* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
| X GT 1.2E19 for shifted, or
otherwise input as unshifted PIB8. If STK ever puts a version number in their record, this heuristic will be removed. The media size variables are now formatted MGBYTES and the VPOs to HEX. See also Changes 17.195 and 17.332. Mar 16: The low four bytes of VTV timestamp now create STCxxSEQ, STCxxHID, STCxxVTI, and STCxxFLG for subtypes 13, 14, 18, and 19 thanks to STK documentation.
VMACORAC record with CONNID=CICS have KERNNUM=0, but it should be Mar 10, 2000 KERNNUM=1 because there is a valid Kernel segment in the record. When KERNNUM=0, MXG does not read the kernel segment, causing missing values for all kernel variables. Now, MXG detects and corrects the Oracle error and will read a Kernel segment even if the count field is wrong. This error has been reported to Oracle. An additional problem is being investigated: CPUTCBTM and CPUSRBTM fields are zero with the 7.3.3 release, and so far, Oracle has no knowledge of any change, but this text will be updated when more is known. Thanks to Yvonne McDonough, USDA NITC, USA. Thanks to Leslie Arndt, USDA NITC, USA. Change 18.033 Member ASUMTAPE was finally tested under ITSV and it too ASUMTAPE contained hardcode references to PDB.xxxxxxxx that are Mar 10, 2000 now changed to their &PDBMXG..xxxxxxxx symbolic names.
EX028INK -MXG error: VCD segment was not input for 'DC' or 'DE'. EX028INL It would have helped if IBM and documented those subtypes EX028INM in NPM 2.4 tables 73/74 or NPM 2.5 tables 80/81! Change FORMATS to ELSE IF 0D0X LE NPMSUBTY LE 0DEX THEN COFTYPE='VCD'; IMAC28 -MXG error: VCS segment has two flavors, and the last 4 VMAC28 variables, VCSMAXFX/CURFX/MAXEC/CUREC are now input only VMXGINIT if VCSVDTYP=2; MXG code was restructured for VCS segment. Mar 10, 2000 -IBM error: APAR OW37758 for NPM 2.4 corrected error in subtype DDx record, but that APAR is not included in the NPM 2.5 release: new APAR AW43578 documents the error. Change 16.336's circumvention still works to protect 2.5. -New datasets from new subtypes: Subtype Exit Dataset Description 14x 028INK NPMINNRT Router Interval 15x 028INL NPMINNRP Cisco Pool Interval 16x 028INM NPMINNIT Router Interface Interval -The NCD and SCD segments' logic was revised to only input the IP address for IP resource, and not input SLUNAME, SLUPU, SLNKNAME and NPMNAME for IP resource type. Thanks to Jerome Vitner, Experian Information Solutions, USA. Thanks to Cendrine Pezier, ABS Technologies, FRANCE. Change 18.031 Uninitialized variable messages for MROTRAN and DB2TRAN ASUMUOW caused no actual problem, but they are now initialized in Mar 9, 2000 the first DO; loop in the last DATA step. Thanks to Freddie Arie, Texas Utilities, USA. Change 18.030 IMS Log processing with APPC transactions caused negative ASMIMSL5 SERVICTM and N+1 rather than N transactions per schedule. ASMIMSL6 After label P031, after USING statement, insert: Mar 9, 2000 TM QLGUFLGX,QLGUAPPC APPC? BO DROPMAP SKIP APPC REDUNDANT 31 This has only been tested with IMS 5.1 data. If you have APPC under IMS 6.1 and have validated this change, or if you find any problems under IMS 6.1, please contact MXG. Thanks to Kenneth D. Jones, ISM, CANADA. Thanks to Carl Meredith, SAS Institute ITSV, USA. Change 18.029 Variables were labeled, times were re-formatted and the VMACMIM label (MS) was removed and fields divided by 1000 to put Mar 3, 2000 durations in units of TIME13.3. Mar 6, 2000 Mar 6: Uninit variable messages MIMCELL-MIMVCF were in first 18.01, but had no impact. Variable MIMMPOS is now labeled and variables MIMADTCR/MIMHDTCR/MIMATMCR/MIMHTMCR are no longer kept Thanks to Chris Weston, SAS Institute Cary, USA. Change 18.028 Partition Data Report was updated for APAR OW37565, in ANALRMFR the body of the Partition Report, where 'CP ' or 'ICF' Mar 3, 2000 will be shown for Processor TYPE. Mar 7, 2000 Mar 7: VMAC75 replaced VMAC71,VMAC77 in line 938. Change 18.027 APAR OW41317 retrofits the OS/390 R2.9 multi-system VMAC7072 enclave support (which added data fields in type 72) Mar 3, 2000 back to OS/390 R2.7. Change 17.385, in MXG 17.17 and later already supports those fields; this change is only for documentation; no change was made.
VMAC21 at the end of the record (i.e., compatibly) that replace Mar 3, 2000 the existing two-byte fields for SIOCOUNT and BLKSIZE. Mar 7, 2000 Change 18.025 If you did not put all SuperSession maintenance on, its VMACNAF SMF record still contains CENT=39 instead of CENT=01 Mar 3, 2000 caused MXG's circumvention algorithm to be off by one day. Remove the term (CENT-38)*86400 from each of the datetime adjustments when CENT=39 and Y2K dates after March 1, 2000 will be correct. Thanks to Joseph J. Faska, Depository Trust, USA.
VMAC80A in RACFTYPE=39 segment (PROGRAM NAMES, PERMIT). The line Mar 3, 2000 SKIP=RACFDLNN=64-3 must be SKIP=RACFDLN-RACFDLNN-3; Mar 9, 2000 Mar 9: The debugging PUT statement was deleted. Mar 14, 2000 Mar14: Variables KW21AU04/05 and KW21GL04/05 formatted.
VMAC7072 have been LENGTH $3 instead of $2, since it contains CP Mar 3, 2000 or ICF. This error prevented ASUM70PR from recognizing and removing ICF processors from the CEC capacity. Change 18.022 Variable DURATM was added to those MIM datasets that VMACMIM contained start and end time; it previously had been Mar 3, 2000 kept only in the MIMTAPE dataset, but MXG intends to keep DURATM in all interval datasets. Only the TYMIM2, TYMIM4, and TYMIM6 are event datasets. Mar 6 note: Uninitialized variables messages MIMCELL-MIMVCF are created but have no impact; labels were removed.
Change 18.021 APAF Release 4.6 added a new field Series Name to the VMACAPAF Pool segment, causing the second and subsequent pools' Mar 1, 2000 data to be trashed. Insert IF LENPOOL GE 36 THEN INPUT APAFSENM $EBCDIC16. @; add APAFSENM='SERIES*NAME' to the LABEL, and to the KEEP= list for dataset APAFPOOL, and then notice and correct the _KAPAFLC to be _KAPAFPO for the APAFPOOL dataset. The other APAF data sets appear to have been unchanged by this release Thanks to Art Cuneo, Health Care Service Corporation, USA.
VMXGDEL but revisions to this %VMXGDEL macro now supports that Feb 29, 2000 syntax. I had not provided an option for compressing of a sorted dataset, but redefining the dataset's _L macro: MACRO _Ldddddd ddname.dataset (compress=yes) % is now valid for compressing individual datasets. VMXGDEL now uses PROC DATASETS MT=ALL so that if you made a SAS dataset to be a SAS view, both will be deleted Thanks to Alex Torben Nielsen, TeleDanmark EDB, DENMARK. Thanks to Michael L. Kynch, International Paper, USA. Change 18.019 The vendor of BETA93 writes non-Y2K-Ready date values in VMACBETA the BETAJOBT SMFSTAMP8 value, with 00dddF format for Y2K Mar 1, 2000 with no century bit turned on. The SMFSTAMP8 input was broken into Time and Date pieces, the Date is now protected with MXG's DATEJUL algorithm, and BETAJOBT is then reconstructed from its (protected) pieces. Thanks to Frank d'Hoine Nationale Bank van Belgie, BELGIUM.
VMAC30 may be eight blanks, or may have only the first of JES3's Feb 29, 2000 8-character job class, because IBM changed SMF30CLS, the old one-byte job class field. Before IBM added SMF30CL8 in 1993, SMF30CLS was INPUT into the 1-byte MXG variable JOBCLASS. In support for IBM's 8-byte SMF30CL8 field for JES3, MXG increased JOBCLASS to 8-bytes, read SMF30CLS into the first byte of JOBCLASS, then input SMF30CL8 into variable JOBCLAS8, and used this logic to prevent a blank JOBCLAS8 value from overriding a non-blank JOBCLASS: IF JOBCLASS LE ' ' AND JOBCLAS8 GT ' ' THEN JOBCLASS=JOBCLAS8; But now in JES2 R2.8 OS/390 records, SMF30CLS is blank and SMF30CL8's first byte contains the JES2 jobclass, and now in JES3 R1.3 OS/390 records, that 1-byte SMF30CLS field that used to be blank is not blank, so the MXG logic had to be changed to match the IBM Incompatibility: IF JOBCLAS8 GT ' ' THEN JOBCLASS=JOBCLAS8; Variable JOBCLASS is length 8 in TYPE30xx datasets built by VMAC30 code (because I can't tell if a type 30 record is from JES2 or JES3), and kept as length 8 in JES3's BUILDPD3's datasets, but with JES2's BUILDPDB, I can keep the variable JOBCLASS in only one byte. Thanks to Allan Koch, Ford Motor Company, USA. Thanks to Rich Anderson, SAS Institute Cary, USA. Thanks to Ken Hansen, Exxon, USA. Change 18.017 The UTILRMFI utility is used to examine any overlap of UTILRMFI CPU time in your RMFINTRV workload definitions. It has Feb 29, 2000 been updated to match the VMXGRMFI workloads, and uses type 30 and type 72 records. See its comments for usage. Change 18.016 Variable TYPEIOML was blank in dataset TYPE70 because of VMAC7072 an extraneous line IF SMF70RAN THEN that should have Feb 26, 2000 been deleted. Fortunately, TYPEIOML is always '3090' and is unlikely to have been used in any reports! Thanks to Bill Shanks, SEMA, UK.
VMAC38 S38CUSER should have been $EBCDIC4. instead of $EBCDIC8. Feb 25, 2000 -Additionally, TYPE38 failed if you tried to read the VSAM SMF file because OFFSMF was not added to each offset calculation. Thanks to Kenneth D. Jones, ISM, CANADA. Thanks to Harald Seifert, Huk-Coburg, GERMANY.
BUILD001 in SAS Version 7 or 8; it caused a warning message that BUIL3001 had no other impact, and CODEPASS=2 itself had no impact BUILDPDB on MXG, and was only specified to suppress a V6 message BUILDPD3 that suggested you should try CODEPASS=2! Now it will be Mar 1, 2000 specified only under SAS V6 execution. Thanks to Mike Hoey, Ameren Services, USA.
ASUMTALO USER= option to send all of the datasets created during Feb 25, 2000 ASUMTALO's execution to the same DDname) fails with error V6: You cannot open WORK.TYPETALO.DATA for output access with member level control because WORK.TYPETALO.DATA is in use by resource environment SASEDSNX V8: MEMBER lock is not available for WORK.TYPETALO.DATA lock held by another user because views and datasets of the same name cannot exist in the same data library. The logic was revised and temporary dataset/view names that do not conflict are now used, and SPINTALO is no longer a view to avoid conflict. Thanks to Sharon L. Nagy, Comerica, USA Thanks to Robert K. Hare, Comerica, USA Change 18.012 The calculation of RESPSTD (Std Deviation of Response) TRND72GO did not protect for TRANS=0 before the divide by TRANS, Feb 24, 2000 causing DIVIDE BY ZERO note on the log (but the dataset created was still valid). Now divide is protected. Thanks to Mike Hoey, Ameren Services, USA.
VMACNTSM message because MXG was expecting SnaDlc1 as the string. Feb 24, 2000 The test was changed to SNADLC*, and the label of the SNAADAPT dataset was also corrected. -Variables DATABASE and APPLETLK appear in both LENGTH $32 and in INFORMAT 16.2 statements, but the SAS compiler did not flag the inconsistency. Both were removed from the numeric INFORMAT statement as they are indeed characters. Thanks to Richard C. Clarke, Freemans PLC, ENGLAND
ANALDSET _NULL_ is not in the data statement". Change the DATA Feb 19, 2000 statement and insert the needed _NULL_ so it reads: DATA _VAR1415 _VAR30 _VAR64 _NULL_ _SMF _CDE1415 _CDE30 _CDE64; Thanks to Ed Long, Fidelity Investments, USA.
RMFINTRV of Report PGNs or Reporting Classes to define workloads VMXGRMFI (either in your IMACWORK definitions, or in your Feb 18, 2000 tailored %VMXGRMFI invocation your RMFINTRV member) Feb 25, 2000 caused the error message "CPU TIMES DO NOT MATCH" on the SAS log during creation of dataset PDB.RMFINTRV. That error message occurs because either: - you used the new DROPPGN= or DROPSRV= operands, which didn't work correctly until this change to VMXGRMFI, or - you didn't use those new operands, in which case there was double accounting.
and now has six operands to make tailoring easier if you want to use RPGNs/Reporting Classes to define workloads: USEREPRT= Default NO, set to YES to permit Report PGN or Reporting Classes to be used. USECNTRL= Default YES, set to NO to delete all Control PGNs or Service Classes. This is only used if you want to use only Report PGN/Classes, and you don't want to list all of them in the DROPSRV= or DROPPGN= arguments. DROPSRV = List of Service Classes to drop. DROPPGN = List of Control Performance Groups to drop. DROPRSRV= List of Reporting Classes to drop. DROPRPGN= List of Report Performance Groups to drop. To be able to use either Reporting Classes or Report PGNs to define your workloads in dataset RMFINTRV, you MUST use these new arguments in your member RMFINTRV's invoke of VMXGRMFI to drop the Control PGN or Service Class observations that correspond to the RPGN/Reporting Class observations you want kept in your workloads, to avoid double accounting and that error! You must also ensure that every SUBSYS has a default Report PGN or Reporting Class so that all work in the corresponding Control Performance Group or Service Class is recorded in type 72 records.
because it contained a hardcode OUTDATA=PDB.RMFINTRV, in the actual %VMXGRMFI invocation in member RMFINTRV, but ITSV expected the symbolic &PDBRMFI..RMFINTRV so that ITSV could redirect to the WORK library. The real bummer is that that OUTDATA= operand is not needed in the actual invocation, and was added only as a documentation example and reminder of the default output dataset name! Remove that operand from the %VMXGRMFI invocation, and the expected symbolic value will be used. -This change also corrects an unrelated problem: when no PERFGRP number was needed to be specified, a blank value caused VMXGRMFI to fail. While using a bogus number like 9999 circumvented the problem, now a blank can be used, but using a null value ( // ) fails. Use / / instead. Thanks to Normand Poitras, ISM, CANADA. Thanks to Jon G. Whitcomb, Great Lakes Higher Education Corp, USA.
VMAC7072 used to create several variables and should have been Feb 17, 2000 kept. Thanks to Frank Cortell, Credit Suisse Group, USA. Change 18.007 SPIN logic was revised to correctly pick up class three ASUMUOW wait times, and DB2ELAP (elapsed time in DB2) was added Feb 29, 2000 to the final dataset. Previously, if a UOW was SPUN, the DB2 Wait buckets and transaction counts were not put in the record in the SPIN dataset. Thanks to Brandon Persinger, The Gap, USA. Thanks to Allan Winston, MBNA, USA.
VMACQAPM 02Sep99, records from this century will have dates in Feb 16, 2000 year 1900 instead of year 2000. The QAPMCONF file must match the century of the data you want to process, since MXG gets AS400CC "Century" value from QAPMCONF. However, you can circumvent the missing CC=1 value in your QAPMCONF file by resetting the value of the macro variable after you have invoked _TQAPCON: %INCLUDE SOURCLIB(VMACQAPM); _TQAPCON
RUN; /*<<= insert to set cc=1 */
etc...
Thanks to Kim Johnson, Bank of America, USA. Change 18.005 There is an end-of-comment */ missing from the second TRNDSMFI line of this trending example. Feb 15, 2000 Thanks to Carl Kyonka, Enbridge Consumers Gas Company Ltd, CANADA. Change 18.004 Type 79 subtype 15 IRLM Long Lock record has SMF79TRN=2 VMAC79 but there is no product section, so STARTIME, SYSPLEX, Feb 17, 2000 DURATM and other product fields do not exist in TYPE79F. Feb 22, 2000 In addition, the second triplet is SMF79ASS rather than SMF79MCS, so the input logic for triplets was revised. Finally, variables are now formatted to HEX when they do not contain printable characters, and R79FTRNM is now kept and labeled. Note that this type 79 subtype 15 record replaces the type 74 subtype 100 dataset that MXG creates in VMAC74. That subtype 100 was a test version that doesn't exist. Thanks to Johannes-Matthias Laveuve, DVG, GERMANY. Thanks to Jurgen Scherwa, DVG, GERMANY.
VMACDB2 and they cause "UNEXPECTED BUFFER POOL ID" messages to be Feb 11, 2000 printed, because MXG did not know about . Those new buffer pools are output in the detail by-pool datasets DB2STATB and DB2ACCTB (although you have to have removed the comment block in member EXDB2ACB to cause MXG to create observations in DB2ACCTB), but the new pools counts were not included in DB2ACCT or DB2STAT datasets. In DB2ACCT/DB2STAT, MXG creates four sets of buffer pool variables: QB1xxxx for BP zero, QB2xxxx for BP one, QB3xxxx for all other 4K buffers, and QB4xxxx for all 32K buffers. Rather than create yet another set of summary counters for the 8K and 16K buffer pools, I have added the 8K and 16K buffer counts to the QB4xxxx variables in DB2ACCT and DB2STAT, since you can always retrieve the individual buffer pool counts from DB2ACCTB/DB2STATB. This change replaces the statement: ELSE IF 80 LE QBACPID LE 80 THEN DO; with ELSE IF 80 LE QBACPID LE 89 OR 100 LE QBACPID LE 109 OR 120 LE QBACPID LE 129 THEN DO; for DB2ACCT, and replaces this statement: ELSE IF 80 LE QBSTPID LE 80 THEN DO; with
ELSE IF 80 LE QBSTPID LE 89 OR 100 LE QBSTPID LE 109 OR 120 LE QBSTPID LE 129 THEN DO; for dataset DB2STATS. Thanks to Joachim Mayer, Amadeus Data Processing GmbH, GERMANY. Change 18.002 Omegamon V500 records were deleted simply because the VMACOMCI test for version number did not include V500. Add Feb 7, 2000 AND FOCVER NE 'V500' and the records will be read. Thanks to Peter Webber, CIS, UK. Change 18.001 The _Bdddddd BY List macro was not sufficiently robust to VMAC1415 remove all duplicates for some datasets, so the BY list VMAC30 of variables has been revised so that you can PROC APPEND VMAC42 to combine possibly duplicated data and then use the MXG VMAC6156 _Sdddddd sort macro (which uses the revised _Bdddddd BY VMAC64 list macro) to eliminate duplicates. However. there are VMAC7072 four datasets that contain intrinsic duplicates; there VMAC74 are repeated observations for the same set of BY vars VMAC89 for the first four datasets listed (with asterisk) that VMAC115 will require further research and discussion with IBM. VMAC116 *MQMACCT _BTY116 now SYSTEM MQMSSSID QWHCAID VMACACC QWHCCV QWHCCN SMFTIME VMACTCP *TYPETCPA _BTYTCPA now SYSTEM APIJOBNM APISTART Feb 18, 2000 APILCLPO APILCLPN APILOCAL APIREMOT SMFTIME *TYPE42DS _BTY42DS now SYSTEM STARTIME JOB READTIME DSNAME DEVNR *TYPE30MU _BTY30MU now READTIME JOB JESNR INITTIME SUBTYPE PRODOWNR PRODNAME PRODQUAL PRODID SMFTIME TYPEACC _BTYACC now SYSTEM READTIME JOB SRHSTEPN SRHDDN MESSAGE SMFTIME MQMBUFER _BTY115B now SYSTEM MQMSSSID QPSTPOOL SMFTIME
TYPE30OM _BTY30OM now READTIME JOB JESNR SUBTYPE STEPNR SUBSTEP OMVSOPI OMVSOUI OMVSOSI SMFTIME TYPE892 _BTY892 now SYSPLEX SYSTEM PRODOWNR PRODNAME PRODFEAT PRODID PRODSTFL PRODREGS SMF89IST SMFTIME TYPE89 _BTY89 now SYSPLEX SYSTEM PRODOWNR PRODNAME PRODQUAL PRODID MULCUFG SMF89IST SMF89UST SMFTIME TYPE74SY _BTY74SY now SYSPLEX SYSTEM STARTIME R742SNME R742SDIR R742STCN SMFTIME
TYPE74ME _BTY74ME now SYSPLEX SYSTEM STARTIME R742MSYS R742MGRP R742MMEM SMFTIME TYPE72SC _BTY72SC now SYSPLEX SYSTEM STARTIME SRVCLASS R723SCSN R723SCSR SMFTIME
TYPE64 _BTY64 now SYSTEM READTIME JOB CATNAME ENTRNAME SITUATN COMPONT DDNAME VOLSER1-VOLSER3 SMFTIME
TYPE6156 _BTY6156 now SYSTEM READTIME JOB SMF6XFNC CATNAME ENTRNAME GENNO VERNO VOLSER1-VOLSER5 SMFTIME PDB.SMFINTRV _BTY30UV add SMFTIME DESCENDING INTBTIME MULTIDD EXTRADD TYPE1415 _BTY1415 now SYSTEM READTIME JOB OPENTIME DDNAME DSNAME UCBSEGNR SMFTIME
TYPE1718 _BTY1718 now SYSTEM READTIME JOB DSNAME VOLSER NEWNAME SMFTIME This is work still in progress; I need to formally open a problem with IBM to discuss the duplicate records that we've identified. This note updated 1Mar00. Thanks to Abakah Nori, United Parcel Service, USA. LASTCHANGE: Version 18. =========================member=CHANGE17================================ /* COPYRIGHT (C) 1984-2000 MERRILL CONSULTANTS DALLAS TEXAS USA */
MXG Version 17.11 was dated Feb 2, 2000, thru Change 17.395. First MXG Version 17.11 was dated Jan 31, 2000, thru Change 17.394. First MXG Version 17.10 was dated Jan 20, 2000, thru Change 17.375. MXG Version 17.09 was dated Dec 22, 1999, thru Change 17.339. Yüklə 28,67 Mb. Dostları ilə paylaş:
|