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

VMXGVTOF variables are created. Both members were effectively

Jul 10, 1997 replaced by DCOLLECT: see TYPEDCOL/JCLDAYDS, although

the archaic members are still useful, especially if you

need to know the physical location of extents, since the

DCOLLECT records only contain total sizes of datasets.

Thanks to Mark Zelden, CIBER Network Services, Inc, USA.

Change 15.153 Cosmetic. The LABELs for LPAR variables are now grouped

ASUM70PR by LPAR number, so all variables for LPAR 0 are together

Jul 9, 1997 for ease in viewing; the global CPU and Percentages are

also grouped together, as the 1st variables in ASUM70PR.

Thanks to Jean Quinkert, Inland Steel, USA.
Change 15.152 Formats $MGHEX2H, $MGHEX4H, and $MGHEX8H are added to

FORMATS MXG to print hex character variables's blank values as

Jun 29, 1997 blanks instead of '40's. See MXG Tech Note, NL 32.
===Changes thru 15.151 were included in MXG 15.03 dated Jun 30, 1997===
Change 15.151 Installation tailoring code that defines your SHIFT value

IMACSHFT has added an example of a table of holiday's dates that

Jun 29, 1997 can be used to change Prime/Non-Prime Shift observations

to Weekend Shift, if that is what your manager says to

do because the numbers are skewed. (Personally, I prefer

to leave the shift asis, and observe the week-to-week

difference between weeks with holidays and weeks without,

but MXG is all about freedom and doing it your way!)

The example is commented out, so this change did not

really change anything except comments in IMACSHFT.

Thanks to Mark Zelden, CIBER Network Services, Inc, USA.
Change 15.150 INPUT STATEMENT EXCEEDED with Omegamon VTAM 200 IRNUM=12

VMACOMVT record; variables ON12OQ, ON12SP, and ON12SS must be

Jun 28, 1997 INPUT as &PIB.4. instead of &PIB.8., and variable ON12DX

must be INPUT as $EBCDIC4. instead of &PIB.8. This is

the first instance of anyone processing this subtype.

Thanks to Rich Anderson, SAS Institute Cary, USA.

Change 15.149 MXG 15.03 corrections uncovered in final final tests:

TESTWWW -Step TESTOTHR of JCLTEST6 fails because of the extra

DIFFDB2 close-parenthesis in the DATA statement in member TESTWWW

TRND72GO (twice I caught this and thought I had fixed it!).

NEWSLTRS -"UNINITIALIZED VARIABLE BHHITRAT" in building DB2STATB;

EXNTCMBU BPHITRAT was spelled BHHITRAT in one LABEL statement in

IMACEXCL member DIFFDB2, causing variable BPHITRAT to be created

Jun 28, 1997 without a label. See Change 15.xxx.

-TRND72GO statement LENGTH SERVICE 8 MSOUNITS; caused a

very strange %MACRO compiler error; but the correct

statement should be LENGTH SERVICE MSOUNITS 8;

-Four Thousand lines were deleted from member NEWSLTRS;

The text of changes in the Change Log in Newsletters

TWENTY-SEVEN and TWENTY-EIGHT should have been removed

from member NEWSLTRS, as that text is kept in member

CHANGESS, and a redundant copy is not needed!

-Comment in EXNTCMBU changed to CMPQCMBU vice CMPQBUSU.

-Comments in IMACEXCL were clarified extensively.

Thanks to Freddie Arie, Lone Star Gas, USA.
===Changes thru 15.148 were included in MXG 15.03 dated Jun 26, 1997===
Change 15.148 Omegamon BSC data had incorrect GMTOFFTM calculated; The

IMACICOC variable OMGMTCHR was truncated to only one byte because

Jun 26, 1997 it was in the INFORMAT ... $NOTRAN. statement; it should

be removed from that INFORMAT statement.

Thanks to Dale Wiles, Perdue Farms, USA.
Change 15.147 NTSMF Version 2.0.d requires MXG 15.03 dated Jun 26, or

VMACNTSM at least MXG 15.02 plus this change. MXG coding errors

Jun 26, 1997 will cause error messages "UNEXPECTED INSTANCE FIELDS" or

INVALID DATA FOR MO IN LINE .... and/or STOPOVER ABEND.

-The LENGTH statement VERSION $4. must be changed

to VERSION $5.

-The two tests for Reduced or Extended Header must be

changed to read:

IF VERSION GE : '2.1' THEN INPUT /*REDUCED HEADER*/

and

IF '2' LE : VERSION LT : '2.1' THEN INPUT /*EXTENDED*/

instead of the tests for '2.0.d' and LE '2.0.c'.

Thanks to Chuck Hopf, MBNA, USA.
Change 15.146 IBM Item RTA000099957 (see DB2 Technical Note 2 in NL 32)

DIFFDB2 states that the calculation of Buffer Pool Hit Ratio

ANALDB2R is (QBSTGET-(QBSTSIO+QBSTDPP+QBSTLPP+QBSTSPP))/QBSTGET;

Jun 26, 1997 whereas MXG used QBSTRIO instead of QBSTSIO in ANALDB2R.

I have added variable BPHITRAT to the DB2STATB dataset,

and variables B1HITRAT-B4HITRAT to the summary DB2STATS

dataset. I expect to alter ANALDB2R but want to verify

first (and its an extensive change). See the discussion

of why you can get negative values for Hit Ratio in DB2

Technical Note in Newsletter THIRTY-TWO.

===Changes thru 15.145 were included in MXG 15.03 dated Jun 25, 1997===
Change 15.145 Minor cosmetic changes were made to TRND72GO (Labels, and

TRND72GO DROPs), and 3 documentation-only members: ACHAP99, DOCVER

Jun 25, 1997 and DOCVER15 were updated with 15.03 dataset contents,

and counts in AAAAAAAA & COPYRITE now are for the real

MXG 15.03 dated Jun 25, 1997. Only ten tapes to trusted

sites carried the Jun 24, 1997 date.

===Changes thru 15.144 were included in MXG 15.03 dated Jun 24, 1997===
Change 15.144 TYPERDS records created 0 observations; these corrections

VMACRDS were required: INPUT of H15FTP is &PIB.2. vice &PIB.4.,

Jun 24, 1997 the INPUT of H15HSQ is &PIB.4. vice &PIB.2., and the

test IF H15FOS+H15FL GT ... must be changed to

IF H15FOS+H15FL-1 GT ....

Thanks to Shana Bereznay, Southern California Edison, USA.

Change 15.143 This change was NOT made. After planning the change and

BUILDPDB writing the change text, I realized I could not make the

BUILDPD3 change as described without a negative impact on existing

BUILD005 MXG BUILDPDB/BUILDPD3 users: BUILDPDB deletes the STEPS

BUIL3005 dataset from WORK as soon as PDB.STEPS is created (so its

Jun 24, 1997 workspace can be reused), but keeping WORK.STEPS would

require more WORK space - an existing daily job could

ABEND, needing more //WORK space, if I made the change.

I will revise the change and revise its description when

I can figure out how to meet IT Service Vision's request

without impact on existing MXG BUILDPDB jobs.
For IT Service Vision. After MXG's BUILDPDB logic has

created the PDB.STEPS dataset, SAS's IT Service Vision

uses the WORK.STEPS dataset left in the WORK file. But

then MXG logic to create PDB.SPUNJOBS replaces the real

WORK.STEPS with only the spun steps, because I reuse the

_PDBSUM macro that is defined in IMACPDB. I considered

changing that macro in IMACPDB to use a symbolic for its

output dataset, but that implementation would have caused

incompatibility for any user that had tailored IMACPDB!
Instead, by adding PROC DATASETS to rename WORK datasets:

PROC DATASETS NOLIST; CHANGE STEPS=REALSTEP;

%INCLUDE SOURCLIB(SPUNJOBS);

PROC DATASETS NOLIST; CHANGE STEPS=SPUNSTEPS;

PROC DATASETS NOLIST; CHANGE REALSTEP=STEPS;

both WORK.STEPS and WORK.SPUNSTEP will now exist after

BUILDPDB runs. By making this change in MXG, IT Service

Vision installers will have one less option to tailor.

Again, this change was NOT made. These are notes only,

that will likely be revised in the future.

Change 15.142 New exit EXPDB30V is added to the BUILDPDB/BUILDPD3 logic

EXPDB30V to allow tailoring of the PDB.SMFINTRV after TYPE30_V has

BUILDPDB been merged with the Accounting Information. While the

BUILDPD3 IMACINTV exit can be used to tailor TYPE30_V (which

BUILD005 becomes PDB.SMFINTRV), the accounting fields do not exist

BUIL3005 when IMACINTV is invoked, so the new exit allows for

Jun 23, 1997 more flexible tailoring.

Thanks to Diane Eppestine, Southwestern Bell, USA.

Change 15.141 ASMTAPES ML-14 populates fields in the Tape Mount record

ASMTAPES (COMPATIBLY) and provides some enhancements to the AR

Jun 23, 1997 mode code to prevent the S0C4 abends with swapped out

address spaces. It does not completely conceal the S0C4

abends (that is still planned) but this should improve

the situation significantly.

Change 15.110 describes the corresponding changes to the

VMACTMNT member to process the newly populated fields,

but as the record format was not changed by the monitor,

earlier versions of MXG can read the ML-14-created SMF

records compatible. You will want to use VMACTMNT from

MXG 15.03 or later to exploit and decode the new fields:

The ALOCSTRT is of value when working with tape mounts

for DFHSM where some DDNAMEs (eg., RESIN1, RESIN2,...)

are re-used many times (dynamically allocated) within

one (long-running) step. The DDNAME is of value when

working with tape mounts for DFHSM were some devices

are allocated multiple times with different DDNAMEs

within a step.

Mark also provided corrections for truncation of JESNR to

four digits in the mount record processing; he found that

the ASMTAPES logic did not replace ALCSTIME with ALISTIME

for INTERVAL records cut after a day boundary has been

crossed; he provided his ASM and SAS code to to add the

above fields; and he found a nit in one series of tests!

Mark is a fine example of an MXG CodeShark!

Thanks to Mark Keimig, VF Information Technology Services Inc, USA.
Change 15.140 Support for new fields in the DFSMSrmm TYPEEDGR extracts

FORMATS that were COMPATIBLY added at some time in the past.

VMACEDGR Dataset EDGRDEXT has new variables:

Jun 23, 1997 RDABEND RDCAT RDCRTJBN RDDCNAME RDDDNAME RDDSSIZE

Jun 26, 1997 RDLABNO RDMDMVID RDRETDAT RDSCNAME RDSGNAME RDSTEPNM

RDVRSJBN RDVRSNAM RDVRSR RDVRSTYP

and new format $MGEDGRT is added to member FORMATS.

Dataset EDGRSEXT has new variable RSBMEDN.

Dataset EDGRVEXT has new variables:

RVABEND RVACTERA RVACTINI RVACTNOT RVACTREP RVACTRET

RVACTSCR RVBMEDN RVDSNREC RVHOMTYP RVLUDEV RVMVMODE

RVNEXTYP RVNLOC RVOBMEDN

27Jun97: Variables RVNVOL RVNPVOL and RVMDMVID were

labelled and added to KEEP for EDGRVEXT dataset.

Thanks to Sam Bass, McLane Co., USA.
Change 15.139 When Landmark maintenance TT00032 was applied after the

TYPETMON March cum tape was installed (to fix a Landmark 0C4 in

TYPEMON8 their dump program), the dump program creates a garbage

Jun 23, 1997 first record, causing MXG to USER ABEND 1099. The bad

record can be deleted by changing the test for dictionary

record deletion to include this garbage record:

IF TMMDREC='DD' OR (TMMDPROD='00'X AND TMMDREC='0081'X)

THEN DELETE;

Thanks to Marcia Ketchersid, Price Waterhouse LLP, USA.
Change 15.138 Report Performance Groups or Reporting Classes now CAN be

IMACWORK used in member IMACWORK to define workloads in dataset

RMFINTRV PDB.RMFINTRV. This redesign separately sums the Control

Jun 22, 1997 Performance Groups (Compat Mode) or the Service Classes

(Goal Mode) into the existing variable CPUTM, and then

sums all of the CPU times from your workload definitions

into new variable CPU72TM. If CPU72TM does not exactly

match CPUTM, then either you have double accounting

(CPU72TM GT CPUTM), or you have failed to map all of your

workloads in your IMACWORK logic (CPUTM GT CPU72TM); MXG

will print an error message on the SAS log if the times

do not match. The new design relocated the DELETE logic

from member RMFINTRV into the revised IMACWORK (since you

will have to change that delete logic if you want to use

RPGNs or Reporting Classes in your workload definitions),

but you do not have to change your IMACWORK member until

you want to use the new facility. The new RMFINTRV will

detect that you are still using an "old" IMACWORK member,

and will continue to delete Report Performance Groups and

Reporting Classes until it detects you are using the new

IMACWORK member. Thus this change is compatibly made!

The optional code to directly read SMF records that is

commented out in RMFINTRV was updated to match BUILDPDB.
Change 15.137 Pre-15.03-QA job found missing labels for variables in

VMACSTRS these members.

TRNDDB2S

VMACTMDB

VMACLMS

TYPEIMS

ASUMDB2A

ASUMDB2B

Jun 22, 1997
Change 15.136 This member for reading Web Proxy logs in Common Log

TYPEWWW Format was revised to conform to MXG style (LABELs,

Jun 22, 1997 FORMATS, LENGTH, and name changes of temporary fields,

and some statements were consolidated. Variable RDRTS

was renamed WWWRDRTS to avoid name conflict.
Change 15.135 Trending for Workload Manager Goal Mode Service Class

TRND72GO dataset TYPE72GO into TREND.TRND72GO now exists, in

Jun 20, 1997 this preliminary version. A revision that will map your

Jun 25, 1997 old TRND72 Performance Group data into the new TRND72GO

is planned so you can track before/after Goal Mode.

Jun 25 note: Cosmetic revisions (LABELs, DROPs) added

after first (Jun 24) MXG 15.03 was created.

Thanks to Mike Billy, Banc One, USA.

Change 15.134 Support for IXFP (Iceberg/IBM RVA) new subtype 6 and 7

EXICEDDS creates three new datasets:

EXICESNP Subtype Dataset Description

EXICEUTL 6 ICEBRGSN SNAPSHOT - EXTENTS SNAPPED

IMACICE 6 ICEBRGDD SNAPSHOT - DDSR EXTENTS

VMACICE 7 ICEBRGUT Space Utilization

Jun 19, 1997 Note that APAR OW25126 corrects errors in subtype 7 in

Mar 30, 1998 sites when Snapshot microcode is installed on an RVA

(RAMAC Virtual Array) subsystem. Put its PTF on!

===>>> Text updated March 30, 1998. This change created an

INCOMPATIBILITY in dataset ICEBRGDE. Before this change,

variable ICEVERS was a numeric and stored in 4 bytes, but

this change (in MXG 15.04) changed ICEVERS to character

of length one in dataset ICEBRGDE, and then created new

new variable ICEVERSN as a numeric stored in 4 bytes in

datasets ICEBRGSN and ICEBRGDD. In retrospect, exactly

why I did it this way is unclear, but I suggest for now,

simply drop the variable ICEVERS from ICEBRGDE, since it

really is not needed to be kept (it's a record version,

but I think there's only been one and it's always a 1?).

Edit member IMACICE and replace the MACRO _KICEDEL %

null definition with

MACRO _KICEDEL DROP=ICEVERS %

to drop the variable from dataset ICEBRGDE to avoid the

conflict. I think ITSV and SAS/CPE can tolerate the

absence of ICEVERS better than its conflicting presence.

This is a preliminary answer, and will be updated if it

does not resolve the conflict.

Thanks to Dan Almagro, Automobile Club of Southern California, USA.

Thanks to Ken Williams, Sun Life of Canada, ENGLAND.

Change 15.133 In sysplex, timestamps converted from "GMT" to "LOCAL"

VMAC30 were off by the number of "leap seconds", currently 20.

VMAC110 These "GMT" time values are actually "Absolute" values

VMACDB2 that include leap seconds; MXG erroneously "floored" the

Jun 18, 1997 delta value to get only the GMT offset hours to convert

"GMT" to "Local". Now, MXG keeps the full delta between

"Absolute" and "Local" to properly convert to local; if

the GMT offset value itself is not exact, it shows the

number of leap seconds in your system (CVTLOS).

Records GMT Offset Comment

30's SMF30IST-INTBTIME Calculated, leap seconds

42 st 6 S42JDGMO Provided, leap seconds

72 GMTOFFTM Provided, leap seconds

100-102 SMFTIME-THISSTCK Calculated, leap seconds

110 MCTMNTAD Provided, leap seconds

Unless you were comparing events with your wrist watch,

you might not have noticed this 20 second error in the

timestamps, but see the discussion in MVS Technical Notes

in Newsletter THIRTY-TWO for synchronization impact.
In addition, this change now converts SYNCTIME to local

time in type 30; it was already converted from Absolute

to local in RMF datasets, so now it is consistent!
Change 15.132 MXG 15.01 or MXG 15.02 cosmetic errors. TESTOTHR brings

VMXGHSM TYPEWWW in, but the FILENAME statement causes dataset not

TYPEWWW found under MVS, so it was commented out. Uninitialized

Jun 18, 1997 variable MCTFLG3 message was corrected by spelling it as

MCTFLGS in the LABEL statement.

Thanks to James McCullough, Economical Mutual Insurance Co., USA.

Change 15.132A DB2 Trace dataset T102S106 was not correct for new fields

VMAC102 added to QWP4 section in DB2 3.1, 4.1, and 5.1. The new

Jun 17, 1997 fields include MXG-constructed QWSU_SEC Service Unit per

CPU-Second value of the machine on which this DB2 Subsys

is executing!

Thanks to Jon Fritz, Levi Strauss & Co., USA.

Change 15.130 Variable SMF94ETO existed in MXG 14.14 but did not exist

VMAC94 with MXG 15.01 or 15.02, while variable SMF94EPM had the

Jun 17, 1997 TOTAL*EJECTS but was labeled POST*EJECTS. The real IBM

documentation now shows that only SMF94ETO should exist

and is the TOTAL*EJECTS, so MXG variables are no longer

created.

Thanks to Robert Miles Standish, Dean Witter Trust Company, USA.
Change 15.129 Label for DB2 QBSTDMC and QBnTDMC variables is now

VMACDB2 QBSTDMC='DATA MANAGER*THRESHOLD*REACHED'.

Jun 17, 1997

Thanks to Joseph Fordham, UPS, USA.

Change 15.128 Dataset TYPE116 MQ Series variable QWHCTNO is now labeled

VMAC116 QWHCTNO='CICS*THREAD*IDENTIFICATION*ADDRESS' instead of

Jun 17, 1997 Thread Number, as IBM now explains it is a control block

address of the thread, so '004dF3D4'x is a valid value.

The address can be reused, so it is not a unique value.

Thanks to Helmut Roese, COM Gmbh, GERMANY.

Change 15.127 AS/400 variable AS400SYN is missing if length of SYSTEM

VMACQAPM is less than 8 characters. The original logic is now:

Jun 16, 1997 ELSE IF GKEY=' S' THEN DO;

LENSYS=MIN(LENGTH-6,8);

INPUT @7 GDESS $VARYING8. LENSYS @;

AS400SYN=GDESS;

CALL SYMPUT('AS400SY',AS400SYN);

END;

Thanks to Len Marchant, Coca Cola Enterprises, USA.
Change 15.126 Counting the Average and Maximum Logged On TSO Users each

ANALCNCR 15 minutes (using PDB.JOBS TSO Session records) is done

Jun 15, 1997 in a new example in the comments in member %ANALCNCR.

Thanks to Dan Gilbert, Bergen Brunswig, USA.

Change 15.125 The detection of VIO was based solely on DEVNR='7FFF'x

VMACUCB but as this could now be a legitimate device address with

Jun 15, 1997 4-digit UCBs, the test for VIO in VMACUCB was changed to

require DEVx also. For type 30 DD segments,

the DEVCLASS for VIO is zero, but in type 14/15 records,

the DEVCLASS is whatever device was installation defined

for VIO space calculations. Fortunately, there is a bit

set in type 14/15 records for VIO, which MXG uses to set

DEVICE='VIO' even though DEVCLASS is non-zero.
Change 15.124 Support for APAR OW25263 adds variables JFCTDSI1 (bits

VMAC1415 for 128 track mode, "1/2 inch/320 meter particle media",

Jun 15, 1997 and JFCTDSI2 (bits for Compaction=YES) for 3590's.
Change 15.123 Comments were revised to describe IMACKEEP as archaic;

IMACKEEP its function was replaced by the MXG Exit Facility that

Jun 15, 1997 is described in Change 10.175.

Thanks to Thierry Van Moer, Comparex Information Systems, BELGIUM.

Change 15.122 Label for variable R744SCN should have been spelled

VMAC74 CONTENTIONS.

Jun 15, 1997

Thanks to Bob Gauthier, American Stores Company, USA.

Change 15.121 DB2 Trace IFCID=125. Variables QW0125NR and QW0125RI are

FORMATS now INPUT as "&IB.4." instead of "&PIB.4." and formats

VMAC102 MGD125N and MGD125R decode their negative values:

Jun 15, 1997 VALUE MGD125N

-2 = '-2:NO STG OR MAX EXCEEDED'

;

VALUE MGD125R

-1 = '-1:RETRIEVE NOT ATTEMPTED'

-2 = '-2:NO STG AVAILABLE'

-3 = '-3:NUM RIDS EXCEEDED MAX LIMIT'

;

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

Thanks to Harmuth Beckmann, Karstat AG, GERMANY.
Change 15.120 Change 14.300 (which only affected old RMDS 1.3 and 1.4)

VMACRMDS was not made as documented, causing variables RMDSARN and

Jun 12, 1997 RMDSARI to be missing; the change is now made!

Thanks to Steve Colio, CIGNA Corp, USA.

Change 15.119 You cannot use Tandem's ftp program to upload Measure's

ADOCTAND "Structured" data files because Tandem's ftp sends the

UTANDSTR entire file including header, data records, and trailer.

Jun 17, 1997 Products like BCOM and Connect Direct do understand

Tandem's structured files and will send only the data

records, so they don't have this problem!

If you are stuck with Tandem's ftp, you will have to

convert the Structured file into an Unstructured file

and then you can use their ftp to upload that data.

Member ADOCTAND has been updated with a Tandem-supplied

example that will create an unstructured file from the

structured MEASCOM data files using Tandem utilities.

In addition, although I hope you never need it, member

UTANDSTR will create Unstructured Measure files from

Structured ones that have already been shipped to MVS.

You will have to supply the LRECL of each file in the

program, and will have to EDIT the dataset names, and

you must use RECFM=FB and LRECL=4096 when you upload

their structured file with their ftp.

If Tandem ever changes their ftp program to recognize

their own records, the double processing can go away.

Thanks to Tim Crocker, AT&T UCS, USA.

Thanks to Jack Driscoll, Tandem, USA.
Change 15.118 TYPE94 record without APAR OW27369 caused INVALID AUDIT

VMAC94 SECTION error; the test IF SMF94SDL GE 76 THEN INPUT must

Jun 6, 1997 be changed to IF SMF94SDL GE 84 THEN INPUT.

Thanks to Will Evans, Consolidated Freightways, USA.

Change 15.117 Changes in NPM that were overlooked long ago:

VMAC28 -LCSLxxxx variables added to CSL segment for HPR bytes and

Jun 4, 1997 frames transmitted, received, and discarded, and LCSLCON5

thru LCSLCON7 flag bytes are now kept.

Change 15.116 TLMS 5.3 (archaic, 5.4 has been out for some time) dates

VMACTLMS were not converted completely; the conversion code from

Jun 4, 1997 the 5.4 code was moved to the 5.3 section of VMACTLMS

Thanks to Jim Van Arsdale, IBM Global Services, USA

Change 15.115 MONTHBLD failed for dataset TYPE77, because WEEKBLDT did

WEEKBLDT not have the correct BY list (but WEEKBLD was correct!).

Jun 4, 1997 WEEKBLDT now matches BUILDPDB/WEEKBLD/MONTHBLD.

Thanks to Mike Rouncville, Rose's Stores, Inc, USA

Change 15.114 TMON/DB2. Previously un-tested subtypes "DE" and "DW" had

VMACTMDB coding errors that surfaced when real data was read.