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

Dataset OPC24D_B has new variables MTDRESE and MTDRESQ.

Dataset OPC24_25 formerly contained subtypes 2,3,4 and 5

(MT0TYPE) of the Type 24 record, but this change creates

new dataset OPC24_2 for the subtype 2, and now OPC24_25

contains only subtypes 3,4, and 5. The new OPC24_2

dataset exit name is EXOPC24Y because EXOPC242 is already

in use.

Thanks to Randy Shumate, LEXIS-NEXIS, USA.

Change 14.076 Variable MSOUNITS is now stored as 5 bytes instead of 4,

VMAC30 because a value such as 2,804,486,170 is truncated to

Apr 11, 1996 only 2,814,486,016 when stored in 4 bytes.

Thanks to Aldo Valenti, SAS Italy, ITALY.

Change 14.075 If KEEPALL=YES was specified, and some of the parameters

VMXGSUM were not used in your %VMXGSUM invocation, the logic to

Apr 10, 1996 reconstruct the strings failed since it should never have

been executed. This change makes KEEPALL=YES work as it

was intended, and provides improved error checking.

Thanks to Tom Elbert, John Alden Insurance, USA.

Change 14.074 If TRNDLPAR has less than one week's data (i.e., first

GRAFLPAR time it's being used), strange graphs occurred because

Apr 10, 1996 MXG assumed more than one week's data. This change uses

the "best fit" algorithm of GRAFTRND so that all points

represent a monday or a start of the week.
Change 14.073 ERROR: VARIABLE QWHSIID NOT FOUND with DB2 I/O reports

ANALDB2R (a holdover from before ANALDBTR existed to pair the DB2

Apr 10, 1996 I/O trace records) is corrected by this change.

Thanks to Roland Nieuwenhuizen, SAS Netherlands, THE NETHERLANDS.

Thanks to Paul Ritzen, KLM, THE NETHERLANDS.
Change 14.072 Variable CLASRC was added to EREPOBR dataset (to identify

ADOCEREP which record created the observation), and ADOCEREP was

VMACEREP updated to describe datasets EREPOBL and EREPHDR.

Apr 10, 1996

Change 14.071 Dataset DB2STATB previously had zero observations unless

EXDB2STB you tailored exit member EXDB2STB, because I thought the

Apr 10, 1996 dataset could be large. In fact, there is only one obs

for each buffer pool for each interval, and thus I have

removed the comment block around the OUTPUT _LDB2STB;

statement, so observations will now always exist.

Thanks to Chuck Hopf, MBNA, USA.
Change 14.070 IBM APAR OW19251 addresses RACF and the year 2000; since

VMAC80A the RACF database only has three bytes for date, IBM has

Apr 10, 1996 declared that RACF dates with YY of 71 or higher will be

interpreted as year 19YY and dates with YY of 70 or less

will be interpreted as year 20YY. Further information is

to be in APAR OW19683. In examining VMAC80A, two dates

were not converted from Julian to SAS. After the INPUT

of variables REVOKDTE and RESUMDTE (two separate places),

insert:

IF REVOKDTE GT 0 THEN REVOKDTE=DATEJUL(REVOKDTE);

IF RESUMDTE GT 0 THEN RESUMDTE=DATEJUL(RESUMDTE);
Change 14.069 TYPE99_2 now contains one observation for each period in

VMAC99 each service class that had activity; previously only the

Apr 10, 1996 first period's data was output.

Before the INPUT @SM992CPO insert DO _I_=1 TO SM992CPN;

After the %%INCLUDE SOURCLIB(EXTY99U2); insert

SM992CPO=SM992CPO+SM992CPL; END;

I discovered that the SRVCLASS variable that I INPUT

from the Period Data Section must instead be INPUT from

the Class Data Section (i.e., where SM992NAM was input),

and I now INPUT new variable SM99PCNM where SRVCLASS was

INPUT, labeled SM99PCNM='INTERNAL*SERVICE*CLASS*NAME' and

is kept instead of SM992NAM. The Internal Service Class

Name only appears in type 99 records, with names like

$SRMDInn for discretionary periods, or $SRMBEST, $SRMGOOD

and $SRMDUMP names.

Thanks to Sridhar Gopalaswami, Barnett Technologies, USA.

Change 14.068 MXG only output the first QLST segment in DB2STATS, but

DIFFDB2 there is one segment for each Remote Destination used in

EXDB2STR each interval. Now, the QLSTxxxx variables in DB2STATS

IMACDB2 will contain the sum of all remote activity to all remote

VMACDB2 locations during the interval, while new dataset DB2STATR

Apr 9, 1996 contains one observation for each Remote Location during

each interval with the detail QLSTxxxx counts. Member

DIFFDB2 was altered to also deaccumulate DB2STATR.

Thanks to Peter Li, John Deere, USA.

Thanks to Ralph Baechle, John Deere, USA.

Change 14.067 Variable QBSTHBE in DB2STATB and variables QBnTVPL,

DIFFDB2 QBnTHPL, and QBnTHBE for n=0,1,2,3 in DB2STATS are not

Apr 2, 1996 not accumulated values but are endpoint values, so their

DIF() statements were removed from DIFFDB2. (QBSTHPL and

QBSTVPL were removed from DB2STATB by Change 13.269).

Thanks to Peter Li, John Deere, USA.

Change 14.066 Validation of TYPE88 with data corrected INPUT STATEMENT

VMAC88 EXCEEDED RECORD LENGTH errors Input of SMF88SDL should

Apr 2, 1996 have been &PIB.4. vice &PIB.2., and the +2 after SMF88SAB

must be +4, SMF88RVN is now input as &NUM.2 vice &PIB.2.,

and SMF88LIT is input TODSTAMP8 and formatted, and new

field SMF88EFS (undocumented in SMF manual) is input.

Thanks to Lee Borgman, Boeing Information Services, USA.
Change 14.065 APARS OW08565 and OW10584 adds SYNCTIME and statistics

VMAC28 (CPU utilization, Fast Program Storage Utilization, and

Apr 1, 1996 Buffer Storage Utilization) for each of the (up to 18!)

processors in the 3746/900 in the NAC segment, and adds

TIC utilization per frame and per byte sent and received

in the CSL segment.

Thanks to John Astle, National Australia Bank, AUSTRALIA.

Thanks to Sandro Aiello, National Australian Bank, AUSTRALIA.

Change 14.064 Using Tape instead of DASD for ANALDSET fails; the step

ANALDSET added in MXG 13.02 reused the same DDNAME. The lines

Apr 1, 1996 DATA SORTSTEP.SORTSTEP; SET SORTSTEP.SORTSTEP must be

DATA TYPE30.TYPE30_4; SET SORTSTEP.SORTSTEP; and a new

data step was added after that step with

DATA SORTSTEP.SORTSTEP; SET TYPE30.TYPE30_4;

Thanks to Bill Bechtloff, Acxion, USA.
Change 14.063 DASDMPL value 1000 times too large in dataset TYPE42DS.

VMAC42 RESPTIME is in millisec, not sec, so equation must be:

Apr 1, 1996 DASDMPL=RESPTIME*DASDRATE/1000;

Thanks to Xiaobo Zhang, Insurance Services Office, USA.

Change 14.062 Just before the "DATA PDB.ASUMAPAF;" statement, insert:

ASUMAPAF PROC SORT DATA=TEMPAPA2;

Apr 1, 1996 BY SYSTEM STARTIME LPARNUM LPARNAME;

to protect sort order if you increase the number of

MDF domains.

Thanks to Scott Snyder, Card Establishment Services, USA.

Change 14.061 Variables MVUCDATE and MVUCTIME need ?? modifier (just

VMACEDGS like change 13.124) to prevent INVALID DATA messages when

Apr 1, 1996 there was no last user change date. In addition, dataset

EDGSOREC and EDGSOVOL were not correct. After the +7,

insert @; IF MOCFLG EQ '08'X THEN INPUT before MOOWNSUR,

replace IF LENGTH-COL+1 GE 4 THEN INPUT with ELSE INPUT

and change IF MOVOLNO GT 0 THEN DO to IF MOVOLNO GT 0

AND MOCFLG NE '08'X THEN DO....

Thanks to Emmanuel Chenay, Toyota Motor Europe, BELGIUM.
Change 14.060 UCICSCNT CICS diagnostic tool was enhanced to count the

IMACEXCL different record sizes found in your CICS transaction

VMAC110 (type 110 subtype 1) data that you can verify if all of

UCICSCNT your regions have the same EXCLUDE/INCLUDE definitions.

Apr 1, 1996 Comments in IMACEXCL were improved for multi-line INPUT

statements for the same CICS field number, and debugging

logic in VMAC110 was strengthened.
Change 14.059 TYPE72GO variable VALDSAMP was incorrect, which caused

VMAC7072 the PCTDLxxx variables to sometimes be wrong. The correct

Mar 29, 1996 equation should have five terms:

VALDSAMP=PCTUSCUS+PCTDLTOT+PCTDLUNK+PCTDLIDL+PCTDLPQU;

In addition, VALDSAMP is added to the KEEP= list for the

TYPE72GO dataset.

Thanks to Bruce Widlund, Merrill Consultants, USA

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

Change 14.058 TYPE71 statistics for Shared Page Groups added in ESA 5.1

VMAC71 by IBM but missed by MXG until now. New variables are:

Mar 29, 1996 SHPG-IN/OU-AU/EX - Page in/out from AUX and ESTORE

SHPGSYxx - Shared Page Groups in the system

SHPGCSxx - Shared Page Groups in CSTORE

SHPGESxx - Shared Page Groups in ESTORE

SHPGAUxx - Shared Page Groups in AUX STORE

SHPGFXxx - Shared Page Groups Fixed

SHPGLOxx - Shared Page Groups Fixed below 16MB

where xx = AV,MN,MX for average, min and maximum.

Thanks to Don Deese, (CPExpert), Computer Management Sciences, USA.
Change 14.057 Support for STK's NearOAM (formarly NearImage) user SMF

EXTYNOAM record creates dataset TYPENOAM with one observation for

IMACNOAM each individual request to retrieve an object. Their SMF

TYPENOAM record can contain up to 100 requests from a single tape;

VMACNOAM MXG creates one observation per request, with the start

Mar 28, 1996 and end times and duration of each request, and the name

of the tape dataset being processed.

Thanks to Fiona Crane, Sun Alliance Insurance Group, ENGLAND.

Change 14.056 Cosmetic. Test macro _TESTINTX defined in VMACVMXA is

JCLTEST6 too long and is renamed _TESTINX to conform with 8-byte

VMACVMXA limit of SAS. There was no change to JCLTEST6, but that

Mar 27, 1996 MXG job encountered this error in VMACVMXA.

Thanks to Philip Sills, IFF International Flavors & Fragrance, USA.
Change 14.055 STK's HSC Subtype 08 record comes in two (incompatible)

VMACSTC lengths of 38 or 40 bytes, but MXG Change 14.001 knew of

Mar 27, 1996 only the 38 byte record. STK expanded a one byte thrice

defined field into three separate fields. Either length

record caused INPUT STATEMENT EXCEEDED error message.

To fix, conditionally execute the INPUT statement by

inserting IF LENGTH-COL+1 EQ 18 THEN before input,

then block replicate the input, change the condition to

ELSE IF LENGTH-COL+1 GE 20 THEN

and then remove the +M1 before fields STC08CID, STC08MAG

in the GE 20 input.

Thanks to Herb G. Strozinsky, Burlington Northern Santa Fe RR, USA.

Thanks to Sudie Wulfert-Schickedanz General American Life Ins, USA.
Change 14.054 Support for Netview FTP (File Transfer Protocol) subtype

FORMATS 51x, written when the Server Finished Request Processing

VMACFTP creates FTP25X dataset with bytes transferred, CPU time,

Mar 27, 1996 begin and end, as well as compression factor.

APAR PN82953 corrects documentation.

Thanks to Herb G. Strozinsky, Burlington Northern Santa Fe RR, USA.

Change 14.053 Variables LUNETID PCSESSID VILUNAME in NSPYLU dataset are

VMACNSPY trashed, because two new fields, LUATTCHS and LUUSRSPS

Mar 27, 1996 were not input by MXG. After the line that INPUTs

Mar 28, 1996 variable LUNRSPSS, and before the @; that ends that

INPUT statement, insert

LUATTCHS &PIB.2. /*ATTACHES*INITIATED*ONLY LU 6.2*/

LUUSRSPS &PIB.2. /*USER*RESPONSES*(BOTH6.2+NON-6.2)*/

then add the two variable names at the end of the NSPYLU

KEEP= list, just before the _KNSPYLU macro reference.

Additionally, variable NSPYNETW was trashed, as six bytes

were inserted in July 95 that were not in the March 95

documentation for Netspy 4.6. To circumvent, after the

line that INPUTs SGTRGT4, insert a line with +6 to

skip six bytes. The actual change creates new variable

NSPYDBKY, Database Key from the Netspy Header, but does

not keep the new variable (until someone tells me what

use it is in which Netspy datasets!).

Thanks to Dennis Longnecker, State of Washington OAC, USA.

Change 14.052 HSM ABARS short FSRTYPE=15 record caused INPUT STATEMENT

VMACHSM EXCEEDED RECORD error. While I am still investigating

Mar 15, 1996 what changed, to circumvent the error and skip the bad

record, insert IF LENGTH-COL+1 LT 274 THEN RETURN;

in two places, immediately preceding both occurrences of

INPUT @15+OFFSMF

WFSRJOB $EBCDIC8. /*REQUESTING*JOB*...

Thanks to Robert A. Burns, The Travelers, USA.

Change 14.051 Variable ELAPSTM was not created in dataset TYPE72GO Goal

VMAC7072 Mode, causing RMFINTRV to have missing values for all of

RMFINTRV the response (TSORESP,BATRESP) when you migrate to WLM

Mar 15, 1996 Goal Mode, because ELAPSTM is the numerator of response.

Just adding ELAPSTM to KEEP list for TYPE72GO won't work.

Thanks to Waldemar Schneider, SAS Institute Cary, USA.

Thanks to Kevin Carpenter, First of America Services, USA.
Change 14.050 INVALID DATA FOR BETASTRT and BETAEND with BETA 1.6.5 for

VMACBETA subtype=21 because MXG failed to skip over three reserved

Mar 13, 1996 bytes. Find the statement ELSE IF SUBTYPE=21 THEN DO;

Then find the statement about 21 lines later reading:

BETASTRT SMFSTAMP8. /*S20TRES1*RESERVED*/

Change the line to read

+3 BETASTRT SMFSTAMP8. /*S20TRES1*RESERVED*/

Thanks to Mr. Hasselmann, Braun AG, GERMANY.

Change 14.049 Changes to keep DATETIME in Change 14.028 should also be

BUILDPD3 made in BUILDPD3 and BUIL3005 (for JES3) to be consistent

BUIL3005 with BUILDPDB and BUILD005 (for JES2). Variable DATETIME

Mar 13, 1996 is kept and labelled in PDB.JOBS, PDB.STEPS and PDB.PRINT

and LENGTH statements were also made consistent.

Thanks to Chris Weston, SAS Institute Cary, USA.

Change 14.048 Variable ULOGADAE should have been INPUT as &PIB.4.3 vice

VMACCOMP &PIB.4.2.

Mar 13, 1996

Thanks to Jens Schlatter, Independent Consultant, GERMANY.

Change 14.047 DB2 Trace dataset T102S096 variable QW0096SN should have

VMAC102 been INPUT as &PIB.4. vice &PIB.2.; variables QW0096SC

Mar 13, 1996 and QW0096SK should be &PIB.2. vice $EBCDIC1.

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

Thanks to Harmuth Beckmann, Karstadt, AG, Germany
====Changes thru 14.046 were included in MXG 14.01 dated Mar 7, 1996===
Change 14.046 Validation of TMON for UNIX at a real site with real data

VMACTUX uncovered these corrections.

Mar 7, 1996 -Variable INTERVAL added to KEEP= for TUXCPU and TUXDISK.

-Variables UTIME, STIME, UTIMETOT, STIMETOT are divided

by 100.

-For data from AIX, memorymgmt fields swapins, swapouts,

filefault, swapfault, and faultiorate do not exist, and

system fields swapin_rate and swapout_rate do not exist,

and must be commented out in your input statement (until

I find a permanent solution!).

-Comparing the sum of utime and stime from process records

was typically 70%-80% of the usertm from cpu records for

HP G50, but with AIX, stime+utime is always greater than

the usertm, usually about 150%, and utime by itself with

AIX is 75-80% of usertm.

-Nowhere is TMON/UNIX is CPU model/speed/size or memory

size kept.

Thanks to Dan Sidorik, SmithKlein Beecham, USA.

Change 14.045 The //LKED.SYSLMOD DD statement must be moved to the end

ASMRMFV of the member for the assembly and link edit to execute

Mar 6, 1996 without error.

Thanks to Jay Beeler, Key Services, USA.

Change 14.044 A truncated DB2 3.1 type 101 SMF record caused INPUT

VMACDB2 STATEMENT EXCEEDED because MXG validation of the triplet

Mar 6, 1996 (offset, length, number) was not as robust as it is now.

Change the test now reading:

IF 0 LT OFFPROD LT LENGTH AND NRPROD GT 0 THEN DO;

to read

IF 0 LT OFFPROD LT LENGTH AND NRPROD GT 0

AND OFFPROD+LENPROD-1 LE LENGTH THEN DO;

The record may have been truncated during copying after

creation, or DB2 might have created the bad record; the

site is still investigating, and this note will be

updated when more is known. Now, MXG will detect this

class of bad records, tell you they are there while

avoiding the STOPOVER error, and keep on truckin'.

The actual change was made to all forty statements of the

preceding form, one for each record segment type.

The bad record had a product section length (in the

triplet) of 152 bytes, but there were only 114 bytes

left in the record from the start of the PROD data

section. The PROD data section contained a valid

76-byte header 1 subsection, and the beginning of a

header 2 section with expected length of 76 bytes, but

only 38 bytes exist.

Thanks to Dan Null, Caterpiller Inc, USA.

Change 14.043 DCOLLECT dataset DCOLSG (Storage Group definitions) has

VMACDCOL variable DSGGBKUF too large, because MXG was off by one

Mar 6, 1996 byte. After the IF DCOLRTYP='DC' THEN DO; statement,

insert new statement M8=-8; Then change the line that

inputs DSGGBKUP so that it reads:

+M8 DSGGBKUP $EBCDIC1. +7 /* label ... */

(DSGGBKUP is a redefinition of DSGCNFRM).

Thanks to Mike Moss, Royal Bank of Scotland, SCOTLAND.

Change 14.042 -INVALID DATA for TIGETMCT or TIFREMCT because the fields

TYPETMON are documented as XL8 and I assumed they were floating

Mar 6, 1996 point (like all the other Landmark XL8 fields) and input

them as &RB.8. Since all of my test data had zero values

I could not tell. Well, it turns out they are fixed and

not float, and their input must be changed from &RB.8. to

&PIB.8., and the two lines after the @; that multiply by

1000000 must be deleted.

-INVALID DATA FOR TIAPREQ TIAPWAT and/or TIAPEXMT result

when your CICS guy uses new TMON TCE 1.3 programs

TMONCNVT or $DBUTIL to convert old TMON TCM 8.3 records;

trashed records of the wrong length (2276) are created by

the Landmark program when you cross versions. Although I

think the Landmark program should have recognized their

inconsistency and ABEND their convert, they don't, so MXG

will now test for wrong length and warn you on the log

that bad records have been found and deleted.

Thanks to Rudra Rishy Maharaj, Stelco Inc, CANADA.

Change 14.041 -Variables CPUTYPE and CPUVERSN are now added to TYPE72GO

EXTY72GO observations. Retained from the last type 70 record, the

VMAC7072 variables were kept in TYPE72 PERFGRP data, so it makes

Mar 5, 1996 sense to keep them also in the TYPE72GO SRVCLASS data, so

Mar 27, 1996 that old reports can be used for WLM reporting.

Apr 23, 1996 -Variable R723RTYP was added to TYPE72GO KEEP= list.

-Change 13.228 was not actually implemented in MXG 13.13,

but its logic (in EXTY72GO to OUTPUT TYPE72GO only if the

service class or report class had activity during the

interval) was finally added in MXG 14.01. Subsequently,

its test for R723TYPE IN(1,2,4) was changed in MXG 14.02

to test for R723TYPE IN(1,2,4,5), because report classes

were being output with all zeroes for report classes

whose associated service classes were never initiated

(e.g., CICS report classes defined on a CICS backup).

-Because R723CRCA is frequently blank even when resources

are recorded for a Service Class, and because it is used

to set R723RTYP values 4 or 5, the logic to set the value

of R723CRCA was changed to now read:

IF R723CADF='1.......'B OR SERVICE+ACTIVETM GT 0 THEN

R723CRCA='Y';

Thanks to Brenda Rabinowitz, Prudential-Bache Securities, USA.

Change 14.040 TCP/IP release 2.2.1 prior to APAR PN40511 was a user SMF

VMACTCP record with ID greater than 128, but if you used ID=118

Mar 5, 1996 for this old release, you caused MXG to print UNEXPECTED

TCP/IP DATA message notifying you that MXG had deleted a

record. However, if the bad record happened to have "S"

in column 69, MXG failed with INPUT STATEMENT EXCEEDED.

Correcting SMF TYPE eliminates the error (and with that

APAR and later releases of TCP/IP, you no longer can set

the ID, it is fixed at 118), but I added protection to

avoid the STOPOVER condition.

Thanks to Jim Majors, Barnett Technologies, USA.
Change 14.039 Altai & CA job schedulers can't restart a multi-step SAS

MXGSAS job due to the DISP=(MOD,PASS) on the //NULLPDS DD

Mar 4, 1996 statement in the JCL procedure to execute SAS, because

you cannot restart a step that needs passed datasets.

However, if you could change PASS to DELETE, the step

would be restartable, and since NULLPDS exists only to

satisfy JCL syntax for the LOAD and SASAUTO symbolic

parameters, and since NULLPDS is never actually opened,

it would be safe to change the PASS to DELETE. However,

when PASS was changed to DELETE, now a second execution

of the PROC fails with ABEND 213-04, and the matter is

still under investigation. Stay tuned.

Thanks to Neil Ervin, Huntington Services, USA.
Change 14.038 Consistency clean-up. Datasets DCOLAI and DCOLCN did not

VMACDCOL have variable ZDATE in their KEEP= list, but now they do.

Mar 3, 1996
Change 14.037 MAINTLEV 8 of the MXG Tape Mount and Allocation Monitor.

ASMTAPES Two principal changes: Use of the CVT to load the return

ZASMTAPE address of the SRB routine, so we will no longer be

Mar 3, 1996 dependent on the SRB DSECT to properly exit the SRB, and

far more stringent use of the lockword; the SRB will now

exit immediately if it is unable to obtain the lockword,

whereas prior versions assumed we were serialized and

continued to process. MAINTLEV 7 has failed only once at

only one site (but it was in its SRB routine, while in

FRR), and it may have caused another ASID to fail! Dumps

of that failure did show the exposure in our SRB routine

that are now corrected.

Thanks to Paul Hill, Midland Bank, ENGLAND.
Change 14.036 Support for type 6 ESS segment decoding is added and is

IMAC6ESS externalized in new IMAC6ESS. By default, no fields in

Mar 3, 1996 the ESS segments will be decoded until you remove the

comment block in this member. You will also want to set

the desired LENGTH of the new variables to be kept, and

if you want the ESS variables in your PDB.PRINT, you will

also need to update _PDB_6 macro in IMACPDB.

Thanks to Martin Wieland, Neckermann B.V., THE NETHERLANDS.

Change 14.035 Format MGD145S (for DB2 trace IFCIDS 124, 145, and 183)

FORMATS should map binary, not hex values, so the X's on both

Mar 3, 1996 sides of the equal sign were removed.

Thanks to Joseph J. Faska, Depository Trust, USA.

Change 14.034 NDM subtypes 'SI' and 'ST' were not correct. In the 'SI'

VMACNDM section, insert +4 between INPUT and NDMUNODE, and

Mar 2, 1996 change $EBCDIC8. after NDMUID to $EBCDIC64. In the 'ST'

section, change $EBCDIC8. after NDMUID to $EBCDIC64.

Worse, I discovered NDMTIME and most other NDM datetime

variables were missing. Thirteen lines that now read