change 13.261 SAP Journal segments in type 110 records caused error

VMAC110 INPUT STATEMENT EXCEEDED, or INVALID DATA FOR HH, or did

Nov 8, 1995 not read in all segments in the SMF record, because MXG

Jan 13, 1996 did not anticipate that SAP would create journal segments

with only a header,

(found for the YISA APPC host-to-host connection

application, with JCSPTRAN='YISA', JCRLL=30, so there

is no data - these segments may be output into a new

dataset if there is usefulness, and this fix still

leaves them available in the EXCICJRN exit),

or with JCRUTRID not containing 'SA',

(found for a most strange segment between other 'SA's,

containing a Global Performance Interval segment with

MCTSSDID=2 which belongs in a subtype =1 record and

is normally output into CICSYSTM from that subtype!;

UPDATED Jan 13. SAP Technical Support has not responded,

but the second problem is circumvented in Change 13.323.

Header only segments are skipped over with these changes:

-Delete the line ... INPUT +6 SAPTEST $EBCDIC2 @LOC @;

-Change IF SUBTYPE=0 OR SAPTEST='SA' THEN DO UNTIL ...

to IF SUBTYPE=0 THEN DO UNTIL ...

-Change IF JCRUTRID='SA' THEN DO;

to IF JCRUTRID='SA' AND JCRLL GE 250 THEN DO;
Unrelated to the above errors, variable APPLID was added

to the KEEP= list for the CICSSAP dataset so CICSSAP can

be analyzed for each CICS region.

Thanks to Jens Schlatter, EDP Consulting Schlatter, GERMANY.

Thanks to Norbert Korsche, OMV, AUSTRIA.

Thanks to Paolo Carloni, Agip petroli SPA, ITALY.

Thanks to ????, Deutsche Post AG, GERMANY.
Change 13.260 RMDS 1.4 records may cause INVALID ARGUMENT TO MDY AT ...

VMACRMDS because only some MDY() functions were protected for the

Nov 8, 1995 'strange' values MO=99 and DD=99.

Feb 26, 1996 -Now, all uses of MDY() are protected with logic of the

form:

IF YY GT 0 AND (YY NE 99 AND MO NE 99 AND DD NE 99)

THEN xxxxDATE=MDY(MO,DD,YY);

-In addition, INVALID DATA FOR MM can occur, because only

some INPUTs of HH MM and SS were protected with the ??

modifier. Now, all fields input with &NUM are preceded

by the double-question-mark modifier.

-Finally, all HMS() functions are now protected with

IF 0 LE HH LE 24 and 0 LE MM LE 60 and 0 LE SS LE 60

logic to prevent invalid arguments to HMS() function.

-The error does not occur with current RMDS 2.1 or later.

Note added Feb 26, 1996: The support for RMDS 1.3/1.4

also deleted the two tests:

IF RMDSACT='D' and RMDSORG NE 'A' THEN RMDSACT='T';

IF RMDSACT='U' AND RMDSORG EQ 'V' THEN RMDSACT='S';

because the activity codes of 'T' and 'S' do not exist

in RMDS version 2.1.

Thanks to Ambat Ravi Nair, Trident Infotech Pte Ltd, SINGAPORE.

Change 13.259 MXG 13.06-13.07 only. ABARS enhancement validation:

EXHSMWWV -UNEXPECTED IDHMSMDS RECORD FOUND because the line now

IMACHSM reading ELSE IF DSRVSR='VRS' THEN ... should have been

VMACHSM ELSE IF DSRVSR='VSR' THEN ....

Nov 8, 1995 -INPUT STATEMENT EXCEEDED RECORD LENGTH on ABARS subtype

Nov 15, 1995 because the four fields WFSRML0U,WFSRML1U,WFSRML2U, and

WFSRTOTU at the end of both ABARS segments are now INPUT

as $EBCDIC1. instead of &PIB.4. MXG now decodes the unit

of space value (blank, K, M, etc.) and converts the space

used during ABACKUP variables WFSRML0S,L1S,L2S,TOTS into

bytes, and are formatted with MGBYTES to print pretty.

-Variable WFSRABCC is now input as $EBCDIC4. vice &PIB.4.

-IBM clarified several issues. Space units of K,M,G,T are

1024 (as expected, but since IBM used 1000 for hardware

"K", and since the ABARs documentation did not say, we

had to ask!). Also, ABACKUP VERIFY does cut a shorter

record that does not contain the space information fields

while ARECOVER creates a longer record, but zeroes out

the space information fields.

-New HSM dataset HSMWWVOL is created, but with zero obs

until you remove the comment block in member EXHSMWWV.

(This dataset will contain one obs for each volser that

was used by ABARS for backup, and I perceive little need

for that information; the useful ABARS information is in

the HSMWWFSR dataset, one obs per ABARS event.)

Thanks to Michael E. Friske, Fidelity Savings, USA.

Change 13.258 Very obscure, and only for the early users of XMXGSUM.

XMXGSUM Change the second occurrence of NUMPOS= from &HYPHEN1 to

Nov 6, 1995 &HYPHEN2. Would have caused an OUT OF MEMORY error.
Change 13.257 Variable AVGQUETM should not have been in the keep list

VMAC7072 for dataset TYPE72GO, as that field is from the subtype 2

Nov 6, 1995 (RMF Monitor III) record, and is output only in TYPE72MN.

Thanks to Don Deese, (CPExpert), Computer Management Sciences, USA.

Change 13.256 Variable QBGAGN should be kept only in DB2ACCTG, and not

VMACDB2 in DB2ACCT, and it should not have been summed during the

Nov 6, 1995 creation of DB2ACCTG. It is the pool ID, not a counter!

Thanks to Chuck Hopf, MBNA, USA.

Change 13.255 Tandem variables CnMISSES was repeated in INPUT; the two

FORMATS instances are now named CnRMISS and CnWMISS for Read or

VMACTAND Write misses. C1BLKS is no longer divided by DURATM, as

Nov 6, 1995 it is blocks allocated, not blocks moved in interval.

Format MGTANDS was updated with device 1Fx and will now

print un-coded values in hex rather than decimal.

Thanks to Steve Smith, BGS Systems, USA.
Change 13.254 Support for TOPSECRET Release 4.4 and 5.0, INCOMPATIBLE,

VMAC80 CAUSES INPUT STATEMENT EXCEEDED RECORD error, because the

VMAC80A new release sets a value of 44X or 50X for RACFVRSN, but

Oct 31, 1995 MXG does not know in advance what value TOPSECRET will

use! Add OR RACFVRSN=44X OR RACFVRSN=50X to the test

for TOPSECRET in both members. (44X exists in 4.4 data;

5.0 is not out yet, so I am gambling in advance that they

will use 50X for that version when it's released!).

Thanks to Mark Paulson, Maurices Incorporated, USA.

Thanks to Sarah Gartner, Hudson's Bay Company, CANADA.

==========MXG Version 13.07, dated Oct 30, 1995, thru 13.253==========
Change 13.253 SAP 5.0.E creates invalid journal segments, which caused

VMAC110 INVALID DATA FOR HH and a hex dump of the record, but MXG

Oct 27, 1995 successfully skipped over the invalid segments and output

the valid ones. SAP had no one in tech support on Friday

to discuss their error before MXG 13.07 was built, but I

will pursue this with them later and update this note.

To eliminate the hex dump and message, insert ?? after

HH and MM following the INPUT of JCSPTASK.

Update: See Change 13.323.

Thanks to Paolo Carloni, Agip petroli SPA, ITALY.

Change 13.252 $AVERS SMF record variables SAVPAGES,SAVBLKS,SAVTBLKS are

VMACSAVR now created; these fields existed at the end of the SMF

Oct 27, 1995 record, but were not populated until this user went to

use them! The vendor, Software Engineering of America,

will fix the error in SAVRS V4.0A.33, and interim fix

number S40AF166 (a ZAP) is available from them now that

will populate these fields.

Thanks to Bill Hamilton, Scottish Widows, SCOTLAND.

Change 13.251 Support for STK's SILO SMF HSC View Subtype 8 record now

EXSTCHSV creates dataset STCHSV for every successful VIEW command

IMACSTC initiated by HSC.

VMACSTC

Oct 26, 1995

Thanks to Cheryl Howard, Wachovia Corporation, USA.

Thanks to Rodney L. Reisch, Wachovia Corporation, USA.
Change 13.250 DURATM was added to TSO/MON datasets by Change 13.089,

VMACTSOM but it was often missing! After the second statement

Oct 26, 1995 INTRVTM=TSOMDUR; insert DURATM=INTRVTM;

Thanks to Dan Squillace, SAS Institute Cary, USA.

Change 13.249 Support for MODEL204 Release 3.0 INCOMPATIBLY added five

IMACM204 fields to the SINCE record. Since there is no release

VMACM204 number in their record, you MUST update member IMACM204

Oct 25, 1995 in your USERID.SOURCLIB to use this new IMACM204, as it

now also defines macro _M204VER which tells MXG to read

Release 2 or Release 3 records (default is for Rel 3).

Thanks to Mark Wessel, Population Census & Surveys Office, ENGLAND.
Change 13.248 Summarization of IDMSTAS dataset from IDMS Perfmon into

ASUMIDMS PDB.ASUMIDS is provided by this user contribution, which

Oct 25, 1995 mimic's ASUMCICS algorithms to create response time and

resources by CV_NUM DC_USER DC_LTERM TASKCODE ADSODLGN.

Thanks to Richard S. Ralston, Whirlpool, USA.
Change 13.247 CICS/ESA 4.1 or later now contains the GMT offset, so MXG

VMAC110 can finally convert the STRTTIME/ENDTIME in CICSTRAN to

Oct 25, 1995 your local time of day. The INPUT of MCTMNTAD is changed

from &PIB.4. to &IB.4., and this line inserted:

MCTMNTAD=1.0485582324*MCTMNTAD;

to convert from CICS timer units to seconds. Then, in

the CICS/ESA 4.1 section for CICSTRAN and CICSEXCE, the

logic IF MCTMNTAD GT . THEN DO;

STRTTIME=STRTTIME+MCTMNTAD;

ENDTIME =ENDTIME +MCTMNTAD; END;

was inserted to add the (negative in USA) GMT offset.

NOTE: IF YOU HAVE TAILORED member EXCICTRN to convert the

CICSTRAN timestamps (as was described in Newsletter 27),

YOU MUST REMOVE OR REVISE YOUR conversion logic so that

it only converts non-4.1 records. For example, you could