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

TYPE16 record), and uses those selected TYPE16 obs to

create a table lookup that is then used to re-read SMF

and select only the matching SMF TYPE1415 records; the

site could thus find the dsnames of all SORTIN datasets.
Program ANAL30 reads SMF to create only TYPE30_4 obs for

steps you selected (this example, programs IEBCOPY and

IEBGENER), creates the lookup table, which is then used

to select all of the non-VSAM TYPE1415 and VSAM TYPE64

datasets that were opened by that step.
These examples are easily extended to select other

pairs of data where one event defines a timerange, and

the matching events must occur within that timerange.

I've been meaning to write this example technique for

some time; this specific request precipiated the program!

Thanks to David Goldstein, EDS, AUSTRALIA.

Change 19.256 Variable UMRECRD in dataset DCOLMIGS is now formatted by

FORMATS the new $MGDCORF format to decode the Record Format when

VMACDCOL UMRECRD is printed or displayed. And variable DCERECRD

Jan 4, 2002 in dataset DCOLDSET is now kept, and formatted with the

new $MGDCORF format, also displays the Record Format (and

eliminates the need to combine the same information from

the variables DCDRECFA/B/C/F/J/S/U/V to get it!

Thanks to Wanda Prather, The John Hopkins Applied Physics Lab, USA

Change 19.255 Comments only. If you add NPM TYPE28 processing to your

DIFF28 BUILDPDB, you must either add %INCLUDE SOURCLIB(DIFF28);

Jan 2, 2002 in member EXPDBOUT (to deaccumulate the NPMINPMT data) or

if you want all NPM datasets written to your PDB library,

then you would instead add _S28 in member EXPDBOUT, as

that "product" sort macro sorts (and deaccumulates where

necessary) all datasets created by TYPE28 code.

The DIFF28 member contains only _S028IN7, the "dataset"

sort macro for NPMINPMT, which is included in _S28 macro.

DIFFxxx members exist only for products with accumulated

data fields, and MXG always invokes the DIFFxxx member in

both the TYPExxx and TYPSxxx members, to hopefully assure

that you always have legitimate (i.e., deaccumulated)

values. But now, the DIFF logic is contained in the

"dataset" sort macro, so the DIFFxxx members contain only

the specific datasets that must be deaccumulated.

Thanks to Bruce Hewson, CitiCorp, N.A., SINGAPORE.
Change 19.254 Support for the configuration record (NTCONFIG dataset)

VMACNTSM adds variables DCSDURATM (DCS Interval) and DSCVRYRT

Jan 2, 2002 (Discovery Record Types) - existing fields not decoded,

and variable SUMRYVER (Summary Version Number), to be

added to the record when this data file was created by

NTSMF's summarization program, which is also the flag

that this input file contains summarized data.
Change 19.253 Variable LABEL in TYPETMNT was wrong; the INPUT location

VMACTMNT of TMNTLAB should have been the first byte, with the

Dec 30, 2001 second byte unused.

Thanks to Mike Jacques, Branch Banking & Trust, USA.

Change 19.252 "New" DFSMS/rmm variables include Creating Program Name,

VMACEDGR Total Block Count across all volumes, and "Last Used"

Dec 30, 2001 Program/Job/Step/DD/DEVNR are now INPUT and kept in

TYPEEDGR. They were added by IBM by OW40710 in 1999.

Thanks to Joe Lodyga, Whirlpool Corporation, USA.
Change 19.251 Use of long-variable-names was not fully supported in

VMXGSUM VMXGSUM; names could be truncated to eight bytes. Since

VMXGINIT MXG does not use long variable names, this only impacted

Dec 27, 2001 use of VMXGSUM with long variable names in SAS V8.

Dec 30, 2001 To accomplish this change, TEMPnn libnames will be built

Jan 4, 2002 with ENGINE=V8SEQ under V8, and ENGINE=V6SEQ under V6,

Jan 20, 2002 and TEMPnENG is set blank if TEMPnn is not used.

Corrected logic if KEEPALL=NO was used.

-In member VMXGINIT, new global macro variable MXGLEN

defaults to 4, and it is used by VMXGSUM (instead of the

hardcoded 4) to set the length of variables on the output

side of VMXGSUM, for those rare cases where you need to

increase kept length, by using %LET MXGLEN=8; before

your %VMXGSUM invocation. NO LONGER TRUE.

PARAGRAPH WAS CHANGED BY CHANGE 19.272 in JANUARY.

-PROC SQL's scope was limited to look only at LIBNAMEs;

under very unlikely circumstances, it could read all SAS

datasets on a multi-volume tape data library; nothing

failed, but the job took longer than expected.

Thanks to Paul Gillis, Pacific Systems Management Pty, AUSTRALIA.

Change 19.250 The Logically Swapped ESTORE in variable ESFRLSAV was

VMAC71 wrong; ESFRHSAV was being added instead of subtracted.

Dec 27, 2001 ESFRLSAV=ESTORE-(ESFRTOAV+ESFRHSAV); is now used.

Thanks to Peter Webber, CIS, US.

Change 19.249 Major revision in support for Web Logs.

EXWWWCOO The original TYPEWWW was completely restructured and now

EXWWWIIS creates multiple datasets, and more are likely over time.

EXWWWLOG -Multiple web log data formats are automatically detected,

EXWWWREF and multiple "Log Event" datasets will be created where

EXWWWURQ the contents of the logs are sufficiently different. An

IMACWWW observation in the "Log Event" dataset is created for

VMACWWW each web log record that is not filtered. At present:

VMXGINIT

TYPEWWW Dataset Description of Web Log

TYPSWWW WWWLOG Netscape/I-planet,Common Log Format,Websphere

Jan 7, 2002 WWWIIS Microsoft IIS (old and new versions)

-"Multiple segment" datasets are created for content items

when there can be more than one item in a log event. The

individual segments can be mapped back to their parent

log event observation using the ZTIME/WEBSEQNR variables.

At present, these multiple segments are decoded:
Dataset Description of contents

WWWCOOKE Cookies, COOKNAME and COOKVALUE pairs

WWWREFER Referer text, may be revised.

WWWURQRY URI Query, URIQNAME, URIQVALU pairs.

-By default, MXG Sorts invoke the NODUP option to remove

all duplicates, but duplicate log events occur (refresh

same page in same second), and to allow you to override

and keep duplicates, macro variable &SASNODUP is defined

in VMXGINIT (Default is NODUP) and it is used in VMACWWW

so you can supress duplicate removal in web logs by

using %LET SASNODUP=; in your //SYSIN
-There is a lot of logic in VMACWWW that will eventually

be documented in ADOCWWW, and there will be new exits to

allow filtering of what records are kept during INPUT, as

web log data can be voluminous. Current notes:

Variable HOST is LOWERCASED, so that all of the casing

spelling variants (WWW.MXG.COM, www.mxg.com) will be

combined.

Variable VISITOR is created from cookies if either the

SITESERVER=ID= or ASPSESSIONID cookie name (COOKNAME)

is found, to track the same user thru your servers. If

neither cookie is found, a VISITOR variable is created

as TRIM(HOST)!!TRIM(CLIENTIP)!!TRIM(USERNAME). Note the

USERNAME exists only for URLs in the authenticated

realm
For web logs with self-contained documentation of their

contents (currently, the IIS "#FIELDS" record and the

Netscape "format=" record), this new architecture reads

those header records to determine what data fields are in

your data records, and in what order, to automatically

process logs with excluded, skipped, optional fields,

and with those fields in any order, transparently!

These powerful algorithms will be easy to use to support

other new data that contain self-documenting headers.

While this support is fully usable now, I anticipate the

expansion of analysis and reporting of this clickstream

data. Unfortunately, at present, there is no way to

correlate the computer resources that are associated with

these web activities. But this is ongoing research, and

additional enhancements can be expected.

Thanks to Tom Gray, SPRINT, USA.
Change 19.248 Change 19.187 was incomplete; an ID statement referenced

ASUMNTIN non existent variables and was removed.

Dec 21, 2001

Thanks to Jim Quigley, ConEd, USA.

======Changes thru 19.247 were in MXG 19.08 dated Dec 20, 2001======
Change 19.247 Some ANALDB2R reports failed if all DB2 datasets had zero

ANALDB2R observations. The original design of %VGETOBS macro did

VGETOBS did not differentiate a non-existent dataset from one

VMXGINIT with zero observations. This enhancement to %VGETOBS

Dec 19, 2001 creates new GLOBALed macro variable VGETDSN that will be

blank if the dataset does not exist in the DDNAME, or

will contain the data set name if it exists. If the

dataset is found, an MXGNOTE with name and number of

observations is printed on the log; if not, an MXGWARN is

printed that the requested dataset was not found.

The invocations of %VGETOBS in ANALDB2R were then revised

to exploit the enhancement.

Change 19.246 Sophisticated option for VMXGSUM when two output summary

ADOCSUM datasets at different summary levels are to be created.

VMXGSUM MXG's normal HouseKeeping deletes the MXGSUMx temporary

Dec 18, 2001 datasets, to make disk space available, but with the new

HOUSKEEP=NO specified in your first %VMXGSUM invocation,

no housekeeping is done, so a second %VMXGSUM can then

specify NOSORT=YES to bypass the PROC SORT, and since the

first step (when needed) will be a VIEW, there is no I/O

associated with the output side of the first data step,

which is then read directly by the PROC MEANS.

You have to be very sure of what you are doing; if the

sort order is not correct, the second %VMXGSUM invocation

will fail. Examples of using these options are given in

member ADOCSUM.

Thanks to Chuck Hopf, MBNA, USA.
Change 19.245 Support for APAR OW49808 to RMF Type 70 Subtype 2 CRYPTO

EXTY70X2 record (new in z/OS 1.2, Change 19.176) enhances the data

EXTY70Y2 measures and there are now three datasets created from

IMAC7072 the 70-2 record:

VMAC7072 DDDDDD Dataset Description

VMXGINIT TY7002 TYPE7002 RMF PCI CRYPTO COPROCESSOR PCICC

Dec 18, 2001 TY70X2 TYPE70X2 RMF PCI CRYPTO ACCELERATOR PCICA

TY70Y2 TYPE70Y2 RMF CRYPTO COPROCESSOR FACILITY CCF

These data will become important as they will be used by

the Secure Sockets Layer associated with Websphere and

similar secured systems.

These data have not yet been validated with test records.

Change 19.244 Support for IDMS Log Statistics Records decodes LGRTYPE

EXIDML01- 'F6'x and '76'x record's into eleven datasets, one for

EXIDML11 each STLTYPE (01x-11x), but only the STLTYPE=02 TASK

FORMATS records are fully decoded (in dataset IDMLOG02). Records

TYPEIDML with STLTYPE=01 (SYSTEM) have three non-zero fields, but

VMACIDML do not match their DSECT descriptions, and all other STL

VMXGINIT TYPEs in the test file have no data fields.

Thanks to Hugh O'Neill, Dow Jones, USA.

Change 19.244 This Windows Disk Space Used utility read the output of

UDSKCONT DIR commands to report disk space by direcory and file.

Dec 13, 2001 It now supports Windows NT/2000/XP, which has a different

output format for the DIR command, but you have to tell

it which format you want to read. I use this to find out

why I've run out of space on a PC disk.

Change 19.243 IBM declared field QBACSWU to be 'Reserved' in DB2 6.1,

VMACDB2 but the field appears to continue to contain QBACSWU, so

Dec 13, 2001 the QBACRSVD field is now input into QBACSWU variable,

but please validate and use it with caution.

Thanks to Charles Harbour, John Deere, USA.
Change 19.242 For users of TYPEIMS7 and VMACIMS, variables IMSGTEXT and

VMACIMS OMSGTEXT in IMS01/IMS03 datasets were blank with IMS 6.1

Dec 12, 2001 and wrong with 5.1 that are now correctly input in this

almost-archaic member. Some unnecessary/confusing tests

IF _IMSVERS were removed for readability. Several new

variables were INPUT but not kept; MSGRACUS (RACF User

ID, also "LogonID"), MSGSAFNM, MSGUSIDI, and MSGWLMSC are

now kept in the 01/03 datasets.

The recommended MXG IMS Log processing is in the members

JCLIMSLn, ASMIMSLn, and TYPEIMSA/TYPEIMSB, and they are

not impacted by this change. This site had reports that

were based on the raw IMSnn datasets created by VMACIMS,

so it was updated to preserve their reports.

Thanks to James Seitz, Oklahoma Department of Human Services, USA.

Thanks to Ken Sharpe, Oklahoma Department of Human Services, USA.
Change 19.241 MXG enhancement for Batch Pipes adds these variables

VMAC91 DDNAME JESNR JOB READTIME STEPNAME STEPNR to the

Dec 11, 2001 TYPE91nn datasets for subtypes 11,12,13, and 15, and

increases the number of system names/flags variables

SMF91XYn and SMF91XFn from eight to sixteen to support

sixteen systems in a sysplex.

Thanks to Michael Oujesky, MBNA, USA.
Change 19.240 MXG 19.04-19.07 only. Variables PCHANBY and PNCHANBY can

VMAC73 be always zero in error. Change 19.176 mis-located the

Dec 10, 2001 initialization of those variables for SMF73CMG=0 records.

Thanks to John Gebert, Office Depot, USA.

Change 19.239 Several Coupling Facility variables in TYPE74CF/TYPE74ST:

VMAC74 R744SASQ/R744SSSQ/R744SQSQ/R744SDSQ/R744FCSQ/R744FSQU

Dec 6, 2001 were all sums of squares that are now divided by 1000000.

Variables R744FTIM/R744FCTM were incorrectly divided by

4096. Our test data had all zeros, so these incorrect

conversions were not previously caught.

Thanks to Bill Cool, EDS Federal Government, USA.
Change 19.238 Variable DEVNR is now created in dataset TYPE42VL as a

VMAC42 numeric field with format HEX4; variable SMF42DEV in that

Dec 6, 2001 dataset is a 4-byte character device number in hex, but

Dec 18, 2001 DEVNR is numeric in all other datasets, and is needed to

merge TYPE42VL with other data.

Dec 18: SMF42DEV input changed to $CHAR and logic to

INPUT the DEVNR variable was revised to work.

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

Change 19.237 Variable LOCLINFO in PDB.STEPS was incorrect. Its value

BUILD005 in a step record was being overlaid with LOCLINFO from

BUIL3005 the other job-level records, so if you put different data

Dec 4, 2001 in the STEP records, it was lost. Now, the step value

is preserved for each step.

Thanks to Diane Eppestine, Southwestern Bell, USA

Change 19.236 Values for SMF73CPD (Channel Path Description), decoded

FORMATS by MXG Format MG073CD, were updated for new '14'x,'15'x,

Nov 26, 2001 values for HSSI and Ethernet Open System Adapter Channel.
Change 19.235 Observations were created in new dataset MQMCFMGR only by

VMAC115 accident; the test IF SM1152O2 GT 0 THEN DO; should have

Nov 26, 2001 been coded as IF SM1152OC GT 0 THEN DO;

Thanks to Martin Lee, Safeway Stores PLC, ENGLAND.

Change 19.234 MXG Software (all versions) supports the euro symbol, as

none there are no MXG variables that contain financial values

Nov 23, 2001 in euros, dollars, or any other units of currency.

Of course, you can always create financial variables

in your reporting and billing programs, and there, you

can use the SAS FORMAT EUROw.d, which complies with

the European Monitary Union rules and displays an "E"

as the euro symbol, without any special characters.

But be careful with EUROw.d (and DOLLARw.d) formats;

you must not underspecify the field width w, or SAS

will truncate from the left and your euro/dollar sign

will not be printed (as shown in SAS documentation!):

With a value of x= E1254.71 euros

using format will print as

EURO6.0 E1,255

EURO5.0 1,255

EURO5.1 1255

EURO6.1 1254.7

Thanks to K. Strange Bruun, IBM Denmark A/S, DENMARK.
Change 19.233 For NT Trending, default DDNAMES could not be changed, as

BLDNTPDB the user tailoring could be overwritten by TRND defaults.

TRNDNTIN Now, user tailoring of DDnames works as expected.

TRNDNTLD

Nov 23, 2001

Thanks to Greg Jackson, National Life Insurance of Vermont, USA.

Change 19.232 The new IHDRTMO2 exit that was supposedly added by Change

IHDRTMO2 Change 19.154 was created but not included in VMACTMO2.

VMACTIMS This change adds the %INCLUDE of IHDRTMO2 in VMACTMO2; it

VMACTMO2 is called after the Transaction Header or Interval Header

VMACTMV2 has been INPUT. Comments in IHDRTMO2 were updated and

Nov 20, 2001 now include an examples of tailoring, either by EDIT of

Nov 23, 2001 the IHDRTMO2 member, or by using the %LET MACTMOH= in

your //SYSIN input.

-Similarly, the INCLUDE of IHDRTIMS and IHDRTMV2 have been

added to their appropriate VMAC member.

Thanks to Jerry Urbaniak, Acxiom CDC, USA.
Change 19.231 Sample MQ Series reports were updated. Originally, the

ANAL115 ANAL115 member contained an MXG "program" that was read

Nov 20, 2001 with a %INCLUDE, and then executed with the syntax:

%INCLUDE SOURCLIB(ANAL115);

Now, the ANAL115 member defines %MACRO ANAL115;, so your

%INCLUDE statement is now replaced with:

%ANAL115;

to execute with all defaults, or with

%ANAL115(PDB=SMF,REPORTS=ALL);

if you want to change the input to use SMF data, select

reports, etc. See the complete list of arguments in the

comments in the ANAL115 member. The member was changed

into a %MACRO so that we could provide you with arguments

to tailor the execution and provide report selection:

- The %INCLUDE(ANAL115); statement now would not fail,

but it would only read the member and compile the

%ANAL115 macro, but it would not "do" anything, since

it does not invoke the actual %MACRO name.

- You don't need the %INCLUDE(ANAL115); statement now,

because when SAS sees that %ANAL115; statement in your

//SYSIN (because MXG set options SASAUTOS=SOURCLIB and

MAUTOSOURCE in its CONFIGV8/AUTOEXEC), SAS will read,

compile, and execute the %MACRO defined in ANAL115.

There are extensive IBM notes about measuring/monitoring

MQ series in the comments of this report member.

Thanks to Bruce Widlund, Merrill Consultants, USA.

Change 19.230 -Using %LET MACKEEP= to tailor the DCOLLECT part of the

DAILYDSN DAILYDSN program (used in the JCLDAYDS example to read

VMXGINIT DCOLLECT and TMS data for "Daily Dataset Accounting")

Nov 19, 2001 did not work. The second execution of %VMXGINIT inside

DAILYDSN was re-executing the %LET MACKEEP=; statement to

set the MACKEEP macro variable back to blank. Further

examination of the %GLOBAL statement documentation showed

A macro variable created with a %GLOBAL statement has

null value until you assign it some other value. If a

global macro variable already exists and you specify

that variable in a %GLOBAL statement, the existing

value remains unchanged.

so there is no need to initialize macro variables to null

value if the variables is in a %GLOBAL statement, and the

removal of those %LET xxxxxx=.; (null value) statements

permits multiple execution of %VMXGINIT.

-However, the purpose of the user's MACKEEP= tailoring was

to make DAILYDSN only create the four DCOLLECT datasets

that are used in DAILYDSN, so that member now has an

example of a %LET MACKEEP= statement that only keeps the

four DCOLxxxx dataset that are used by DAILYDSN:

DCOLDSET DCOLVOLS DCOLMIGS DCOLBKUP

Thanks to Joseph Stoll, University of Wisconsin Hospital, USA.
Change 19.229 Support for RSD Folders ASTRE Auditing User SMF Record

EXRSDAUD creates new RSDAUDIT dataset with member TYPERSDA, and

FORMATS new formats MGRSDAU and $MGRSDFO. One set of fields,

IMACRSDA Folder Type is not documented, but will eventually be

TYPERSDA decoded with another format.

TYPSRSDA Note that the other RSD Folder SMF records are supported

VMACRSDA in TYPERSDF.

VMXGINIT The tested version is RSD/Folders V 4.1.

Nov 17, 2001

Thanks to Christa Neven, KBC Bank, BELGIUM.

Change 19.228 MXG 19.01-19.07 only, execution under ITSV, received this

VMACSYNC ERROR: VARIABLE UNIT1 HAS BEEN DEFINED AS BOTH CHARACTER

Nov 16, 2001 AND NUMERIC, because MXG 19.01 unintentionally changed

those variables from CHAR to NUM, but ITSV's dictionary

(righteously!) expected them to still be character.

This change creates variables UNIT0-UNIT99 as CHAR $4

variables, as they were in MXG 18.18, but it also creates

new numeric variables DEVNR0-DEVNR99, that should be used

instead of UNITn variables, so that your devices are in

numeric order: 181x..18Ax..A18x..B41x... If you use the

character unit address, you get A81x..B41x..18Ax..181x..

Variables DEVNR0-DEVNR99 are FORMATed HEX4, so they

print the same value as the UNITn variables.

Thanks to Bob Funk, Progressive Insurance, USA

Change 19.227 Fields added to HSM User SMF (usually 240/241) last year,

VMACHSM compatibly, in what were reserved bytes. These new MXG

Nov 15, 2001 variables now exist in HSMFSRST dataset:

FSRFLG3 ='FSRFLG3*REQUEST*FLAGS'

FSRTIMS2='TIME WHEN*PREPROCESSING*COMPLETED'

FSRTIMM1='TIME WHEN*DATA MOVEMENT*STARTED'

FSRTIMM2='TIME WHEN*DATA MOVEMENT*COMPLETED'

FSRTIME1='TIME WHEN*HOST PROCESSNG*STARTED'

FSRHOST ='HOST*IDENTIFIER'

Those new timestamps for phases of HSM may be useful.

I chose to keep only the one-byte flag variable FSRFLG3,

to save DASD space, but these tests can be used for the

bit-level-field-names, some of which may be important:

FSRFLG3='1.......'B FSRFVINI RECOVERY REQUEST

FSRFLG3='.1......'B FSRVXPL1 DS EXPIRED FROM ML1

FSRFLG3='..1.....'B FSRFXPL2 DS EXPIRED FROM ML2

FSRFLG3='...1....'B FSRFEXBV BACKUP VER BEING EXPIRED

FSRFLG3='....1...'B FSRFBKTP BACKUP VER DELTD IS TAPE

FSRFLG3='.....1..'B FSRFEXDT DELETED BY EXPDT/MGTCL

FSRFLG3='......1.'B FSRRECON MIGRATE DUE TO RECONNECT

Thanks to Judith Bruner, Wal-Mart, USA.
Change 19.226 There was a missing end-of-comment */ in this member.

ASUMTCPT

Nov 15, 2001

Thanks to Dan Riggs, Wachovia, USA.

Change 19.225 Support for Landmark's The Monitor for TCP/IP v1.1 has

EXTMTCIA been tested with actual data for all subtypes! These new

EXTMTCIC datasets are created from the dumped TMON records:

EXTMTCIE dddddd dataset description

EXTMTCIF TMTCIA TMTCIA TMON TCP/IP IA API TERM

EXTMTCIG TMTCIC TMTCIC TMON TCP/IP IC TCP GROUP

EXTMTCIL TMTCIE TMTCIE TMON TCP/IP IE EXCEPTION

EXTMTCIM TMTCIF TMTCIF TMON TCP/IP IF INTERFACE/ICMP

EXTMTCMO TMTCIG TMTCIG TMON TCP/IP IG PING RESULTS