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

triplet in the header which points to all the deadlock

resource info sections

- each deadlock task has only one deadlock resource

section associated with it.

This support has been tested, but only with a small number

of test records.

Thanks to Malcolm Sare, Australian National, AUSTRALIA.
Change 10.160 UTILGETM (used in step SMFSMALL of JCLTEST6 to create the

UTILGETM SMFSMALL sample file now will select ten of each subtype

Sep 5, 1992 of all SMF records, including user records. Previously,

only types 22,30,39,70-79 subtypes were selected. This

implementation makes use of _TEMPORARY_ arrays to avoid

a limitation in SAS 6.07 of 32768 variables in a data step

(that limitation is to be fixed in the November, 1992,

SAS maintenance tape), but you may wish to note that the

variables in _TEMPORARY_ arrays are not similarly limited.

Since this type of array did not exist in SAS Version 5,

UTILGETM tests its operating environment and uses the new

(more robust) logic only under Version 6; the old logic is

unchanged under Version 5.

Thanks to Chuck Hopf, Primerica, USA.

Change 10.159 Although we don't recommend using these archaic read VTOC

FMXGUCBL programs (instead, see member ASMVTOC/VMXGVTOF), they will

VMXGVTOC now work under SAS 6.07. The EXEC ASMHCL statement must

VMXGVTOR be changed so the PARM.C parameter precedes the PARM.L

Sep 3, 1992 parameter, and the //SYSIN following that EXEC must be

changed to //C.SYSIN. Additionally, the function calls in

VMXGVTOC and VMXGVTOR using =FMXGUCBL() was changed to

=UCBL() because functions can only have 4-character names

in SAS Version 6.

Thanks to Bob Smith, Nissan Motor Corporation in the USA.

Change 10.158 FORMAT NOT FOUND error with DB2 SQL Trace reports if no

ANALDB2R T102S105 or T102S107 records (the records that map DBID &

Sep 3, 1992 OBID to their names) were found. While these records are

always written at Trace start-up, the condition can occur

when you process records in the middle of a trace session.

Also, even when the format could be constructed, if there

was no match with your OBID/DBID value, an 8-byte string

(SYSTEM ID + QWHSSSID) was printed. Now, the format is

always created, and for a no-match, the string of SYSTEM,

QWHSSSID, DBID, and OBID is printed. This required some

reformating of some of the SQL reports.
Selection by PLAN, CORRNAME, etc., has now been added to

the Record Trace report.

Regardless of the order in which reports are specified,

ANALDB2R executes and prints the reports alphabetically.

The SORTBY= parameter was ignored in subsequent reports,

if a report such as PMACC03 was requested that had a non-

changeable sort sequence, because of incorrect parsing

that has now been corrected by DB2RSORT macro changes and

invoking this macro for the reports whose sort sequence

can be changed.

Additional changes that will reduce WORK space requirement

are still in development.

Thanks to Chuck Hopf, Primerica, USA.
Change 10.157 (MXG 10.1 only). Assembler error MSGAREA not found cause

ASMVTOC LSPACE UCB=(4),EXPMSG=MSGAREA,MF=I should have been

Sep 2, 1992 LSPACE UCB=(4),MF=I

The EXPMSG=MSGAREA was left over from debugging. Sorry!

Thanks to Sean Chaffee, Amadeus Dataprocessing Gmbh & Co, GERMANY.
Change 10.156 ASMVVDS may fail with User 666 ABEND processing a VVDS on

ASMVVDS a volume which is SMS managed multivolume with guaranteed

Sep 2, 1992 space. The Diagnose VVDS command does not work until you

install IBM APAR OY48199 (PTFs UY74336-339).

Thanks to Chris Taylor, Wachovia Operations Services Corp., USA.
Change 10.155 NPM 1.5 incorrectly documented format of new variables

VMAC28 LLBSSTIM and LLBSPTIM as packed decimal (PDTIME4.) fields,

Sep 2, 1992 but actual data now shows they are PIB4. fields. Change

the PDTIME4. to PIB4. after both variables.

Thanks to Jerry Kleinheinz, Mortgage Guarantee Corp, USA.
Change 10.154 TYPEMONI (archaic Landmark Version 7) does not contain

TYPEMONI APPLID, but ASUMCICS expects that variable, so sites still

Sep 2, 1992 using TYPEMONI (remember, TYPEMON8 is for Versions 8/9)

must add APPLID to the KEEP= list for MONITASK, and must

insert APPLID=SYSID; after the @; and before FILTEM=0; .

Thanks to Russ Weid, CUNA Mutual Insurance Group, USA.

Change 10.153 Step accounting variable SACCT1 is now added to the PDB

IMACPDB datasets JOBS, STEPS, and PRINT, so that Started Tasks

Sep 1, 1992 (which can have only step accounting fields) can be

tracked to account group. This involved adding SACCT1 to

the ID statement in macro _PDBSUMS, and adding SACCT1 to

the _PDBADD2 and _PDBADD3 macros.

Thanks to Mike Skopec, Platinum Technology, USA.
Change 10.152 The //LIBRARY DD in the first step should have specified

JCLTEST6 DISP=(NEW,CATLG) instead of NEW,PASS so that this MXG

Sep 1, 1992 format library is both built AND cataloged!

Thanks to Basil Hines, Rogers Data, CANADA.

Change 10.151 Some DB2 graphs may not have been produced because the

GRAFDB2 statement IF EOF THEN CALL SYMPUT... should have been only

Sep 1, 1992 CALL SYMPUT.... (It worked in testing, because in each of

the cases, the last observation was selected, but if the

last observation was not selected, &MXGOBS was not set.)

Thanks to Andre Henry, AG 1824, GERMANY.

Change 10.150 WSF 3.3.6 causes INPUT STATEMENT EXCEEDED error because

VMACWSF the accounting extension in 3.3.6 contains only three of

Sep 1, 1992 the seven fields added in 3.4.1. To process 3.3.6 data,

you will need to find BOTH occurrences of ACCXPFMR $CHAR8.

and insert the following line after each occurrence:

@; IF LENGTH-COL+1 GE 148 THEN INPUT

so that those following fields are only input if the data

is in the record. There is no error with 3.4.1 data.

Thanks to Frank Sullivan, Woolworth plc, ENGLAND.
Change 10.149 DB2 variables CORRNAME and CORRNUM are determined by the

IMACDB2 connection type (TSO, IMS, CICS, etc.), but prior to DB2

Sep 1, 1992 2.3, the only way to identify connection was by testing

the value of QWHCCN for you site's IMS jobname in macro

_DB2CORR defined in IMACDB2. This macro has now been

redesigned to take advantage of new (in DB2 2.3) variable

QWHCATYP which explicitly identifies the connection type.

(See member ADOCDB2 for all possible values of QWHCATYP.)

The macro definition now reads:

MACRO _DB2CORR

IF (QWHSRELN GE 2.3 AND (QWHCATYP=5 OR QWHCATYP=6 OR QWHCATYP=9))

OR

(QWHSRELN LE 2.2 AND (QWHCCN =: 'IMS1' OR QWHCCN =: 'IMS2'))

THEN DO; /* IDENTIFY IMS CONNECTIONS */

CORRNAME = SUBSTR(QWHCCV,5,8);

CORRNUM = SUBSTR(QWHCCV,1,4);

END;

ELSE

IF (QWHSRELN GE 2.3 AND QWHCATYP=4)

OR

(QWHSRELN LE 2.2 AND (QWHCCN =: 'CICS' OR QWHCCN =: 'PROD'))

THEN DO; /* IDENTIFY CICS CONNECTIONS */

CORRNAME = SUBSTR(QWHCCV,5,4);

CORRNUM = SUBSTR(QWHCCV,1,4);

END;

ELSE DO; /* FOR ALL OTHER CONNECTIONS */

CORRNAME = SUBSTR(QWHCCV,1,8);

CORRNUM = ' ';

END;

%

Thanks to Mr. M. Praet, AG, GERMANY.

Change 10.148 Three DCOLLECT variables in the DCOLBKUP (backup) dataset

VMACDCOL have incorrect values. UBALLSP, UBUSESP, and UBRECSP were

Sep 1, 1992 1024th of what they should have been. Find the line

UBALLSP=1024*UMALLSP;

and the two lines following. Change the "UM" in all three

lines to "UB" to correctly convert the variables.

Thanks to John Ellis, Commercial Union, U.K.
Change 10.147 The MXTDELAY dataset built by these sample CICS reporting

ANALCICS programs is not used by any report (it was a holdover from

ANALMONI CICS 1.6), and the dataset is no longer created.

Aug 31, 1992

Thanks to Kevin Batten, Roses Stores, Inc, USA.
Change 10.146 The new Appended IMS log processing program ASMIMSLG now

ASMIMSLG works with IMS 2.2 as well as IMS 3.1. Sites still using

Aug 31, 1992 IMS 2.2 need only to comment out a single statement:

MVC ORGENT(8),MSGRACGP

(because MSGRACGP is a field name that does not exist in

the IMS 2.2 DSECT as it was added in IMS 3.1), and then

use the IMS 2.2 macro libraries for the assembly. Note

that you will need to have two load libraries, one for

the IMS 2.2 version of MXGIMSLG, and one for the IMS 3.3

version if MXGIMSLG, to keep them separate. The comments

in ASMIMSLG were revised to include IMS 2.2 installation.

Thanks to Warren Hayward, TJX, USA.

Change 10.145 Landmark Monitor for CICS variable TAMRCNT is now INPUT

TYPEMON8 as PIB4. instead of PIB4.6.

Aug 18, 1992

Thanks to David Taylor, Owens & Minor, USA.

Change 10.144 NETSPY NSPYREC=N records with NCPELTYP=41 (29X) cause an

VMACNSPY INPUT STATEMENT EXCEEDED RECORD LENGTH error. Find the

Aug 18, 1992 line with NSPNEWVC PIB4. (note this is the line with

PIB4., not the earlier line with PIB2.), and delete it.

Thanks to Aubrey Tang, Government Insurance Office, AUSTRALIA.
Change 10.143 Amdahl APAF support did not keep variable BALANCE in the

VMACAPAF APAFDOMA data set, but now it does. Note that you specify

Aug 18, 1992 values 0 to 9 or A on the hardware screen, but the value

in the SMF record is 0 thru 10!

Thanks to George Scott, Rockwell International Corporation, USA.
Change 10.142 The "Appended" IMS log processing introduced in MXG 10.1

ASMIMSLG is almost perfect, but even perfection can be improved.

TYPEIMSA There were actually two errors in that release: LOGONID

Aug 12, 1992 was sometimes blank, and STRTTIME/ENDTIME were not reset

to GUTIME/OENQTIME, causing INPQUETM/SERVICTM/RESPNSTM

to be missing/incorrect. LOGONID was an omission in the

ASMIMSLG programs that create MXGIMSLG, and the time value

error was a correction in the back end of TYPEIMSA.

In addition, the Fast Path processing (IMS log 059x) that

was in TYPEIMS (the "old" way) has now been added to both

ASMIMSLG and TYPEIMSA. Moreover, Ashland's six-months use

of their contribution caused them to add enhancements to

the dynamic table allocation algorithms that are described

in comments in ASMIMSLG (and new, cleaner messages are now

printed on the SYSOUT for MXGIMSLG execution).

Thanks to J. Cary Jones, Philip Morris, USA.

Thanks to Pete Shepard, Ashland Oil, USA.

Thanks to Jeffrey S. Crum, Ashland Oil, USA.

Change 10.141 NPM changes in MXG 10.1-Change 10.106- cause INVALID DATA

VMAC28 FOR NTMPDUTH because PIB4 was mistyped instead of PIB4.

Aug 12, 1992 Change both occurrences of 'PIB4 ' to 'PIB4.'. Once this

was corrected, the values for NTNPDUTH/NTNOCCTH were noted

to be '7FFFFFFE'X, because NPM folks use that hex value as

their "infinity" (the real maximum is '7FFFFFFF'X since

the first bit is the sign bit). However, since MXG stored

these variables in LENGTH DEFAULT=4 variables, and since

the maximum hex value storable in four bytes is '7FFFFF00'

MXG was truncating the "infinity" from 2,147,483,646 to

2,147,483,392. To correct this truncation of infinity,

LENGTH NTNPDUTH NTNOCCTH 5; was added. These NTN fields

are PIUs (PDUs) or bytes (Octets) threshold values at

which NPM is to cut interval records. By specifying the

infinity value, the site is telling NPM to never write an

interval record based on these counters, and instead to

write timed interval records.

Thanks to Scott Ashby, U.S. Customs Data Center, USA.

Change 10.140 An interesting discovery was made by comparing the pagein

RMFINTRV counts in TYPE71 with the sum of the page in counts in the

Aug 7, 1992 TYPE72 observations (SET TYPE72; IF PERFRPGN=.; , to keep

only control performance group data). The TYPE71 PVTNPIN

was 46,912 page ins, while the sum of the TYPE72 PAGEINS

was 45,838, which means that the delta, 1074, page ins

were not counted in the address space data. This may be

a coding error in RMF code, or could be an indicator of a

special category of paging. IBM can't explain it, yet.

Thanks to Sandi Whitmyer, USA Funds, USA.

Change 10.139 The PRUWTR package, from Prudential Insurance, is used to

TYPE6 manage spool printing, and writes an SMF Type 6 record.

Aug 7, 1992 The READTIME value put in that record is incorrect; it is

taken from JCTCNVON, the converter on event, instead of

being taken from JCTRDRON, the correct value of when the

job's JOB card was recognized. Because the wrong time was

in READTIME, these type 6 records did not match up to any

of the other job records (BY READTIME JOB JESNR), and the

PDB.PRINT observations from PRUWTR did not have ACCOUNTn

values, and the PDB.JOBS observations did not include the

PRUWTR lines. Since PRUWTR is also source code, your JES

person can change the value from JCTCNVON to JCTRDRON, and

then the PRUWTR type 6 records will match other records.

Thanks to Norm Lacoe, Confederation Life, CANADA.

Thanks to Beth Wells, Confederation Life, CANADA.
Change 10.138 ROSCOE creates records for two new "ROSCOE monitors". The

VMACROSC JCL-CHECKER (MONITOR JCK) is identical with the JCL RECORD

Aug 6, 1992 and JCK records are added to the ROSCOJCL dataset. The

Documentview creates observations in ROSCOVPE that can be

recognized as documentview because VPEMONNM='DOC'.

Thanks to Norbert Korsche, OMV-AG, AUSTRIA.

Change 10.137 MVS/ESA XCF Type 74 record INPUT STATEMENT EXCEEDED

VMAC74 RECORD LENGTH and STOPOVER ABEND, because of an MXG error.

XMAC74 Find INPUT @OFF742MO, and change the immediately following

Aug 4, 1992 statement from DO _I_=1 TO NRDDSS; to DO _I_=1 TO NR742MO;

(make sure you change only this DO statement; there are

two other occurrences of DOs to NRDDSS that are correct!)

Thanks to John Nicholls, Health Insurance Commission, AUSTRALIA.

Change 10.136 MXG 10.1 Tape Mount Monitor ABENDs with S 55F at startup.

ASMTMNT Code added in 10.1 to support MVS/ESA 4.2 dynamic devices

Aug 4, 1992 perturbed the contents of Register 1. When the SYSEVENT

TRANSWAP was issued, the non-zero value in R1 is taken as

the address of an ECB to wait upon, but as it was now not

in the address space of the MXGTMNT, the ABEND occurred.

The fix is simple. Insert SR R1,R1 immediately

prior to the SYSEVENT TRANSWAP statement to zero Register

One, which tells TRANSWAP not to post any ECB address, and

thereby avoids the abend. This error was not in MXG 9.9.

Thanks to Andrew Jeppeal, Marshalls, USA.

Thanks to Matt Gallion, Tenneco, USA.
Change 10.135 DB2 Audit Detail Report isn't produced when more than one

ANALDB2R report was requested, or when the SORTBY operand was used,

Aug 4, 1992 because of a mislocated %END statement. Find the first

DATA _NULL_ after the %MACRO PMAUD02 statement, and insert

a %END; statement immediately before that DATA _NULL_

statement. Then find %MEND PMAUD02; and delete the %END;

statement immediately preceding that %MEND statement.

Thanks to Tom Hubbard, Westone Bancorp, USA.

Change 10.134 NPM variable PCTBUSY, for full duplex lines, contains the

VMAC28 maximum of Primary (outbound from NCP) or Secondary

Aug 3, 1992 (inbound to NCP) utilization, but PCTBUSY won't identify

when one direction dominates. MXG added two new variables

PCTPRBSY/PCTSCBSY which contain the utilization in each

direction. An additional network variable, PCTNEGPL, the

percentage of polls that were negative polls, was added as

it is useful in identifying network response problems.

Thanks to Scott Taylor, Primerica, USA.
Change 10.133 Some observations in PDB.JOBS can have JELAPSTM missing,

IMACPDB even though variable INBITS contains a 'J', indicating a

Aug 1, 1992 type 30_5 record was found. (If there is no 'J' in the

third position of INBITS, then JELAPSTM is expected to be

missing, since it is calculated from the 30_5 record.)

This error affected only jobs with more than 1476 unique

DD segments in their type 30_5 record, and resulted from

the absence of variable MULTIDD in the _PDB30_5 macro that

is defined in member IMACPDB. MULTIDD='Y' identifies the

"extra" type 30 record created where there are more than

1476 DD segments in the record, and is used in BUILDPDB/3

logic to recognize these pseudo job (and step) records.

Because it was not in TYPE30_5 in BUILDPDB, JELAPSTM was

inadvertently missing, and also variable RESTARTS was not

accurate for these jobs. To correct the error, variable

MULTIDD was added to the _PDB30_5 definition in IMACPDB.

You could workaround the error by adding the following

statement in your reporting program that uses JELAPSTM:

IF JELAPSTM=. AND JINITIME NE . AND JTRMTIME NE . THEN

JELAPSTM=JTRMTIME-JINITIME;

Thanks to Jim Cummings, First Interstate Bank of Oregon, USA.
Change 10.132 CICS Journal data can be sent to the type 110 SMF record.

EXCICJRN Some sites modify CICS to send data (such as logon/logoff)

Jul 30, 1992 to a journal and direct that journal to SMF. This change

creates a new exit which allows you to gain control (for

example, to process a user journal record). If you think

you need to use this exit, call for further assistance.

Note that Change 10.164 adds support for the SAP journal

record from SAP accounting, and does not use EXCICJRN.

Thanks to John Ebner, SystemHouse, USA.
Change 10.131 Amdahl's MDF architecture supports 16 LPARs, but MXG only

ASUM70PR supported 8 in its summarization and trending. (A message

TRND70PR alerts you that more than 8 were found.) This change

Jul 27, 1992 adds variables for the additional 8 possible LPARS.

Thanks to Jeff McFadyen, Ministry of Correctional Services, CANADA.
Change 10.130 Calculation of tape length for 3420 (reels) in VMACTMS is

VMACTMS5 slightly incorrect. The IRG (inter record gap) of 0.6" is

Jul 27, 1992 for 1600BPI tapes, and IRG is only 0.3 inches for 6250s.

For a full tape of 5200 blocks at 32760 blksize, the 0.6"

value caused a 2400 foot tape to be reported as 2530 feet!

Thanks to Malcolm Sare, Australian National, AUSTRALIA.

Change 10.129 SAS 6.07 under CMS has serious problems for MXG users.

Jul 27, 1992 The concat of sourclibs (USERID.SOURCLIB, MXG.SOURCLIB)

did not work, although it is now reportedly corrected.

The workaround is to make a complete copy of SOURCLIB

and make your tailoring changes in that copy until SAS

fixes the problem so you can reinstall with your changes

isolated in the USERID.SOURCLIB.

Standard label tapes cannot be read under CMS SAS 6.07,

but SAS is working on the problem with Systems Center.
Change 10.128 BUILDPDB under CMS SAS needs correction. The new macros

BUILDPDB VMXGDKRN/VMXGDKRW must be renamed VMXGDKN/VMXGDKW because

BUILDPD3 CMS SAS is limited to only seven character macro names.

Jul 27, 1992 Additionally, execution under CMS SAS 5.18 will require

the addition of a FILE DEF for SMFTEMP, and the DCB=SMF in

the FILE SMFTEMP statement in BUILDPDB must be changed to

DCB=VB LRECL=32756 BLKSIZE=32760 since CMS cannot create

VBS files.

Change 10.127 JCLPDB6 in MXG 10.1 will fail with DATASET PDB.TYPETMNT

BUILDPDB NOT FOUND error, because ASUMTMNT is included in JCLPDB6

BUILDPD3 sample program, but the intended automatic creation of the

Jul 24, 1992 TYPETMNT (Tape Mount Monitor records, created by ASMTMNT),

did not get into MXG 10.1. Now, BUILDPDB/BUILDPD3 will

create TYPETMNT dataset, so that if you install MXGTMNT to

create SMF records, you will now only have to update the

IMACTMNT member in your USERID.SOURCLIB to tell MXG what

SMF record number you assigned, and your tape mount times

will automatically be in your PDB library.

Thanks to Jeff McFadyen, Ministry of Correctional Services, CANADA.
Change 10.126 Variable QWHSLUUC should have been spelled QWHSLUCC in

ANALDBTR the _VTRCMN macro definition for DB2 Trace processing.

Jul 24, 1992

Thanks to Christophe Faure, SAS Institute, FRANCE.

Change 10.125 Variable DS4VTOCE was input but not kept, and was not

VMXGVTOC labeled. Add it adjacent to DS4VTOCI in the keep list,

VMXGVTOF and add DS4VTOCE='VTOC*EXTENT*DESCRIPTION' adjacent to

Jul 24, 1992 DS4VTOCI in the LABEL statement.

Thanks to Christophe Faure, SAS Institute, FRANCE.
Change 10.124 IBM incompatibly changed the PSF type 6 SMF record, but

VMAC6 the change was not documented! Bin counts were added to

Jul 22, 1992 the APA subsection (and the added length was NOT added to

SMF6LN4!), and the ESS subsection, formerly only in the

JES2 type 6 SMF record, now exists in the PSF record; this

unexpected segment caused incorrect values for the APA

subsection variables DOCLENFT/FONT,OVLY,PGSG USED-LOAD.

The simple fix is to relocate the ESS section DO group

(22 lines) which starts with

IF SECTIND='...1....'B AND OFTONXT ... /* ESS SEG */

to just after the end of the PSF section DO group, which

starts with

IF SECTIND='..1.....'B AND OFTONXT ... /* PSF-APA */

This change also added BINnBNCT variables for 8 bins.

The APAR which apparently made this change is OY30945.

Thanks to Kevin James, Cornhill Insurance, ENGLAND.

Thanks to Tim Sparrow, Cornhill Insurance, ENGLAND.
Change 10.123 Variables R791BPIN,PINE,BPNE,CTAR, and R791VAL were added

VMAC79 to TYPE791 dataset. The test IF SMF79ASL GE 96 before the

Jul 21, 1992 input of R791HIN was changed to IF SMF79ASL GE 98. The

MVS/ESA 4.2 value for R791FMCT is occasionally too large

(129MB for INIT with WM/LW); IBM is investigating.

Variables R794PPIA/LPIA were added to TYPE794 dataset,

and redundant LCUID code was removed from TYPE79E logic.

Thanks to Steve Saunders, Talbots, USA.

Change 10.122 Total page movement between central and expanded storage

VMAC71 is EXTREADS (from ESTORE to CSTORE) plus PGMVTOEX (from

Jul 21, 1992 CSTORE to EXTORE). The sum of EXTREADS+PGMVTOEX is useful

in comparing before and after effects on memory changes.

Thanks to Sigfrido Perdomo, Metropolitan Life, USA.
Change 10.121 Datasets ILKAST20 and ILKAST21 had invalid data because

VMACILKA the offset was incorrect, and the documentation was wrong.

Jul 21, 1992 Change the line reading:

Aug 18, 1992 FPTDCSOF=FTPDCSOF-3+OFFSMF;

to read:

FPTDCSOF=FTPDCSOF+SMFACDOF;

Change the line reading:

FPTDCTOF=FTPDCTOF-3+OFFSMF:

to read:

FPTDCTOF=FTPDCTOF+SMFACDOF;

Change the +3 following F20DTTY PIB1. to +1