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

EXTMSTMS have CPGM (Creating Program Name) and SMSMC (SMS Mgmt

IMACTMS5 Class Name). Dataset TMS.TMS has new variables LPGM

VMACTMS5 (Last Used Program Name), ROBTY/ROBID Robotic flags,and

Jun 13, 1995 B1INT/B1DIS (B1 Security Integrity/Disclosure labels).

In addition, this revision added the "_L, _K" tailoring

macros in member IMACTMS5 (See Change 10.175), and the

EXTMSxxx Exit Members so TMS is now consistent with other

MXG processing members. These additions should be

transparent, and are needed only if you want to tailor

the contents of the TMS datasets.

Thanks to Lonnie Melamed, SHL, CANADA.
Change 13.082 DB2 3.1 only. No Statistics Reports, DB2STATS has zero

DIFFDB2 observations. The MXG logic demanded there be data in

Jun 13, 1995 DB2STAT0,DB2STAT1, and DB2STAT2 to OUTPUT to DB2STATS,

but DB2STAT2 only has observations if your DB2 uses the

new Hiperpools! The logic was changed from:

IF (QWHSRELN LT 2.9 AND IN0 AND IN1)

OR (IN0 AND IN1 AND IN2);

to:

IF IN0 AND IN1;

so that DB2STATS is created whether or not there are obs

in DB2STAT2.
I'm awaiting doc of a case where non-zero obs in DB2STATS

becomes zero obs in DB2TEMP2.

Thanks to Siegfried Trantes, IDG Informationsvararbeitung, GERMANY.

Change 13.081 ANALBLSR uses the EXCPDASD variable created by ANALDSET,

ANALDSET but ANALDSET did not sum the MULTIDD='Y' observations in

Jun 13, 1995 the STEPS dataset, so some calculations in ANALBLSR were

greater than 100%! A new data step was added to ANALDSET

to sum EXCPDASD, and also to output only one observation

per step. Insert after PROC SORT DATA=TYPE30.TYPE30_4...

DATA SORTSTEP.SORTSTEP;

SET SORTSTEP.SORTSTEP;

BY SYSTEM READTIME JOB SORTTIME;

DROP TOT_EXCP;

IF FIRST.JOB THEN TOT_EXCP=0;

TOT_EXCP+EXCPDASD;

IF LAST.SORTTIME THEN DO;

EXCPDASD=TOT_EXCP;

TOT_EXCP=0;

OUTPUT;

RETURN;

END;

A future MXG enhancement will provide a separate optional

routine that can be invoked during BUILDPDB or TYPE30

processing to sum all variables in the MULTIDD='Y'

observations into a single observation (applies to the

TYPE30_V, TYPE30_4, TYPE30_5, and TYPE30_6 datasets).

Thanks to Jerry Layne, Virginia Power, USA.
Change 13.080 Support for Sterling SAMS Storage Automation & Reporting

EXSAMSES System's user SMF record creates three new data sets:

EXSAMSTO SAMSTOT - Total pools, volumes, space avail and used

EXSAMSVO SAMSVOL - Per volume space stats, pool(s) of volume

IMACSAMS SAMSESOT - Per pool space statistics.

TYPESAMS SAMS Version 3.2.0 documentation/data was validated.

VMACSAMS

Jun 12, 1995

Thanks to Shaheen Pervaiz, Acxiom, USA.
Change 13.079 -DB2 Statistics Summary PMSTA01 was wrong or not printed,

ANALDB2R especially if INTERVAL=SHIFT was specified; this error is

Jun 12, 1995 fixable as described below.

-Only the first 60 bytes of the SQL text was printed in

the Audit reports; too many lines were changed to print

the change here.

-The PMSTA01 error can be corrected with:

Find the %VMXGSUM(INVOKEBY=PMSTA01 - ANALDB2R) segment.

- Change the argument KEEPIN=MINTIME MAXTIME,

to read KEEPIN=MINTIME MAXTIME STRTTIME,

- After the argument SUMBY= change

OR &PMSTA01 NE YES %THEN BEGTIME; to read

OR &PMSTA01 NE YES %THEN STRTTIME;

-Find the first COMMITS = SUM(Q3STCOMM, ....

then change the previous DURATM=ENDTIME-BEGTIME;

to read DURATM=ENDTIME-STRTTIME;

Thanks to Joe Richard, MBNA, USA.

Thanks to Chuck Hopf, MBNA, USA.

Thanks to Mitch McKenna, SAS Institute Europe, GERMANY.

Thanks to Victor Chacon, NYNEX Mobile Communications Co., USA.

Change 13.078 Disregard this change. It was reported that Change

VMAC110 12.324 causes SAP 5.0.c to get INVALID DATA FOR HH MM,

Jun 12, 1995 but &PK fields can't have invalid values! The problem

turned out to be a user problem, and there was no change.

Change 13.077 SAP Variable STCTIMTR bad. Because of MXG Change 12.324,

IMACICSA MXG 12.12A and MXG 13.01 and later expect SAP 5.0 data,

Jun 12, 1995 while MXG 12.12 and earlier expect pre SAP 4.3F records.

(SAP 4.3F changed STCTIMTR, 4.3J split STCFLAG/STCTASK.)

You will have to EDIT IMACICSA and verify (or change) the

INPUT of STCTIMTR (Start time) to match your SAP release:

pre-4.3F: STCTIMTR PDTIME4.

4.3F,4.3J,5.0: STCTIMTR &PIB.4.3

MXG Change 12.324 changed the INPUT of STCTIMTR from

PDTIME4 to PIB4.3 in member IMACICSA, thinking it was an

error, but it was a change between SAP 4.3J and SAP 5.0.

Unfortunately, it does not appear there is any way to

detect which version of SAP wrote the record, so you must

manually update MXG to tell me what you have!

Thanks to Norbert Korsche, OMV, AUSTRIA.

Thanks to Ralf Menz, COMLAB, GERMANY.

Thanks to Jens Schlatter, EDP Consulting Schlatter, GERMANY.
Change 13.076 ANALALL selects and prints all SMF records from a job,

ANALALL but had not been recently updated for new job-related SMF

VMAC24 records. Member IMACJBCK, the "job checking" exit that

VMAC33 is used to select by JOB name, should have been included

VMAC36 by every job-related record processing, but had not been.

VMAC42 Several VMACs create multiple datasets from different

VMAC59 subtypes, but only some subtypes are job-related, so the

VMAC60 ANALALL logic now prevents creation of those unneeded

VMAC78 datasets. These SMF records processing VMACs were hit:

VMAC80 24,33,36,42,59,60,78,80,91,101.

VMAC91

VMACDB2

Jun 12, 1995
Change 13.075 Support for WYLBUR Accounting from ACS Wylbur. This SMF

EXTYWYLA record is similar to TYPEWYLB (WYLBUR Accounting from OBS

IMACWYLA Wylbur), but five variables are added, and fields are in

TYPEWYLA different locations, hence the need for these new members

VMACWYLA to create new dataset WYLBURA.

Jun 12, 1995

Thanks to Matt Anderson, Texas Tech University, USA.
Change 13.074 Further performance improvement. LENGTH DEFAULT=4; is

XMXGSUM now set in first DATA step, saving //WORK space.More

Jun 11, 1995 diagnostics added internally.

Thanks to Diane Eppestine, Southwestern Bell, USA.

Change 13.073 PDB.JOBS variables ABEND and CONDCODE cannot be used for

VMAC30 ABEND analysis; instead, PDB.STEPS must be used, because

Jun 9, 1995 a) programs abend, not jobs, and using PDB.STEPS lets you

identify which program was involved, so you can

separate inhouse-written programs from externally

acquired programs. Furthermore, counting ABENDs by

program will identify a trend that might be lost if

the same program abends many times in different jobs;

there may be only one ABEND per JOB so the program

would be overlooked if you counted by jobs.

b) multiple abends can occur in a job - which one do you

want in PDB.JOBS - the first, last, or which?, and

c) the type 30 subtype 5 record values are inconsistent:

variable CONDCODE is the value from the final step,

not the last step which failed, and

variable ABEND relates to the entire job, not a

specific step, based on bits in TERMIND1.

-However, an MXG logic error can cause type 30 subtype 5

to have ABEND='SYSTEM' for a job which had ABEND='USER'.

The correction is to replace

IF COMPCODE>=32768 THEN DO;

with these five lines:

IF TERMIND1='1.......'B THEN DO;

ABEND='SYSTEM';

CONDCODE=COMPCODE;

END;

ELSE IF COMPCODE>=32768 THEN DO;

Thanks to ???, Benetton, ITALY.

Change 13.072 APAR OW08641 for NPM Version 2.2 discusses support for

FORMATS LU 6.2 sessions acting in half-duplex flip-flop mode, and

VMAC28 adds new APPC values for Session Type in LSCDSTYP, so the

Jun 9, 1995 variable is now formatted by new format $MG028ST.

Note added Mar 27, 1998: The format $MG028ST was created

by this change, but the variable LSCDSTYP was not given

the $MG028ST format until Change 16.037!
Change 13.071 OS/400 Version 3.1 only, the DSARM and DSTYPE variables

VMACQAPM are reversed. The IBM documentation shows the order was

Jun 8, 1995 changed, so I coded to agree with the documentation, but

now real data proves that the documentation was wrong,

so now the code agrees with the data! Change the last

pair of INPUT's of DSARM and DSTYPE so that the order is

DSARM, then DSTYPE. There are three pairs of INPUTs, and

now, all will have DSARM before DSTYPE.

Also, two references to variable SPRES2 were removed.

Thanks to Clark Jennings, Reynolds Metal, USA.

Thanks to Freddie Arie, Lone Star Gas, TEXAS.
Change 13.070 Dataset NDMCT variable NDMDSDSN (Source Dataset Name)

VMACNDM was added to the KEEP= list, the second occurrence of

Jun 8, 1995 NDMDODSN in the LABEL statement was changed to NDMDSDSN,

and the first INPUT of NDMDODSN was changed to NDMDSDSN.

Thanks to Sue Yarker, Midland Bank PLC, ENGLAND.

Thanks to Natasja van Gavel, Alcatel, BELGIUM.

Thanks to Bryant Osborn, NationsBank Technology Services, USA.
Change 13.069 Cosmetic changes. Label for DB2END should be "LAST*DB2"

ANALDB2C (it was "LAST*CICS..."), and incomplete comments were

Jun 8, 1995 completed and clarified.

Thanks to Geert De Gand, Nationale Bank Van Belgie, BELGIUM.

Change 13.068 Change 12.196, when implemented, did not contain the

VMACNSPY ELSE ARSPNET=.; statement, but that line has been

Jun 8, 1995 added so the code agrees with the change.

Thanks to Alan Keeble, British Steel, UK.

Change 13.067 Backlevel SAS (TS404 for sure) causes error VFONF DOES

ANALRMFR NOT FOLLOW VFON0 (because syntax of A-F for variable name

Jun 8, 1995 suffix didn't work right in TS404). Install current SAS,

or you could RENAME each of the A-F variables to 10-16,

and change the ARRAYs to VFON0-VFON16.

Thanks to David Childers, Lowe's Companies, Inc, USA.

Change 13.066 Support for MVS/ESA 5.2 (compatible) SMF record changes.

VMAC24 -Type 24 SMF record variables SMF24LN4,SMF24SAN,SMF24SAC

VMAC30 were added to TYPE24, listing system affinity names; I

VMAC42 keep only the first 8 system names in SMF24SAC.

IMAC42 -Type 30 SMF record added new CPU measures CPUASRTM (the

EXTY42XR Preemptable SRBs and Client SRBs CPU time) and CPUENCTM

Jun 5, 1995 (the Enclave CPU time); note that these two new CPU times

are already included in CPUTCBTM. New variables ENCLACTM

ENCLCPSU and ENCLTRAN provide Enclave transaction active

time, CPU Service units, and transaction counts. Also,

variables IARVPSEC,IARVAPIN,IARVEPIN record page seconds

memory residency for IARVSERV shared CSTORE, and page ins

from Aux Store and Expanded Store. Those variables were

added to TYPE30_V,TYPE30_4,TYPE30_5 & TYPE30_6 datasets

but were not propagated into BUILDPDB/BUILDPD3 as yet.

Finally, Automatic Restart Management Section provides

the element name, type, restart group, and the timestamps

when IXCARM issued REGISTER,WAITPRED,READY, & DEREGISTER;

the ARMS variables were only added to TYPE30_5 until more

is understood about their usage.

-Type 42 SMF new subtype 11 for XRC Extended Remote Copy

creates new dataset TYPE42XR with the logical session

name, and per-SSID activity counts (data mover reads,

with/without data, format/update writes, etc.).

The new segments have been coded, but not all have been

tested with data records that have populated those data

segments, so verify your results.

Change 13.065 MVS/ESA 5.1 TYPE30_V (PDB.SMFINTRV) SET CLOCK causes bad

VMAC30 INTBTIME/INTETIME (interval begin/end time of day) value.

Jun 4, 1995 This note revised Jul 12, 1995 - prior belief that error

Jul 12, 1995 was leapseconds value in sysplex timer is not correct.

Fields SMF30IST/SMF30ISS should be identical, with IST on

local clock and ISS on GMT clock (so that we can measure

the GMT-to-local delta), but one site had deltas between

IST and ISS of 36 seconds for days, then 80 seconds, etc.

After chasing through IBM SMF level 2 for weeks with no

answer, the site's IBMer found an IBM note describing the

culprit: the use of SET CLOCK command:

When a system is IPLed, the GMT is exactly in sync

with the local time, but when a SET CLOCK command is

issued to change the local time to be out of sync with

the GMT (and this also causes the "SMF midnight" to be

recalculated), now jobs run show an IST-ISS delta

exactly equal to the GMT-local delta of the SET CLOCK

command! And the note says that IBM plans only to

document to clarify that changing the local time may

get the local time out of sync with the correct offset

from GMT! Note "Relationship between SMF30ISS and

SMF30IST" by David, document number (?) ADB0627.

MXG variable EXECTM in TYPE30_V becomes negative, and

variables INTBTIME/INTETIME/SYNCTIME are earlier than

all other timestamps in the interval record.

Existing TYPE30_V datasets can have the negative

EXECTM corrected by this recalculation:

EXECTM=INTETIME-INTBTIME;

Since IBM plans no change, I have revised my logic to now

expect this difference and to use it:

1. After using the FLOOR(IST-ISS) to calculate the GMT

offset in hours, I now also calculate any delta in

seconds and add that back into the GMT offset so that

ISS and IET values will be brought up to the IST

clock. After the line:

GMTOFF30=100*FLOOR(SMF30IST-INTBTIME+10)/100);

Insert this line:

GMTOFF30=GMTOFF30+SMF30IST-INTBTIME;

This logic will work when the error is fixed, too!

2. The calculation of EXECTM in the interval record does

not use LOADTIME anymore:

First, find the line:

IF MULTIDD NE 'Y' THEN DO;

and then find the subsequent line reading:

EXECTM=(INTETIME-MAX(LOADTIME,INTBTIME));

and change it to read:

EXECTM=INTETIME-INTBTIME;

(The actual change restructured the logic.)

Thanks to Wolfgang Vierling, Vereinte Versicherungen, GERMANY.

Change 13.064 Multi-UCB type 14/15 records were not properly processed

VMAC1415 due to MXG logic error. The actual order of segments is

Jun 2, 1995 basic, (open date), UCB-segments, (hiperbatch), (ISAM),

(extended segment), and there are two possible extended

segments: compressed statistics or SMS classes. Support

required relocating MXG code so that the hiperbatch, ISAM

and extended segment data are read in before processing

the UCB segments. Since MXG OUTPUTs one observation for

each UCB segment, the hiperbatch and compressed file

statistics appear only once per record, not once per UCB

segment. For multi-ucb records, then, the counts for the

hiperbatch and compressed file statistics will be kept in

the first observation from each record and will be set to

missing in the 2nd and subsequent observations created by

the additional UCB segments. Variable UCBSEGNR was also

added to identify whether observations in TYPE1415 are

created by the first or subsequent ucb segments.

Multi-ucb records are created for concatenated BPAM files

for multi-volume files, and for striped files.

Real data records clarified IBM documentation of the real

order of these segments.

The major impact without this fix was that multi-ucb obs

had SMS storage, data, and management class blank in all

but the last observation for that SMF record (and the

compression statistics could have been incorrect).

Thanks to Paul Dzielak, Smith Barney, USA.

Change 13.063 Variables TTOCPREL,TTOCSUCL have value of *NONE* unless

VMXGHSM the volser started with H; a local tailoring mod was left

Jun 2, 1995 in the code that should have been removed. Delete the

two lines that contain '*NONE*'.

Thanks to Wanda Prather, Johns Hopkins Applied Physics Lab, USA.
Change 13.062 MXG 13.01 only.

VMACNSPY -Remove "$HEX16." from line 124700, as those variables are

Jun 2, 1995 should be $HEX.2. Fortunately no harm was done, but the

variables were stored as 8 bytes instead of 1 byte!

-Labels for VRBMAXWD,VRBMINWD are now SIZE vice COUNT.

-Insert STARTIME=ENDTIME-DURATM; immediately before the

%INCLUDE SOURCLIB(EXNSPYVT); to create STARTIME.

-Variable APPLID was removed from the KEEP= list for the

datasets NSPYVTAM and NSPYVIRT as it does not apply.

-INPUT of variable NSPBPOOL was changed to $EBCDIC4. (it

was &PIB.4.) and it is now labelled 'BUFFER*POOL*NAME'.

Thanks to Rod Fernandes, Albert Heijn B.V., NETHERLANDS.

Change 13.061 All MXG-built formats for hexadecimal values now have

FORMATS OTHER=(| $HEXn. |) - for character variables

Jun 1, 1995 OTHER=(| HEXn. |) - for numeric variables

added in their creation so that undefined values are

displayed in hex. This syntax is only available in

SAS Version 6.07 and later, so if you still must try to

run MXG under archaic SAS 6.06 or more archaic SAS 5.18,

you will have to remove all of those lines from member

FORMATS before you build your archaic format library.

Thanks to Shantha Baran, British Steel, ENGLAND.

Change 13.060 MXG 13.01 only. Change 13.055 was not spelled right.

GRAFLPAR The test IF NAME GT ' ' THEN DO; added by 13.055 should

Jun 1, 1995 be IF LP&ID.NAME GT ' ' THEN DO;

Thanks to Norbert Korsche, OMV AG, AUSTRIA.

Change 13.059 -Label corrections. Variable NSPBOEXP "buffers" misspelled

VMACNSPY and variable CBPFLAG was changed to "TABLE*FLAG".

Jun 1, 1995 -Format correction. Variable NSPXRESO was removed from the

$HEX16 format, as it is a character, not hex, variable.

-MXG 13.01 only, new support for record type X caused

"INPUT STATEMENT EXCEEDED RECORD LENGTH" error. The line

M4=-4; must be changed to M8=-8; and the subsequent

reference to +M4 must be changed to +M8.

Thanks to Norbert Korsche, OMV AG, AUSTRIA.
Change 13.058 ANALDB2R error "BY VARIABLE STRTTIME IS NOT ON INPUT DATA

ANALDB2R SET WORK.DB2DUMRY" resulted from bad MXG logic. Find

May 30, 1995 "%MACRO PMSTA01", and change the next two occurrences of

" BEGTIME" to " STRTTIME" (note, don't change &BEGTIMEs):

First: OR &PMSTA01 NE YES %THEN BEGTIME;

Second: DURATM=ENDTIME-BEGTIME;

Thanks to David Klein, City of New York, USA.
=====Taught in London and Heidelberg, paper at Konigswinter CMG-CE======
Change 13.057 CICS Statistics dataset CICLSRR variables A08BKCTD and

VMAC110 A08BKDTD were incorrectly INPUT as RMFDUR4. Their input

May 4, 1995 lines (two pairs) must each be replaced with:

HH &PK.1. MM &PK.1. SS &PK.1. TH ?? &PD.1. @;

IF TH GE 0 THEN A08BKCTD=HMS(HH,MM,(SS+TH));

ELSE A08BKCTD=.;

INPUT

Thanks to Steve Harris, TVA, USA.

Change 13.056 Change 13.017 (TYPE6 OUTDEVCE for 4-digit Remote) was not

VMAC6 correct. The "5,2" in the SUBSTR should be "6,2". The

May 4, 1995 text of 13.017 has also been changed to be correct.

Thanks to Norbert Korsche, OMV, AUSTRIA.

==Changes thru 13.055 were included in MXG 13.01 dated May 3, 1995===
Change 13.055 Intermittent LPARs (test LPARs) caused strange graphs;

GRAFLPAR The statement IF LP&ID.UPDTM GT 0 THEN DO;

May 3, 1995 should be IF NAME GT ' ' THEN DO;

so the LPAR numbers don't change from hour to hour!

Thanks to Tom Kelman, SunTrust Service Corporation, USA.
Change 13.054 Variable NRSAMPLE was added to the TYPE74CF KEEP= list.

VMAC74 Several calculations on the Coupling Facility require the

May 3, 1995 number of samples.

Thanks to Bruce Widlund, Merrill Consultants, USA.

Change 13.053 If you used TEMP01 or TEMP02 parameters to segregate the

XMXGSUM work files, variables were missing because dataset VARS

May 3, 1995 was not in the DELETE list. In addition, the logic has

been significantly simplified by using SUM(A B C)=A B C

(instead of the archaic SUM= list with X1-Xnn placeholder

dummy variables). This might be the last iteration as

XMXGSUM has now been tested with just about all feasible

combinations, but I will wait until MXG 13.02 before I

actually replace VMXGSUM with this XMXGSUM. I strongly

suggest that you install XMXGSUM in place of VMXGSUM as

described in the comments in XMXGSUM - XMXGSUM is much

more efficient, especially for CICS and DB2 data.

Thanks to Diane Eppestine, Southwestern Bell, USA.
Change 13.052 When ASUMTALO was revised (Change xx.yyy), corresponding

GRAFTALO changes to GRAFTALO should have been made. SYSTEM was

May 2, 1995 removed, ALOCTIME changed to TIMESTMP and the average

tape drives allocated is TAPEDRVS=TAPEDRVS/DURATM;.

Thanks to Tom Elbert, John Alden Life Ins., USA.
Change 13.051 AS/400 3.1 dataset QAPMSYS duration variables SITRNT,

VMACQAPM S6TRNT and SPRES1 are INPUT as &PD.8.3 instead of &PD.8.

May 2, 1995 V 3.1 increased the field width from 6 to 8 bytes but the

change in units to milliseconds was not documented! In

addition, SPRES1 and SPRES2 were renamed SPTRNT and

SPTRNS and added to the KEEP= list for QAPMSYS dataset.

(The "Passthru" SICPU/SITRNT/SITRNS and the "inter"

SPCPU/SPTRNT/SPTRNS variables are both needed to track

total interactive work.) See also Change 13.011.

Thanks to Lynn Marchant, Coca Cola, USA.

Change 13.050 TYPE42DS I/O Statistics Subsection does not always exist,

VMAC42 but MXG input the variables nevertheless, causing really

May 2, 1995 large values for I/O durations and cache statistics.

Insert IF OFFDSIO GT 0 THEN DO;

before OFFDSIO=OFFDSIO-3+OFFSMF;

Before OFFDSAM=OFFDSAM-3+OFFSMF;

insert END;

So the fields will be missing instead of wrong. I am

still investigating why OFFDSIO (S42DSIOO) is zero.

Thanks to Ted Blank, IBM, USA.

Change 13.049 TYPE116 dataset has zero observations. Change both

VMAC116 IF SUBTYPE=1 THEN ... to IF SUBTYPE=0 THEN .... Insert

May 2, 1995 OFFQMAC=OFFQMAC-3+OFFSMF; before INPUT @OFFQMAC

Thanks to E.U.C. Hettinga, Centrum Voor Informatie.., NETHERLANDS.

Change 13.048 XMXGSUM in MXG 13.01 still failed, if the TEMPDD argument

XMXGSUM was used, due to mislocated code when Change 13.032 was

Apr 26, 1995 made. Also, repeated executions of XMXGSUM could fail

because temporary dataset VARS was not deleted at the

end of XMXGSUM.
Change 13.047 ANALTAPE in MXG 13.01 still failed, because Change 13.044

ANALCNCR to ANALCNCR created an error when there was no SUMBY used

Apr 26, 1995 in the %ANALCNCR invocation. (Only ANALTAPE had the

exposure). (See also Change 12.327).