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

VMAC71-VMAC79 as PIB1 instead of PIB2.

Feb 28, 1994

Thanks to Scott Ashby, Wachovia Operational Services Corp., USA.
Change 11.323 MXG 11.09-11.10 only. Change 11.246 added support for

IMAC28 NPM 2.1.0, but in IMAC28, macro _L028NWD should have

VMAC28 spelled its dataset name as NPMNWDWD instead of NPMNWCWD.

Feb 28, 1994 In addition, causing confusion but no execution error,

comments in VMAC28 were misspelled; NPMCLLAN should be

NPMCMLAN, and RMSTR should be RMSTA in all occurrences.

Thanks to Ann Wheeler, American President Lines, USA.
Change 11.322 TYPE72MN variables WSETFIX and WSETASM were incorrect.

VMAC7072 They are now calculated as:

Feb 18, 1994 WSETFIX=FRAMEFIX/AVGUSER;

WSETASM=FRAMEASM/AVGUSER;

Thanks to Jan van Kemenade, Universitair Centrum Info., NETHERLANDS.
Change 11.321 Support for ISOGON's SoftAudit Product Usage File and

EXSFTAM Module Usage File creates two datasets by reading the two

EXSFTAP separate SoftAudit flat files:

IMACSFTA MXG Dataset DDname Description

TYPESFTA SOFTAUDM XPUSAGEM Module Usage File

VMACSFTA SOFTAUDP XPUSAGEP Product Usage File

Feb 17, 1994 Both files will be read if data exists in either DDname:

// EXEC MXGSAS,USER=PDBSFTA

//XPUSAGEM DD DSN=MODULE.USAGE.FILE,DISP=SHR

//XPUSAGEP DD DSN=PRODUCT.USAGE.FILE,DISP=SHR

//PDBSFTA DD DSN=WHERE.YOU.WANT.OUTPUT,DISP=(,CATLG),

// UNIT=SYSDA,SPACE=(CYL,(10,10))

%INCLUDE SOURCLIB(TYPESFTA);

If you only want to process one of the two files, use

DD DUMMY in your JCL for the unwanted file.
Change 11.320 APPC tasks do not go thru JES, so there will be no type

BUILDPDB 26 purge record to match up with APPC type 30 records. As

BUILD006 a result, all APPC work would be held in the SPIN library

Feb 16, 1994 until SPINCNT in IMACSPIN is exceeded. Even if there are

type 6 records for APPC tasks, there is no way to know

they exist, so there is no reason to SPIN APPC tasks, and

therefore, the BUILDPDB logic for outputting APPC tasks

to the PDB.JOBS and PDB.STEPS datasets was revised to

send APPC tasks to the PDB as soon as both a type 30

subtype 4 (step) AND type 30 subtype 5 (job) record have

been found. If there were any type 6 records for the

same JOB JESNR READTIME combination in today's SMF data,

then PDB.PRINT will have the APPC tasks print data with

accounting fields from the type 30. An isolated type 6

record for an APPC task will be output when found, i.e.,

it will not be sent to the SPIN library. The insert:

ELSE IF TYPETASK=:'A' THEN DO;

IF IN30_5 THEN OKFLAG=1;

ELSE IF IN6 AND NOT IN30_4 THEN OKFLAG=1;

END;

Thanks to Don Friesen, B.C. Systems, CANADA.
Change 11.319 AS/400 variable AS400SYN was blank, because the %MACRO

VMACQAPM variable &AS400SY was not passed correctly, and was being

Feb 16, 1994 reinitialized to blanks in QAPMCONF. The three SYMPUTS

in _CQAPCON were enclosed in IF _N_=1 THEN DO; ... END;.

All occurrences of AS400SYN="&AS400SY"; were changed to

read AS400SYN=SYMGET('AS400SY'); Also, the single

occurrence of NRCPUS="&AS400CP"; was changed to read

NRCPUS=SYMGET('AS400CP');

Thanks to Greg Scriba, Budget Rent-A-Car, USA.
Change 11.318 CICS Statistics variable A20E1HWM (Peak Contention Users)

VMAC110 was left out of the KEEP= list for dataset CICCONMR.

Feb 14, 1994

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

Change 11.317 Partial support for LEGENT's MIM Release 4.0 enhances the

VMACMIM MIMTAPE dataset for the new release, but additional work

Feb 14, 1994 is needed to decode new subtypes in this release. This

code is functional, and hence included, but test data for

other subtype is needed before full support is provided.

Thanks to Doug Drain, National City Bank, USA.

==Changes thru 11.316 were include in MXG PreRelease 11.10 dtd 14Feb94==
Change 11.316 Support for LEGENT's NETSPY Release 4.4.

EXNSPYET -Dataset NSPYAPPL, new variables APOTLN62, APOTNN62 count

EXNSPYFR the outbound bytes for LU 6.2 and non-LU 6.2 sessions.

FORMATS -Dataset NSPYLU, new variable RESPNOTC='Y' if response

IMACNSPY time is collected. IF RESPNOTC='N', then variables:

VMACNSPY LRSPHOST LRSPNET WRSPHOST WRSPNET CRSPHOST CRSPNET

Feb 13, 1994 NETRSPNO T1RSPNO T2RSPNO T3RSPNO T4RSPNO

are now set to zero by MXG, as LEGENT says "these fields

may have data in them, however they should not be used

when reporting".

-New Dataset NSPYFRLY for Frame Relay Statistics

-New Dataset NSPYETHR for Ethernet Statistics.

-Several new values for NSPNSUBT are now decoded by

MXG format MGNSPEL, and logic for recognition of which

sub-subtype of the type 'N' was clarified.
Change 11.315 Support for CA's ACF2 Release 6.0 and 6.1 added seventeen

VMACACF2 new variables to type 'V' record, and one to type 'D'.

Feb 12, 1994 Renames of ASSSPCOD to ASSPPCOD and ACFGFOE to ACFGFDEN

correct my misspellings. The use of LENGTH-COL-1 (to know

how many bytes are left in the record) should have been

LENGTH-COL+1 (this could have caused new variables to not

be read in, although no one seems to have noticed!).

In this revision, I have also decoded the LIDREC and the

LIDXARE of the type 'J' record, labelled and formatted

the several dozen new variables, but did not add any of

those variables to the ACF2JR data set; instead, if you

decide you need those variables, you can use the MXG

macro _KACFJR in member IMACACF2 to add them. (Only one

site had requested the LID fields.)

Change 11.314 For developers, this is my recommended test protocol:

Testing

Feb 12, 1994

You need to ALWAYS test it ONE MORE TIME!

When you think your code is done:
Run it once more, as a batch job, and while that job

that job is running, use SPF 3.12 COMPARE to examine

every difference between the before-and-after source

members.
Then,examine the batch job's output:

- The SYStemLOG, for any Operating System warnings,

- The SASPRINT report output, for any differences,

- The SASLOG log output, for any occurrence of each

of these strings (blanks are important!):

"ERROR:" "ERROR :" " UNINIT"

"NEVER BEEN" "NOT FOUND" "CONVERT"

"NOT CATLGD" " NOT " "TRUNCATED"
Change 11.314A MXG 11.09A Only, PMACC02 Report, DB2 Accounting Detail

ANALDB2R Trace was in error due to insufficient testing. The data

Feb 12, 1994 was summarized when it should only have been sorted.

Thanks to Jeff Marsh, Twentieth Century Services, USA.

Change 11.313 OMEGAMON for CICS V550/V551 User SMF record subtype 100

VMACOMCI sub-subtype 2 caused INPUT STATEMENT EXCEEDED RECORD LEN.

Feb 12, 1994 The code expected the same number of segments for EGROUP

as for ETRNAME, but they are unrelated tables. The code

was corrected, but for simplicity both EGROUP and ETRNAME

segments are still output in dataset OMCITRAN; you can

identify which is which by testing :

IF ETRGRPM NE . ==> ETRNAME and ETRGRPM are valid

IF ETRGRPM EQ . ==> EGROUP, EGRPNAM, EGRPESNR are valid

This change supercedes Change 11.115.

Thanks to Ron BLeeden, Jewel Food Stores, USA.

Thanks to Bill Wieland, EDS Westlake, USA.

Change 11.312 Variable AVGENQMS (Average ENQUE time in milliseconds)

VMAC74 was calculated but not KEPT, LABELed nor FORMATted, but

Feb 9, 1994 now it is.

Thanks to Waldemar Schneider, SAS Institute Europe, GERMANY.

Change 11.311 GRAFWORK was enhanced with a new graph of memory usage by

GRAFWORK workload (using the ACTFRMTM-based measure of resident

Feb 7, 1994 memory frame seconds in the xxxxMEMR variables). GRAFWORK

now provides graphic depiction of CPU, I/O, and MEMORY

resource usage by workload.
Change 11.310 GRAFRMFI was revised to include new variables added to

GRAFRMFI RMFINTRV recently, and the internal logic revised to make

Feb 7, 1994 maintenance easier.
Change 11.309 This major revision to VMXGSUM reduces runtime and CPU

VMXGSUM time, by keeping only the variables and datasets that are

ANALDB2R needed during the input for summarization, and executes

ANALPRTR only the steps required (i.e., it will bypass PROC SORT

ASUMCICS if it can). If you have tailored some MXG members that

ASUMDBDS invoke %VMXGSUM, you MUST examine the INCOMPATIBILITY

ASUMDB2A note, below, and you MAY have to update your tailored MXG

ASUMDOS members. If you have used %VMXGSUM in your own reporting

ASUMHPCS programs, you may also be vulnerable to required changes.

ASUMJOBS -Changed the logic for MINTIME= and MAXTIME=. No longer

ASUMTMNT are variables named MINTIME/MAXTIME created; instead, the

ASUMVDEV variable name(s) passed into VMXGSUM are retained, which

ASUMVMON permits bypassing the first data step to reduce costs.

ASUM70PR -NOSORT= parameter was added, which allows the sort to be

DAILYDSN bypassed if you KNOW the data is already in order.

GRAFDB2 -Initialization to protect for "UNINITIALIZED VARIABLE"

GRAFLPAR message (the series of IF X=. THEN X=.; statements) was

TRNDDB2A relocated to execute only once.

TRNDDB2S -New logic automatically figures out what variables need

Feb 7, 1994 to be kept (by looking at all variables that are in any

text of of the list-of-variable parameters), so the _KMXGSUM

change was syntax that was added in MXG 11.09 is no longer used.

revised -The new KEEPIN= parameter is required if you have INCODE=

Mar 6, 1994 specified, and if there are unique variables used in your

INCODE= logic that are not referenced by other VMXGSUM

list-of-variable-parameters (SUM= SUMBY= MIN= ... etc.)

-INCOMPATIBILITY NOTE

If there is an INCODE= parameter (for sophisticated use

you can insert SAS code with this parameter), AND only

if there are variables referenced in your INCODE= logic

that are not referenced by the other VMXGSUM parameters

(SUM= SUMBY= MIN= ... etc.),

Then you MUST add a KEEPIN= parameter to your %VMXGSUM

invocation so that those unique variables exist during

the INCODE= code execution.
These MXG-supplied members had to be changed in the MXG

Source Library because all had un-kept variables that

had to be listed in the KEEPIN= parameter:

ASUMDBDS ASUMDB2A ASUMDOS ASUMHPCS ASUM70PR

DAILYDSN GRAFDB2 GRAFLPAR TRNDDB2A

If any of those 9 members are in your USERID.SOURCLIB

tailoring library, you MUST refit your changes,

starting with the MXG 11.11 member that contains the

required KEEPIN= parameter.
You must scan your USERID.SOURCLIB(s) for any of your

own programs that invoke %VMXGSUM, and see if any of

them meet both conditions (INCODE= and nonkept variable

referenced in that INCODE= logic).

Examine the SAS log of your test runs for UNINITIALIZED

VARIABLES messages; that is a sure sign that you have

INCODE= variables that do not exist!
DO NOT OVERLOOK THIS CRITICAL INCOMPATIBILITY, which

should affect only the very small number of sites that

have tailored the MXG summarization or trending code.

As long as you are executing those 9 MXG members

unmodified from the MXG 11.11 library, there is not any

incompatibility with this change.

-All MXG invocations of VMXGSUM were examined and INVOKEBY

was added so the caller would print on the SAS log.

-Especially with a large input dataset (eg., CICSTRAN with

594,000 observations, the keeping of only the needed

variables significantly saves resources. Using:

%VMXGSUM(INDATA=CICSTRAN.CICSTRAN,

OUTDATA=CICSSUM,

DATETIME=STRTTIME,

INTERVAL=HOUR,

SUMBY=APPLID DATETIME,

FREQ=NUMTRANS,

SUM=IRESPTM TASCPUTM);

Showed these comparisons before and after this change:
Run COMPRESS CPU EXCP Memory DASD

Option SEC count used tracks

Before NO 120 21843 4973K 10699

After NO 83 12920 4902K 943

Before YES 650 25330 4913K 7129

After YES 135 13255 4902K 1111

The elapsed times were also significantly reduced. The

37 minutes required for the "before" compressed run was

reduced to 11 minutes for the "after" compressed run.

Thanks to Chuck Hopf, Primerica, USA.

Change 11.308 Support for Candle's Deltamon SMF record creates new MXG

EXTYDLMN dataset TYPEDLMN, which reports activity (ADD,UPDATE,

FORMATS DELETE, or RENAME) at the PDS member level.

IMACDLMN

TYPEDLMN

VMACDLMN

Feb 7, 1994

Thanks to Chuck Hopf, Primerica, USA.

Change 11.307 ASTEX variable RCHCNT=SUM(RHTRD,RHTRDS,RHTDFW,RHTCSW) was

VMACDMON added after the INPUT of the four arguments. In prior

Feb 7, 1994 versions of ASTEX, RCHCNT was directly input, but now it

must be calculated from its four arguments.

Thanks to Jay Stewart, Honda, USA.
Change 11.306 For MVS execution, MEMSIZE=32MB is now the default value

CONFIG in CONFIG. The default BUILDPDB failed "OUT OF MEMORY"

Feb 7, 1994 with MEMSIZE=24MB, and 32MB protects for the future.
Change 11.305 Variable CMF09UIC from Boole's CMF must be divided by the

VMACCMF number of samples:

Feb 7, 1994 IF CMFHDSAM GT 0 THEN CMF09UIC=CMF09UIC/CMFHDSAM;

Variable CMF05NUM is now format HEX4.; variable CMF19TPG

no longer formatted HEX2.!

Thanks to Joanne Turpie, Department of Labour, NEW ZEALAND.

Change 11.304 Support for OPC/ESA Release 2.1 has added 5 new datasets

EXOPC34 (and corrected INVALID MTD SUBTYPE messages):

EXOPC35 OPC24_H MTD Delete/Change CM

EXOPC36 OPC24_I MTD Hold Operations

IMACOPC OPC34 Catalog Management

VMACOPC OPC35 Backup Event

Feb 6, 1994 OPC36 CP Backup Log

Finding all the IBM manuals for this release was a chore;

comments in VMACOPC identify what is documented where!

And I learned that you cannot order Licensed IBM Pubs if

you use your "IBM Customer Number", but if you give the

same number as your "IBM Enterprise Number", the Pubs

clerk discovers you are authorized for Licensed Pubs!

Thanks to Alan Phelan, Allied Irish Bank, IRELAND.

Thanks to Wolfgang Vierling, Vereinte Versicherungen, GERMANY.
Change 11.303 TYPE71 variable AVLEXTMN, "Minimum ESTORE Available", can

VMAC71 be negative, according to IBM, so its input was changed

Feb 5, 1994 from &PIB.4. to &IB.4. The actual value of the field is

"number of extended storage e-frames currently on the

available sets excluding those reserved for pref steal."

This means that when the value is zero, all of RSM except

for pref steal will accept that there are "no" available.

If the value is less than zero, pref steal will check to

see if there are any in reserve and will use these.

Thanks to Hr. Leineweber, HUELS AG, GERMANY.

Change 11.302 Testing of TYPEVMXA (VM/ESA) under UNIX, OS/2 or WINDOWS

AUTOEXEC found glitches (but this change is NOT needed if you run

VMACVMXA MXG under either the MVS or VM versions of SAS):

VMAC83 -AUTOEXEC had blanks inside quotes, which is no problem

XMACNCCF for WINDOWS or OS/2, but UNIX did not tolerate, so they

VMXGHSM were removed. Also, a new Filename statement was added:

ANALSNAP FILENAME INSTREAM 'C:\MXG\USERID\INSTREAM.SAS'

IMACACCT and must be suffixed .SAS, because it is written to and

SYSLOGJ3 then %INCLUDEd (to build formats "instream").

UTILXREF -Some variables input as $EBCDIC were not character data

VMACACF2 but were binary data and are now input as $CHAR. Some

VMACASXT also required addition of $HEX formats, and most were in

VMACCMA seldom-to-never-used VM data sets.

VMACDB2 -The building of the "instream" PROC FORMAT failed with

VMACHMF INVALID HEX DATA, because of $EBCDIC versus $CHAR input

VMACPOOL format that was corrected.

VMACTSOM -Discovered that EBCDIC algorithm to identify numeric

VMACVMON EBCDIC characters from alphabetic characters, using

VMACVMXA LT '0' ==> alphabetic GE '0' ==> numeric

VMACVVDS is invalid under ASCII execution. EBCDIC algorithm works

VMAC102 because EBCDIC numbers are 'F0'x thru 'F9'x, which is

VMAC24 larger than EBCDIC letters of '81'x thru 'C1'x (lower

VMAC33 case) and 'C1'x thru 'E9'x (caps), but the ASCII numbers

VMAC37 are '30'x thru '39'x, which is smaller than ASCII

VMAC4789 letters '41'x thru '5A'x (caps) and '61'x thru '7A'x

VMAC59 (lower case). The parsing algorithm was redesigned to

VMAC60 scan for non-blank versus blank, and one more difference

VMAC6156 between ASCII and EBCDIC execution is documented!

VMAC80 -Use of $VARYING200. input for printable characters was

VMAC83 resolved. $VARYING acts like $CHAR instead of $EBCDIC,

Feb 4, 1994 so strings that are to be printed need conversion. The

WINDOWS conversion algorithm is:

INPUT variable $VARYINGnn. lenvar @;

variable=INPUT(variable,$EBCDICnn.);

variable=TRANSLATE(variable,' ','80'x);

Note: only the INPUT function should be required, but

The TRANSLATE was unexpectedly needed because the

$EBCDIC was found to convert '20'x ASCII blanks at

the end of the string to '80'x. SAS Institute

suggested that using INPUTC(string,format,length);

instead of INPUT(string,format) would have eliminated

the need for the TRANSLATE() function, but my code

works, whereas (Jun 24, 1997 update:)

when I tried INPUTC under SAS 6.12 TS020 Windows 95

I got these errors pointing to the $EBCDIC128. when

I tried to use VAR=INPUTC(VAR,$EBCDIC128.,LEN);

386-185 Expecting Arithmetic Expression and

200-322 The Symbol is not recognized

underscored for format $EBCDIC128.. Problem

will be opened with SAS: That's the way it is!

The $VARYING conversion required examination of each

occurrence to see if the field is a printable (text)

string, and if so, insert the above conversion code.

(This also required creation of DO group in some cases.)

Thanks to Chris Powell, Vancouver Stock Exchange, CANADA.
Change 11.301 Labels for the four new SYNCSORT variables SYNDSMVL,

VMACSYNC SYNFMAVL, SYNFMALO, and SYNMMUSE were added; originally

Feb 2, 1994 they were not provided in SYNCSORT documentation.

Thanks to John Borland, Citibank, USA.

Change 11.300 An SMF Writer "Simulator" has been added to ANALSMF, to

ANALSMF examine your SMF data and tell you what is the optimum

Feb 2, 1994 CI Size of your VSAM data set. See the MVS Technical

Note in Newsletter TWENTY-FIVE for a complete discussion

of the impact of the wrong CI Size on the SMF Writer.

You need only to EDIT ANALSMF to set the CI Size of your

VSAM file (in macro _MYCISIZ), and then point the SMF DD

to a dumped SMF file, and the report of DASD Space and CI

Writes at different CI Sizes will be produced. Several

additional reports were added to describe how seldom the

SMF Writer actually writes, and the Statistics from

TYPE23 are also printed, along with a tabulation that

shows the contents (count and bytes) of where your SMF

data records come from.

Change 11.299 The example of the ZARA LIST command to create an OUTFILE

TYPEZARA needs to have the SSFN operand added so that "secondary"

VMACZARA files (i.e., second and subsequent files on a volume) are

Feb 2, 1994 created in the OUTFILE. Without SSFN, you only see the

first dataset on each volume in ZARADSN. Correct is:

LIST OUTFILE ACTIVE SSFN $$

In addition, variable LSTTYPE, type of tape (AUTO,ACTIVE,

SCRATCH) was added to ZARAVOLS and ZARADSN datasets.

Also, variable FILTIMEL may contain trash (seems to be

for volumes converted to ZARA from CA-1); the "??" was

added to the INPUT FILTIMEL ?? &PD.4. statement.

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

Thanks to Neil Ervin, Huntington Bank Service Company, USA.
Change 11.298 The JCL in this example to analyze all records from a

ANALALL specific job was partially for SAS 6.08 and partially for

Jan 18, 1994 SAS 5.18, causing confusion for novices. The JCL example

now is for SAS Version 6.

Thanks to Rosie Jergovic, Pacific Bell, USA.
Change 11.297 Variable TIC_UTIL in NETSPY is incorrectly computed. The

VMACNSPY BYTSENT+TRTPBYTS should be BYTSENT*TRTPBYTS in both of

Jan 18, 1994 the equations for TIC_UTIL.

Thanks to Linda Liu, Transamerica, USA.

Change 11.296 This analysis of the size of an SMF file was enhanced to

ANALSMF report SMF type 23 statistics, and the categorization of

Jan 18, 1994 SMF record IDs for JOB-IO now includes all VSAM activity.
Change 11.295 Missing value for FATNUM, UTNUM, or MRONUM cause INVALID

TYPEMON8 DO ARGUMENT message if Change 11.270 was not installed to

Jan 17, 1994 circumvent new format of Landmark data. While 11.270 did

avoid these errors, the missing value protection was now

added, just in case!

Thanks to Hr. Dungl, CA-Inform, GERMANY.

Change 11.294 MXG 11.09A only. Accounting detail report not produced

ANALDB2R if time selection was used and DB2ACCT was used as input.

Jan 13, 1994 Find "%MACRO PMACC02;", then find "MINTIME=QWACBSC;".

Insert after that line the statement "MAXTIME=QWACESC;"

Thanks to Jeff Marsh, Twentieth Century Services Inc, USA.
Change 11.293 Additional refinements in the RMF reporting from MXG data

ANALRMFR were added. The time selection logic was corrected so a

Jan 13, 1994 report for 10am-12am Mon-Fri can be requested, and CPUIDs

thru 15 CPUs is now supported.

Change 11.292 MXG 11.09A only. The KEEP= logic did not include &MEAN.

VMXGSUM The three lines starting with &MAX &MAXTIME need to have

Jan 12, 1994 &MEAN added after &MAXTIME. Only if you invoked VMXGSUM

with the _KMXGSUM operand, and had MEAN= parameter, would

this change have had any effect. Change 11.309 revised.
Change 11.291 Analysis of VSAM datasets were sorted by SMFTIME instead

ANALDSET of the STEP initiate time; logic was added to store and

Jan 11, 1994 retain the SORTTIME of each STEP record in STEPTIME, and

then store STEPTIME into SORTTIME for TYPE64 observations

so VSAM is consistent with non-VSAM sort order.

Thanks to Gary Matney, Twentieth Century Investors, USA.

==Changes thru 11.290 were included in MXG 11.09A dated Jan 10, 1994====
Change 11.290 Member ASUMAPAF, a user contribution, summarizes Amdahl's

ASUMAPAF APAF records from MDF, similar to ASUM70PR summary of

TYPEAPAF IBM's PR/SM records. TYPEAPAF was cosmetically changed

Jan 8, 1994 to use __NRLPS to control all of the LPAR variables to

be kept (previously, some variables were controlled by

__NRLPS and some were kept for all possible 15 LPARS).

The value of __NRLPS is set in IMACAPAF.