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

MXG examples, if your %VMXGSUM uses an OUTCODE= parameter

to re-calculate the DATETIME= variable-name, your results

may be seriously wrong. If your code looks like this:

DATETIME=STRTTIME,

INTERVAL=MONTH,

SUMBY =APPLID DATETIME TRANNAME,

OUTCODE =STRTTIME=DATETIME; DROP DATETIME; ,

it must be replaced with this:

DATETIME=STRTTIME,

INTERVAL=MONTH,

SUMBY =APPLID DATETIME TRANNAME,

ID =STRTTIME,

Prior to 12.084, the DATETIME= variable-name was forced

to LENGTH 8, but when the new ID=variable-name parameter

was added to replace the recalculation in the OUTCODE=

parameter, the automatic forcing of LENGTH 8 was dropped.

All MXG invocations of %VMXGSUM were changed, but I did

not realize nor document this exposure if you did not

make the change.

(Eliminating OUTCODE= eliminates a data step; also, since

12.084+ keeps only the needed variables, the variable

might not even exist to have its length changed.)
When a datetimestamp value is stored in length 4, it can

be truncated by up to 255 seconds, which can cause the

summarization value to be on the prior day, week, or

month!

No code was changed by this documentation-only change.

Thanks to Ahn Ngo, Crowley Maritime Corporation, USA.

Change 13.151 -HP PCS variables NONEW NOKILL and NOSHORT are characters,

VMACHPAI not numerics. Add all three variable names to the $8

VMACHPSU FORMAT statement in VMACHPUX, and add NONEW and NOKILL to

VMACHPUX the $8 FORMAT statement in VMACHPAI and VMACHPUX to fix.

VMACHPUX -Comparing CPU time in the HPUXGLOB and HPUXAPPL datasets

Aug 2, 1995 shows expected comparison (HPUXAPPL records over 90% of

the global CPU time. However, HPUXPROC shows more CPU

time than was recorded in the HPUXGLOB, probably because

the CPU time IN HPUXPROC is in seconds (rather than tenth

of seconds), and it appears that fractional CPU seconds

are written as 1 second. The site has "Short Lived On",

causing records for tasks running less than one second,

which may contribute to the discrepancy. HP is being

contacted concerning the CPU time resolution in HPUXPROC.

Thanks to Dan Sidorick, Smith Klein Beecham, USA.
Change 13.150 MXG 13.04 tape only. Deleted line will cause ABEND.

ASMTAPES After label TIMRPOST, following the WTO statement and

Jul 31, 1995 before the PURGEDQ statement, insert:

L R8,PRGERMTR

MAINTLEV was changed from 4 to 5 to reflect this change.
========MXG Version 13.04, dated Jul 31, 1995, thru 13.149======
Change 13.149 Support for Kodak's ACXIS software for Kodak Optical Disk

ADOCAXC (by Data/Ware Development), Version 2.1.0, thanks to this

EXAXC01X significant user contribution creates ten new datasets.

EXAXC02X AXC01X - AXCIS DATA CREATION SUBTYPE 01X'

EXAXC02I AXC02X - AXCIS INDEX EXTRACTION SUBTYPE 02X'

EXAXC03X AXC02XIS - AXCIS INDEX EXTRACTION SEGMENT XIS'

EXAXC04X AXC03X - AXCIS LOCAL INDEX UPDATE SUBTYPE 03X'

EXAXC05X AXC04X - AXCIS GLOBAL INDEX UPDATE SUBTYPE 04X'

EXAXC06X AXC05X - AXCIS ARCHIVAL SUBTYPE 05X'

EXAXC07X AXC06X - AXCIS USER STATISTICS SUBTYPE 06X'

EXAXC08X AXC07X - AXCIS RETRIEVAL ACTIVITY SUBTYPE 07X'

EXAXC09X AXC08X - AXCIS FAILED RETRIEVAL ACT. SUBTYPE 08X'

IMACAXC AXC09X - AXCIS UNAUTH. ACCESS ATTEMPT SUBTYPE 09X'

TYPEAXC

VMACAXC

Jul 30, 1995

Thanks to Stephen Bell, Westfalisch Lippischen Sparkassen, GERMANY.
Change 13.148 I have never actually published standards for MXG code.

VMACXXXX Many users have contributed code that works, but I often

Jul 30, 1995 have to spend some time revising their code to my format,

because I have never formalized exactly how to write MXG

code. Member VMACXXXX and the other ....XXXX members are

good examples of the basic code structure, but I will now

begin to include specific coding standards that I try to

us, some cosmetic, some functional, for now in VMACXXXX,

but likely to be moved to a different-named member in a

subsequent iteration.

Change 13.147 Support for LEGENT's HyperBuf SMF Record adds new dataset

FORMATS TYPEHBUF, thanks to this user contribution. The original

EXTYHBUF code was tested with Beta Release 4.02 of HyperBuf,

IMACHBUF although I changed some of the code (see change 13.148)

TYPEHBUF and only syntax checked afterwards, as the test data file

VMACHBUF had not yet arrived for validation.

Jul 30, 1995

Thanks to Joseph G. Ogurchak, Westfield Companies, USA.

Change 13.146 Five variables in NDM Connect Direct NDMMC dataset

VMACNDM (NDMTNAME,NDMFNAME,NDMANAME,NDMNRECS,NEMNBLKS) and four

Jul 30, 1995 variables in NDMWO (NDMTNAME,NDMFNAME,NDMANAME,NDMTSKNO)

were wrong because NDMUID was expanded from 8 to 64 bytes

in these records. The MXG test:

(NDMRTYPE='MC' AND LENGTH=168) must be changed to read

(NDMRTYPE='MC' AND (LENGTH=168 OR LENGTH=214)) to fix.

In addition, the actual change now inputs and keeps new

variable NDMWODAT, the undocumented text string at the

end of the WO record. (I suspect my DSECTS are old!).

Thanks to ???, Alcatel, FRANCE.
Change 13.145 Support for Packet/Main SMF record creates 11 datasets

EXPKMNBI thanks to this significant user contribution:

EXPKMNBL PKMNCONN - Connect to TIC

EXPKMNCO PKMNDISC - Disconnect from Packet Network

EXPKMNDI PKMNBIND - Bind Completed

EXPKMNER PKMNUNBN - Unbind processed for user

EXPKMNFE PKMNERR - Error Events

EXPKMNFL PKMNSTUP - Startup of PKTMN Program

EXPKMNFS PKMNSHUT - Shutdown of PKTMN Program

EXPKMNSH PKMNBLOG - Bind Image Log (for Session Data Logging)

EXPKMNST PKMNFLOG - Outbound FMD Logical Message (for ")

EXPKMNUN PKMNFTST - File Transfer Start

IMACPKMN PKMNFTEN - File Transfer End

TYPEPKMN The first five datasets have been validated with data.

VMACPKMN

Jul 29, 1995

Thanks to Mr. Loewenthal, Zonussi Elettrodomestici, ITALY.
Change 13.144 Support for NetCompress SMF record creates six datasets,

DIFFNTCP thanks to this significant user contribution:

EXNTCPAP NTCPINAP - Application Interval - type 03

EXNTCPOP NTCPINSY - System Interval - type 02

EXNTCPRT NTCPSESS - Device Session Start - type 05

EXNTCPSE NTCPSESE - Device Session End - type 06

EXNTCPSS NTCPSTRT - System Startup - type 01

EXNTCPSY NTCPSTOP - System Stop - type 04

EXNTCPOP Since the NTCPINSY dataset contains accumulated fields,

IMACNTCP member DIFFNTCP is automatically included in TYPENTCP,

TYPENTCP Note if you tailor BUILDPDB to add NTCP processing, you

VMACNTCP must also add %INCLUDE SOURCLIB(DIFFNTCP) in EXPDBOUT.

Jul 29, 1995 However, since the documentation does not describe which

fields are in fact accumulated versus end point values,

I had to deduce from the test data which fields to DIF(),

so please validate closely that I got them all!

Thanks to Mr. Loewenthal, Zonussi Elettrodomestici, ITALY.
Change 13.143 TSO/MON 6.1 only, variables TRIVTM NTRIVTM and LONGTM are

VMACTSOM were 38400 times too small; replace the three TU4. with

Jul 29, 1995 &PIB.4. and the values will be correct. (When I validated

the 6.1 changes last fall with test data, I only looked

the counts of transactions, and failed to perform a

reasonableness check on the response time values.)

Thanks to Yingha Mui, Canada Life, CANADA.
Change 13.142 Variables LCSLCBR and LETHCBR were misspelled as the 2nd

VMAC28 occurrence of LSCLCBS and LETHCBS, and the second label

Jul 28, 1995 for LCSLXBYT was deleted.
Change 13.141 Cosmetic changes. Duplicate variable names occurred in

VMACHPUX KEEP= statements, causing no error, but as they really

VMACIAM serve only to confuse, the duplications were removed.

VMACOMCI The MEMBER(VARIABLE) that were changed in each member:

VMACXAM VMACHPUX(EXPORTIM), VMACIAM(IAMAVGRL), VMACOMCI(EAMAXT),

VMAC110 VMACXAM(SLORESP1,SLORESP2), VMAC110(LDGDPSCR),

VMAC7072 VMAC7072(CPUWAITD,CPUWAITE,CPUWAITF), VMAC99(PHSPS).

Jul 27, 1995 (Freddie's diagnostic program, which reads MXG source to

look for inconsistencies, found these duplications.)

Thanks to Freddie Arie, Lone Star Gas, TEXAS.

Change 13.140 DFSORT Release 13 only. Variable ICEINPDS was repeated

VMAC16 four times in the KEEP=, the LABEL, and the INPUT

Jul 27, 1995 statements. The 2nd thru 4th occurrences of ICEINPDS

should have been ICEINNDS, ICEOUTDS and ICEOFLDS.

Thanks to Freddie Arie, Lone Star Gas, TEXAS.
Change 13.139 Almost cosmetic; MXG previously kept only the first 9

DIFFDB2 IFCs and Destinations in DB2STAT0, and printed a message

VMACDB2 ("This is not usually important") if additional segments

Jul 27, 1995 were found in your record. Since this caused a few calls

I elected to increase the kept IFCs and Destinations to

14, since that appears to be the default count. These

counts of activity by IFC and Destination are rarely

needed, but now they exist. See Change 12.279.

Change 13.138 Using TYPEMON8 to process Landmark TMON/CICS Version 1.3

TYPEMON8 will cause the "IT APPEARS YOU ARE TRYING TO PROCESS ..."

Jul 26, 1995 message about decompression exit. That text now reminds

you that TYPETMON is required for Version 1.3 records.

Thanks to Jerry Boggess, County of Sacramento, USA.
Change 13.137 The MICS VM Data Transmission Program's VB output file is

VMACVMXA already supported in MXG VM/ESA member VMACVMXA, by using

Jul 26, 1995 the _VMINPUT macro instead of the _MWINPUT macro.

Only comments were changed by this Change.

Thanks to Ian Davies, Workers' Compensation Board of Alberta, CANADA.
Change 13.136 Variables HMF5BUS and HMF5TIME were not in the KEEP= list

VMACHMF for dataset TYPEHMF5, and they were also left out of the

Jul 26, 1995 LABEL statement in change 13.038.

Thanks to Shaheen Pervaiz, Acxiom CDC Inc, USA.

Change 13.135 This revision to the MXG Tape Mount and Tape Allocation

ASMTAPES monitor provides SRB reliability enhancements and adds

VMACTMNT new diagnostics. This is MAINTLEV 4 of the monitor and

Jul 25, 1995 should resolve several erratic problems with the SRB

Jul 29, 1995 routine (including, possibly, the occasional stall where

the monitor stops writing SMF records). This revision

creates a separate mount record for PROGRAM=IEFIIC, which

is now known to be the program name seen when a job step

is delayed in allocation recovery, and I intend to see if

there is any way to match up this IEFIIC record to its

related job, but I have to get the monitor into test site

hands to create the test data to examine first! Because

these PROGRAM=IEFIIC records represent allocation rather

than mount delays, I chose to delete them in VMACTMNT so

your statistics will not be affected. Sites wishing to

examine these records can remove that test from VMACTMNT.

Thanks to Tom Parker, Hogan Systems, USA.

Thanks to Jim Purdie, BancOne, USA.

Thanks to David Childers, Lowe's, USA.

Thanks to Shaheen Pervaiz, TransUnion, USA.

Thanks to Tracy Blackstone, Kaiser Permanente, USA.
Change 13.134 Date/time selection failed, especially crossing midnight.

ANALRMFR Find and delete each of the 44 lines containing either

Jul 24, 1995 IF TIMEPART("&BEGTIME"DT) LE TIMEPART(STARTIME);

or

IF TIMEPART("&ENDTIME"DT) GE TIMEPART(STARTIME);

Thanks to Norbert Korsche, OMV AG, AUSTRIA.
Change 13.133 Candle's Supersession Release 147 causes MXG message that

VMACNAF "More account fields were encountered ..." is corrected

Jul 24, 1995 by Candle PTF QLV1372.

Thanks to Mr. Schott, Rehau AGu. Co., GERMANY.

Change 13.132 Variable DURATM was neither calculated nor kept in the

VMAC28 NPMVSVGB data set; it was added to the KEEP= list and

Jul 24, 1995 created in the 'VGB' segment processing.

Thanks to Carl Sommer, SAS Institute Cary, USA.

Change 13.131 Several corrections to HSM FSR SMF record processing:

VMACHSM -Variable FSRTBYTW was mis-calculated. It should be:

Jul 24, 1995 FSRTBYTW=SUM(FSRTBYBK,FSRTBYTW);

-The DO I=1 to FSRNENT1 loop should have included all of

the SUM functions and the FSRTVOLS logic, with the

%INCLUDE SOURCLIB(EXHSMFST) and END relocated to just

before the DO on FSRNENT2. Before this change, MXG only

output the counts for the last tape when this FSR had

multiple tapes, and always had TPTYPE='INPUT'.

-Variable FSRTVOLS was added to LENGTH statement as $30.

Thanks to Wanda Prather, Johns Hopkins University APL, USA.
Change 13.130 Internet addresses FTPCLHST FTPCRHST F20LHST and F20RHST

VMACILKA were numeric variables in hex; with this change they are

Jul 24, 1995 character variables of length 15 with numbers and decimal

points between them. Inputting each of the four bytes as

individually using PIB1. and then concatenate the PUT

function of each byte, using:

ADDR=PUT(B1,3.)!!'.'!!PUT(B2,3.)!!'.'!!PUT(B3,3.)!!'.'

PUT(B4,3.));

Note added Aug 1996: Data from before and after this

change cannot be combined, as these four variables are

numeric before and character after. If you must combine

(i.e., year to date), then you must convert these four

variables in your existing accumulation from numeric to

character, using this logic:

MACRO _CONVERT

B1=FLOOR(X/65536);

B2=FLOOR((X-B1*65536)/4096);

B3=FLOOR((X-B1*65536-B2*4096)/256);

B4=MOD(X,256);

ADDR=PUT(B1,3.)!!'.'!!PUT(B2,3.)!!'.'!!PUT(B3,3.)!!'.'

PUT(B4,3.));

%

DATA NEWACCUM; SET OLDACCUM;

DROP FTPCLHST FTPCRHST F20LHST F20RHST;

X=FTPCLHST; _CONVERT; XTPCLHST=ADDR;

X=FTPCRHST; _CONVERT; XTPCRHST=ADDR;

X=F20LHST; _CONVERT; X20LHST =ADDR;

X=F20RHST; _CONVERT; X20RHST =ADDR;

DATA OLDACCUM; SET NEWACCUM;

DROP XTPCLHST XTPCRHST X20LHST X20RHST;

FTPCLHST=XTPCLHST;

FTPCRHST=XTPCRHST;

F20LHST =X20LHST ;

F20RHST =X20RHST ;

Thanks to Mohammad Mourad, AT&T, USA.

Change 13.129 DB2 Statistics Report again, but only individual fields!

ANALDB2R -PMSTA01/PMSTA02 labels for Buffer Pools BP0 and BP1 for

DIFFDB2 GENERAL and READ ACTIVITY were printed as BP1 and BP2.

VMACDB2 In ANALDB2R find GENERAL and make the following change,

Jul 24, 1995 and then find READ ACTIVITY and make the same change:

Jul 25, 1995 %IF &B EQ 1 %OR &B EQ 2 %THEN %DO;

Jul 26, 1995 @1 "BP&B GENERAL "

Jul 27, 1995 %END;

must be replaced with:

%IF &B EQ 1 %THEN %DO;

@1 "BP0 GENERAL "

%END;

%ELSE %IF &B EQ 2 %THEN %DO;

@1 "BP1 GENERAL "

%END;

-The LOG RBA field is printed as a decimal when it is a

hex address, and must be stored as length 8.

In ANALDB2R, find this line with QWSDLR and add HEX12.:

+2 'LOG RBA :' QWSDLR HEX12. ;

In VMACDB2 add variable QWSDLR to the LENGTH statement

to make it length 8 instead of the default length of 4,

and add a line QWSDLR HEX12. to the FORMAT statement.

(I chose HEX12 for an 8-byte field because RBA is a 24bit

address, and because HEX16 produces wrong values - see

SAS Technical Note).

In DIFFDB2, remove the line QWSDLR=DIF(QWSDLR);

-The second of the two lines with this label is wrong:

DBD REQUESTS NOT IN EDM

Change that 2nd label to DBD REQUESTS/DBD NOT IN EDM

The original calculation was always correct.

-The value for REQUESTS FOR PT SECTIONS in the Statistics

summary is incorrect. Find the line reading

+3 'REQUESTS FOR PT SECTIONS '

and change the next line to QISEKTG instead of QISEKT.

Now, find the previous SUM= statement and add QISEKTG.

-Statistics Detail report, page 4, right half of page

(AUTHORIZATION ATTEMPTS thru LOOK AHEAD MOUNT SUCCEED):

The /MINUTE /THREAD /COMMIT columns contain wrong values

(the left hand COL1xxxx variables are printed instead of

the right hand COL2xxxx variables). You must find each

of the report labels for the right hand column, and for

those that are followed by lines with COL1PMIN COL1PTHD.

and COL1PCOM, change the COL1xxxx to COL2xxxx. Twenty-one

sets of three lines were changed.

-Statistics Detail report, page 6, right half of page,

(GET PAGES thru RANDOM GET PAGE RESULTS), same error as

page 4 - file labels and change COL1xxxx to COL2xxxx.

Thanks to Neil Ervin, Huntington Service Company, USA.

Thanks to Cal Cooke, Huntington Service Company, USA.

Thanks to Dean Berry, Farmland Industries, Inc, USA.

Change 13.128 Support for APAR OW13246 (PSF/MVS Release 2.2.0) adds new

VMAC6 variables SMF6BYTE (bytes transmitted) & SMF6TARG (target

Jul 23, 1995 address) to dataset TYPE6, for sites exploiting the new

PSF/MVS Download feature. Note that these new variables

are not automatically carried into the PDB.PRINT dataset

with BUILDPDB, but you can tailor IMACPDB easily to add

them if you need them.
Change 13.127 This change is for documentation of an incompatibility

FORMATS with SAS Version 5; there was no actual change. Beginning

Jul 23, 1995 with MXG 13.01, member FORMATS cannot be executed under

SAS version 5 or SAS version 6.06, nor can it be executed

if the output LIBRARY is a version 5 format library. The

syntax OTHER=(| $HEX2. |) added by Change 13.061 did not

exist in those archaic versions of SAS. However, if you

delete all of the lines in member FORMATS with that

OTHER= syntax, the formats program will execute without

error, even on those old versions.

Change 13.126 This change was revised after Newsletter 28 was printed.

VMACVMXA VM/ESA MONWRITE records written both by Sterling's

Jul 23, 1995 VM/Monitor and by IBMs MONWRITE have changed in VM/ESA

Jul 24, 1995 2.2. MXG calculated the number of control records using

Aug 22, 1995 NRCRRECS=(IPRMMSG2-IPRMMSG2+1)/12; (even though there is

only one control record per block), and verified that the

record was a control record by testing IPARMLF1=8709x,

but the IPARML control block has changed. Sterling writes

records with IPRMMSG1=IPRMMSG2=028x that have MWTCAREA of

all zeroes (so this is not a valid control record, altho

028x is the correct offset to where MWTCAREA should be!),

and IBM writes records with MSG1=MSG2=3Fx that do contain

a valid MWTCAREA, but 3Fx is the wrong offset to it!

(These heuristics were needed in early VM/XA records, but

now that the IPARML has changed, they can and must be

removed with these changes:

-The block IF MOD(IPARMLF1,10000X) NE 8709 ... to END;

was deleted.

-The calculation NRCRRECS=(IPRMMSG2-IPRMMSG1+1)/12; and

the following INPUT @IPRMMSG1+1 @; was replaced with:

NRCRRECS=1;

INPUT @41 @;

(the code added by the July change, between these two

lines was also deleted by this August change).

-After the @; after CAEND is input, insert this code:

IF CASTR=0 AND CAEND=0 THEN DO;

NRCRRECS=0;

DELETE;

END;

With these changes, MXG now skips over the MWTEXTBF area

that contains the IPARML control block, and instead reads

the MWTCAREA directly to determine how much data (if any)

follows this control record.

Thanks to Ian Davies, Workers' Compensation Board of Alberta, CANADA.

Thanks to Connie Mak, Avon, USA.
Change 13.125 Landmark TMON for MVS corrections. The line reading:

VMACTMVS IF JDSTCBS GT 10000 THEN DO;PUT _N_= COL=;LIST;STOP;END;

Jul 23, 1995 was deleted. That debugging code for a bad record (over

Aug 21, 1995 10,000 TCBs in real memory) caused the MXG job to stop

before the end of the input file, and there was no error

message nor condition code set by MXG; the only clue was

a single record hex dump printed on the SAS log.

Variables STARTIME and DURATM were removed from the KEEP=

list for dataset TMVSCC (which is an event record, not an

interval), and SYSTEM=CCCSMFID; was inserted before the

%%INCLUDE SOURCLIB(EXTMVCC);

to correct the blank value for SYSTEM in that dataset.

Also, variable CFIUTYPE was changed from numeric to

character and formatted $HEX8, to be consistent with the

other occurrences of unit type in other TMVS records. I

recognize this may cause a problem if you combine TMVSCF

data sets built before and after this change.

Thanks to Rod Fernandes, Albert Heijn B.B., BELGIUM.

Change 13.124 RMM SMF Record caused INVALID DATA FOR MDUCDATE/MDUCTIME

VMACEDGS if there had been no user change because the PD4 fields

Jul 21, 1995 were nulls. Insert ?? between the variable name and

the &PDB.4. format name in the two lines which input the

variables MDUCDATE and MDUCTIME.

Thanks to Bob Newbury, University College System of Nevada, USA.

Change 13.123 Variables B1DIS B1INT CPGM LPGM ROBID ROBTY and SMSMC are

TYPETMS5 added to the output datasets created by TYPETMS5; these

Jul 21, 1995 variables (new in TMS Release 5.1) were added to VMACTMS

by Change 13.083, but were not carried into the datasets

created by the TYPETMS5 processing.

Thanks to Norbert Korsche, OMV AG, AUSTRIA.

Change 13.122 Variable RACFUSER from type 30 was added to the DSETOPEN

ANALDSET and DSETSUMS datasets created from 14/15/64 records so

Jul 21, 1995 that Auditors can know the submittor of the job that did

the access to the file.

Thanks to Freddie Arie, Lone Star Gas, TEXAS.
Change 13.121 Reading your 3990 Mod 11 Fat DASD with SAS INFILE CVAF

VMXGVTOC option will fail, because the DSCB 5 2-byte addresses are

Jul 19, 1995 3-bytes for those devices. Removing the CVAF option will

circumvent the SAS error, because MXG does not need the

DSCB 5 records to constuct the VTOC datasets. However,

far better than using this old VMXGVTOC technology, the

ASMVTOC/VMGXVTOF pair have been the recommended VTOC tool

in MXG for years. If you now have DFSMS, using their

DCOLLECT is even better, since it includes the VSAM and

VVDS data that previously required you to also run the

ASMVVDS/TYPEVVDS pair. I have removed the CVAF option

from the INFILE statement to prevent this error, should

anyone still be using VMXGVTOC and not read this note.

April 20, 2006 Notes:

1. Member JCLDAYDS and DAILYDSN are the members to

use to capture/bill-for disk space and tapes.

2. SAS Note SN-015522 for SAS Version 9.1.3 does

now recommend specifying the CVAF option in the

INFILE statement to circumvent an 0C4 ABEND in

module UNKNOWN function VXVTC at offset 00082C.

Thanks to Terry Poole, SAS Institute Cary, USA.
========MXG Version 13.03, dated Jul 19, 1995, thru 13.120======