* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
By default, IMACRMFI member does no summarization, and instead creates one observation in PDB.RMFINTRV and PDB.ASUM70PR for each original RMF interval. -If you have tailored member IMACRMFI to set the summary interval (eg, HOUR, SHIFT, etc) you want, then it will be used for both RMFINTRV and ASUM70PR. -If instead you tailored member RMFINTRV to use INTERVAL= or SYNC59= arguments, then you MUST also tailor member ASUM70PR to use those same arguments and values. -With the added variables, the maximum percent CPU busy that a system can reach based on LPAR weights can be calculated LPARMAX=100*(LPARSHAR/TOTSHARE)*(NRCPUS/PARTNCPU); E.g.: CEC with 11 physical engines and with 3 LPARS: 6 engines and weights of 47 in two LPARS 2 engines and weight of 6 in one LPAR LPARMAX = 100 * (47/100) * (11/6) = 86.6 % is the busyiest those six engine LPARs can be, even though PLATBUSY is 100% -Correction in ASUM70PR for Dedicated processors to change IF LCPUDED='Y' AND LCPUWAIT='Y' THEN DO; to now read IF LCPUDED='Y' OR LCPUWAIT='Y' THEN DO; because the logic only worked for Dedicated and LCPUWAIT, which could cause the non-dedicated CPU busy to be wrong. Thanks to Chuck Hopf, MBNA, USA. Thanks to Steve Emson, TESCO, ENGLAND. Thanks to Richard Rich, State of California Teale Data Center, USA.
EXTY8201 creates thirteen new datasets: EXTY8203 TY8201 TYPE8201 INITIALIZATION EXTY8204 TY8203 TYPE8203 STATUS CHANGE EXTY8205 TY8204 TYPE8204 CONDITION CODE 3 EXTY8206 TY8205 TYPE8205 SPECIAL SECURITY MODE EXTY8207 TY8206 TYPE8206 MASTER KEY PART EXTY8208 TY8207 TYPE8207 KEUK PART EXTY8209 TY8208 TYPE8208 CKDS REFRESH EXTY8210 TY8209 TYPE8209 CKDS UPDATE EXTY8211 TY8210 TYPE8210 PKA KEY PART EXTY8212 TY8211 TYPE8211 CLEAR NEW MASTER KEY PART EXTY8213 TY8212 TYPE8212 PUBLIC KEY SECURE CABLE VMAC82 TY8213 TYPE8213 PUBLIC KDS UPDATE VMXGINIT Oct 19, 2001 Change 19.199 Incorrect copying in the weekly phase was corrected by BLDNTPDB inserting a RUN; statement between the PROC COPY and the Oct 19, 2001 %VMXGOPTR invocation (to force macro resolution). Thanks to Greg Jackson, National Life of Vermont, USA Change 19.198 Invalid values for sample counts of 'FFFFFFFF'x for the VMAC7072 PCTDLTOT (R723TOT) and PCTDLQ (R723CQ) and 'FFFFFFFA'x Oct 19, 2001 for PCTDLTDQ (R723CTDQ) caused very small VELOCITY and very large PERFINDX values. MXG now tests those three fields and forces a missing value when large values are found in the IBM record. Updated: Oct 30, 2001: IBM APAR OW50380 fixes the error, at least for basic mode. Thanks to Jiann-Shiun Huang, CIGNA, USA Thanks to Perry Metzel, Hewitt, USA. Change 19.197 -Support for revised SYSTEM object with NRDATA=24, which EXNTEPXY reinstated Total Interrupts/Sec counter to match all of EXNTEXEX the SYSTEM fields that used to exist in NT 4.0. EXNTMEAL -Support for the restructured SQL 2000 objects creates EXNTMECA these seventeen new SQLServer datasets: EXNTMECO NTQLAM MSQACCES NT MSSQL:ACCESS METHODS EXNTMEIM NTQLBD MSQBKPDV NT MSSQL:BACKUP DEVICE EXNTMEM NTQLBM MSQBUFMG NT MSSQL:BUFFER MANAGER EXNTMEOE NTQLBP MSQBUFPA NT MSSQL:BUFFER PARTITION EXNTMEOR NTQLCM MSQCACMG NT MSSQL:CACHE MANAGER EXNTMEPO NTQLDA MSQDATAB NT MSSQL:DATABASES EXNTMEPR NTQLGS MSQGENST NT MSSQL:GENERAL STATISTICS EXNTMESA NTQLLA MSQLATCH NT MSSQL:LATCHES EXNTMESR NTQLLO MSQLOCKS NT MSSQL:LOCKS EXNTMETD NTQLMM MSQMEMMG NT MSSQL:MEMORY MANAGER EXNTMETS NTQLRA MSQREPAG NT MSSQL:REPLICATION AGENTS EXNTMEWM NTQLRD MSQREPDI NT MSSQL:REPLICATION DIST. EXNTMIGA NTQLRL MSQREPLR NT MSSQL:REPLICATION LOGREADER EXNTMIGP NTQLRM MSQREPME NT MSSQL:REPLICATION MERGE EXNTMISC NTQLRS MSQREPSN NT MSSQL:ACCESS METHODS EXNTMISE NTQLSE MSQUSRSE NT MSSQL:USER SETTABLE EXNTMISI NTQLSS MSQSTATS NT MSSQL:SQL STATISTICS EXNTQLAM -Support for 14 new MS Exchange objects create datasets: EXNTQLBD NTMEAL MSEXMEAL NT MSEXCHANGEAL EXNTQLBM NTMECA MSEXCACH NT MSEXCHANGEDSACCESS CACHES EXNTQLBP NTMECO MSEXCONT NT MSEXCHANGEDSACCESS CONTEXTS EXNTQLCM NTMEIM MSEXIMAP NT MSEXCHANGEIMAP4 EXNTQLDA NTMEM MEMORY NT MEMORY EXNTQLGS NTMEOE MSEXOLEV NT MSEXCHANGE OLEDB EVENTS EXNTQLLA NTMEOR MSEXOLRE NT MSEXCHANGE OLEDB RESOURCE EXNTQLLO NTMEPO MSEXPOP3 NT MSEXCHANGEPOP3 EXNTQLMM NTMEPR MSEXPROC NT MSEXCHANGEALACCESS PROCESSES EXNTQLRA NTMESA MSEXMESA NT MSEXCHANGESA - NSPI PROXY EXNTQLRD NTMESR MSEXSRS NT MSEXCHANGESRS EXNTQLRL NTMETD MSEXTRDR NT MSEXCHANGEIS TRANSPORT DRIVER EXNTQLRM NTMETS MSEXTRSD NT MSEXCHANGETRANSPORT STORE DRIVER EXNTQLRS NTMEWM MSEXWEBM NT MSEXCHANGEWEB MAIL EXNTQLSE The duplicate removal logic was validated and BY lists EXNTQLSS were enhanced where needed to ensure removal. There are IMACNTSM actual duplicates in MSEXCONT that are being investiaged. NTINTRV -In NTINTRV, the NTCONFIG dataset is no longer combined in VMACNTSM to PDB.NTINTRV because is is not an interval dataset. VMXGINIT -SP2 for Exchange 2000 added object MSExchangeIS Mailbox Oct 19, 2001 that appears to have the same fields as MS..eIS Public, so the Mailbox records are output in MSEXISPU dataset. Object for MSExchangeIS was restructured with some new and some removed fields. The "Database ==> Instances" object is questionable, and missing fields in Database object have yet to be resolved, but both objects with "Database" in their name are now output into DATABASE. -New Epoxy object creates new EPOXY dataset. -New MicroSoft Gatherer object creates new MIGATHER -New " Gatherer Projects object creates new MIGATHPR -New Microsoft Search object creates new MISEARCH -New " Search Catalogs object creates new MISEARSC -New " Search Indexer Catalogs creates new MISEARSI -New Exchange Server HTTP Extensions new EXCHHTEX
VMACDB2 was added by DB2 Release 6, but was skipped by MXG; now Oct 17, 2001 that variable is INPUT and kept in dataset DB2GBPAT. Thanks to Don Deese, (CPExpert), Computer Management Sciences, USA. Change 19.195 Cosmetic. Some of the error messages for bad segments VMAC108 had incorrect segment name in the text, possibly causing Oct 17, 2001 confusion, if any of those conditions ever occurred. Thanks to Bob Furlong, Chase, USA. Change 19.194 Change 19.101 removed PGMRNAME from the TYPE30_4 step BUILD005 dataset in BUILDPDB processing to protect against blank BUIL3005 values, but that change caused PGMRNAME to not be kept Oct 17, 2001 in PDB.STEPS; by adding PGMRNAME to the PDBADD2/PDBADD3 Apr 5, 2002 macros, the PGRMNAME from the PDB.JOBS dataset will now be copied into the PDB.STEPS dataset. However, PGMRNAME will still be blank if no PGMRNAME was used on the JOB card: PDB.JOBS and PDB.STEPS contains observations for all "jobs": STCs, TSO, OMVS, USS, APPC, etc, all will create "job" records that can have no PGMRNAME value, so variable TYPETASK may explain blank PGMRNAME values. Since BUILDPDB keeps PGMRNAME from only the 30.1 Job Init and 26 Job Purge, if "Spinning" was not enabled in your IMACSPIN, and if only step/job/print without init/purge are found for a job, then variable INBITS won't contain an I or a P, and PGMRNAME will be blank in JOBS/STEPS. 5Apr02: Text revised: why it can still be blank. Thanks to Bob Greer, Nissan North America, USA. Change 19.193 Support for IFCID=225, DB2 Storage Manager Pool Summary VMAC102 Statistics to create dataset T102S225, with lots of stats Oct 17, 2001 on pool sizes and concurrent activity. Thanks to Sim Williams, Fidelity, USA. Change 19.192 New formatted variable S42DSBUF for VSAM Buffer type: FORMATS S42DSBUF='VSAM*BUFFER*TYPE' VMAC42 with values 0:NSR 1:RLS 2:LSR 3:GSR Oct 16, 2001 and new character flag variables ('Y' or blank) are added S42DSEXC='OPEN FOR*EXCP*PROCESSING?' S42DSFXD='NONVSAM*FIXED*LENGTH*RECORDS?' S42DSPL ='PROGRAM*LIBRARY?' S42DSEF ='EXTENDED*FORMAT?' S42DSEFC='COMPRESSED*FORMAT?' to TYPE42DS by decoding the existing S42DSFL1 variable. These new variables were used to examine why the S42AMxxx Access Method Statistics variables are sometimes missing in TYPE42DS. Thus far, all observations with S42DSEXC='Y' - EXCP Access Method do not contain an Access Method Statistics section, but no other pattern has been found to explain why that A.M. segment is not present (S42DSAMO=0) for other datasets. Thanks to Rob D'Andrea, NatWest, ENGLAND. Change 19.191 Variable CPUTYPE was added to the KEEP= list for TYPE70PR VMAC7072 dataset. Oct 15, 2001 Thanks to Helmut Roese, COM Software GmbH, GERMANY. Change 19.190 This example report to calculate per-service-class CPU ANALSRVC busy percentages had a FORMAT PGPCTCAP PGPCTUSE 5.1; that Oct 7, 2001 should have been FORMAT SVPCTCAP SVPCTUSE 5.1; . Thanks to Sudie Wulfert-Schickedanz, Anheuser-Busch, USA. ======Changes thru 19.189 were in MXG 19.04 dated Oct 5, 2001====== Change 19.189 Support for APAR OW49536 (INCOMPAT if you have SPARE CPU VMAC7072 engines on your 2064 CPU). Reserve/Spare CPUs on 2064s Oct 5, 2001 have segments in type 70 records that were being counted as a real CP engine (both in RMF reports and in MXG!). This change tests SMF70ONT, the new LPAR "online time"; only if that value is non-zero will that CP be counted. A missing value in SMF70ONT will be counted; that is a record without the new field that can't have the problem. A non-zero non-missing value in SMF70ONT will be counted, because it was online during the interval. Variable LPARCPUS in dataset TYPE70PR, and the LPnNRPRC variables for each LPAR in dataset ASUM70PR will be wrong and LPAR percentages based on these variables will also be incorrect without this change. Thanks to Al Sherkow, I/S Management Strategies Ltd, USA.
TYPEIMSA variable END7TIME was not corrected, so these variables Oct 5, 2001 IMSSRVTM IMSRESTM IMSSRVTS and IMSRESTS were off from TOD by the value of GMTOFFTM. All are now corrected. Thanks to Daniel Cannon, The Hartford, USA.
ASUMNTIN -NTCONFG file has been added in NTINTRV so KEEPALL=YES can BLDNTPDB be used in ASUMNTIN. NTINTRV -ASUMNTIN now uses KEEPALL=YES for performance reasons. -RUNMNTH processing in BLDNTPDB corrects problem passing Oct 5, 2001 a variable that should not have been. Stay tuned: more enhancements, like RERUN, in progress.
VMAC74 observations from the subsequent, incomplete record when Oct 2, 2001 that record should have been deleted, as it was precedeed by the MXG "Broken Record" message from the previous SMF record. This change only outputs to TYPE74CF when the SUM(SMF744XN,SMF744GN,SMF744QN,SMF744SN,SMF744PN) is greater than zero. For First 19.04: This change relocate the RO section, inserted the missing END; and corrected the comment that has masked the missing END; Thanks to Wally Danielson, Airborne, USA.
EXMPLXGA one per subtype, from Williams Data Systems: EXMPLXIN Subtype DDDDDD Dataset EXMPLXPE 1 MPLXIN IMPLEXIN Interface Statistics EXMPLXPM 2 MPLXSE IMPLEXSE Service Statistics EXMPLXPO 3 MPLXGA IMPLEXGA Gateway Statistics EXMPLXRT 4 MPLXRT IMPLEXRT RTT Statistics EXMPLXSE 5 MPLXPR IMPLEXPR Protocol Monitoring FORMATS These records contains accumulated fields, so the member IMACMPLX both TYPEMPLX and TYPSMPLX invoke the _SMPLX "sort MPLX" TYPEMPLX macro, which sorts and deaccumulates these datasets into TYPSMPLX the default DDname of PDB. Implex is an IP monitor with VMACMPLX 3270 response time monitoring. VMXGINIT
Oct 5, 2001 Thanks to Rolf Meinert, Volkswagen, GERMANY. Change 19.184 MXG Data Sets contain labels with unique syntax DDDDDD:, BLDNTPDB that is used via PROC CONTENTS to control copying of the Oct 2, 2001 daily/weekly/monthly in the BLDNTPDB logic. However, the DATA WEEK.NTxxxxx; SET MON.NTxxxxx ... SUN.NTxxxxx; does not propagate a dataset LABEL, causing RUNMNTH=YES to fail with ERROR: BY VARIABLE _B IS NOT ON INPUT DATASET (because the DDDDDD value is used to create &SUFX, but if there is no DDDDDD value in the label, &SUFX is blank). The circumvention is to use the PDB, instead of the WEEK. as the library to read to get the DDDDDD values. Change %VMXGINV(LIBNAME=WEEK); to %VMXGINV(LIBNAME=PDB) after the comment "BUILD THE MONTHLY" to circumvent. Thanks to Liz Slack, Premera Blue Cross, USA.
VMAC74 The test IF SKIP GT 4 THEN DO; prior to its INPUT should Oct 1, 2001 have been IF SKIP GE 4 THEN DO; Thanks to Philip Foster, CGI Inc, CANADA. Change 19.182 A new alternative report member for GRAFWORK, enhanced to GRAFWRKM support all workload that are now createable in RMFINTRV, GRAFWRKX with an additional neat twist. With GRAFWORK, workloads Oct 1, 2001 are stacked from bottom to top and the order was fixed: Overhead, CICS, IMS, TSO, BATCH, OTHER OTH1-OTH9. But now with GRAFWRKS, if IMACWORK=NO is used, there is total control of the order of stacking (except that the "overhead", or uncaptured, is always at the bottom, so you can order the stacking to map the priorities of your workloads. GRAFWRKM is an example using GRAFWRKX. Thanks to Chuck Hopf, MBNA, USA.
VMAC110 is almost compatible! IBM added two new variables at the Sep 29, 2001 end of the standard transaction segment: PTPWAITM='3270 BRIDGE*PARTNER*WAIT*ELAPSED*TIME' PTPWAICN='3270 BRIDGE*PARTNER*WAIT*COUNT' so that all of the standard IBM fields are still correct with earlier versions of MXG that supported TS 2.1 data. However, if you are processing any of the optional CICS segments, this change is required so that the optional variables are valid: (you ARE processing optional CICS segments if there are IMACICxx members in your tailoring MXG library). This change supports only the subtype 1 record. A later change will support the CICS TS 2.2 Statistics subtype 2 record changes (INCOMPAT: CICDS data moved to STID=60).
EXT120JI Server, Enterprise Edition) which adds new Server Type: EXT120JC -MOFW server supports both C++ and Java BOs (in 3.02) FORMATS -J2EE server supports EJBs only (new in 4.0) VMAC120 The existing subtype 1 and 3 records for MOFW servers are VMXGINIT reused for the new J2EE server, so existing MXG datasets Sep 28, 2001 TYP102SA and TYP102SC from subtype 1 Server Activity and Oct 17, 2001 TYP102SI from subtype 3 Server Interval are changed only by the addition of new variable SM120ST, Server Type, to identify MOFW or J2EE server type. Existing MOFW subtype 2 and 4 could not be reused; new subtypes 5 and 6 create MXG dataset TYP120JC (subtype 5 J2EE Container Activity) and dataset TYP120JI (subtype 6 J2EE Container Interval). -These new records are the first SMF records to contain UCS, or Unicode DBCS (Double Byte) character data in new variables SM120JA8/SM120JB1/SM120JM1/SM120JA, with the names of containers, ejbRoles, methods, and AMCNames of the bean, etc. However, SAS support for UCS (Informat $UCS2Bnnn.) became available in SAS V8, so those new UCS variables will have valid printable names only if SAS Version 8 or later is used to build the datasets.
characters (512 or 1024 bytes of DBCS data in the record) until there is a need for more text, and for simplicity and compatibility, only the first 200 characters are kept, creating a unique new MXG macro variable named UCS2B4, with value: IF &SYSVER GE 8.2 THEN %LET UCS2B4= $UCS2B4 ; ELSE %LET UCS2B4= $EBCDIC2; so that using "&UCS2B4.00" syntax, SAS V8.2+ will INPUT with $UCS2B400. while earlier INPUT is with $EBCDIC200. Code was revised and validated with J2EE data, Oct 17. Change 19.179 RMF III LENGTH=0 cause INPUT STATEMENT EXCEEDED error, VMACRMFV but a rerun does not fail, (but a rerun against the VSAM Sep 28, 2001 file to create the BSAM file will have different records because RMF III VSAM continuously runs and wraps). This change tests and deletes LENGTH=0 records with a message on the log to show how many records were found. Thanks to John F. Gebert, Office Depo Inc, USA.
VMACDB2 Commit Count, but that should have been a numberic value. VMACDB2H Since I can't change the type from CHAR to NUM without a Sep 27, 2001 serious compatability issue, I have created QWHSLUCN as a numeric variable with the Commmit Count. Thanks to Chuck Hopf, MBNA, USA. Change 19.177 Support for IMS 6.1 Fast Path records (INCOMPAT). VMACIMS Fast Path subcodes 01, 03, 36, 37, and 38 were all VMACIMSA changed by IBM; fields were inserted in all records, and Sep 27, 2001 new variables are now created and kept. Support was then Oct 15, 2001 revised Oct 15 to correct VARIABLE STRTTIME NOT FOUND errors introduced by the first change. Thanks to Pete Gain, SAS, ENGLAND.
EXTY7002 -TYPE30xx - bit 2 in field SMF30SFL (Storage/Paging) new EXTY74DU variable IEFUSIME='IEFUSI*CHANGED*MEMLIMIT?' FORMATS in TYPE30_4, TYPE30_V, TYPE30_5, TYPE30_6. IMAC7072 - new vars MEMLIMIT='MEMLIMIT*VALUE' IMAC74 and SMF30MLS='SOURCE OF MEMLIMIT' which VMAC30 is decoded by new MG030ML format: VMAC7072 1:MEMLIMIT SET BY SMF VMAC71 2:MEMLIMIT SET IN JCL VMAC73 3:MEMLIMIT BASED ON REGION=0 VMAC74 4:MEMLIMIT SET BY IEFUSI VMAC78 but SMF30MLS and MEMLIMIT are set missing (and VMAC79 not zero) if SMF30MLS has no bit set (i.e., if VMAC90 MEMLIMIT is not specified). VMXGINIT -TYPE70PR - Existing SMF70BDA is now divided by SMF70DSA Sep 25, 2001 to calculate Average number of LCPUs Active, as the variable was labelled, and existing SMF70ACS is now documented as 4 bytes (z/OS R1.1 doc had ACS 2 bytes with +2 after MAS; I assume this is an IBM doc correction). SMF70ACS is now divided by SMF70DSA. -TYPE7002 - New Dataset for Cryptographic Coprocessor data contains PCCI Cryptographic Hardware execution times and counts for all operations and all RSA-key-generation operations. -TYPE71 - New Swap Reason "In-Real Swap" is added in new variables SWAPIR/EXTAUXIR/LOGAUXIR/LOGEXTIR/ PHYAUXIR/PHYEXTIR. -TYPE72GO - New variable R723CLSC, only in Reporting Class observations, contains the name of the Service Class that last contributed to this Reporting Class; blank if this obs is for Service Class. Note: Previously Reporting Classes did not have Periods, (only Service Classes had periods), but now both can have up to 8 periods. - New variable R723HETR='Y' if this report class period is "Heterogeneous": If the report class contains transactions from a single service class, R723HETR='N' for a "Homogeneous" reporting class. Report classes now can combine transactions running in different service classes, and those "Heterogeneous" have R723HETR='Y'. - New variables R723CAMU/R723CAMD (CAM CRYPTO) and R723APU/R723APD (AP CRYPTO) contain TCB USING and TCB DELAY samples for Cryptographic Asynchronous Message Processor (CAM) and for Cryptographic Assist Processor (AP). - New R723FQD Feature Queue Delay Samples: a TCB was found waiting on a processor feature queue associated with a CPU, are also included in the delay samples in R723CCDE (MXG PCTDLCDE). Note: Feature Queue Using Samples are in MXG variable PCTUSCUS (R723CCUS). -TYPE7204 - Type 72 subtype 4 delay samples have new var R724OR18 for new "In-Real" Swap Reason delays. -TYPE73 - New CPMF Channel Measurement Groups 3 adds new channel traffic variables for HiperSockets, or the IQDIO, a/k/a IQD (Internal Queued Direct Communication) channel type, a high-speed LAN for communications between LPARs in a CEC/CPC. Eleven variables with byte rates per second, for both LPAR and Totals, for both data unit and message unit traffic, plus unsuccessful sends and receive counts, but there is no capacity measure for the IQD. These new variables exist only for SMF73CMG=3: SMF73PDS SMF73PDU SMF73PMS SMF73PUB SMF73PUM SMF73TDS SMF73TDU SMF73TMS SMF73TUB SMF73TUM SMF73PUS -TYPE74CF - New variables added to TYPE74CF with bit masks of paths available and installed and channel path type acronym for each of 8 possible paths in variables R744FPAS R744FPIS R744FPCM R744FTA1-R744FTA8 -TYPE74ST - New variables for waits, wait duration, and square of wait duration for waiting for peer subchannel, for waiting for peer with reserve held, and for waiting for peer completion in these new variables: R744SCSS R744SCST R744SCTC R744SLSV R744SPES R744SPLN R744SPSS R744SPST R744SPTC R744SRSS R744SRST R744SRTC -TYPE74DU - New remote facility data section in subtype 4 Coupling Facility data for your Duplexed CFs creates new dataset TYPE74DU. -TYPE746x - SKIP calculation corrected for 746BL/746DL. -TYPE747x - While decoding of the subtype 7 FICON Director is in place in VMAC74, no datasets are created until I have test data to understand. -TYPE78CF - IBM removed PCTPTHBY and NRCMPTSM fields, so they will now have missing values. -TYPE791 - Variable R791TTOD now is non-zero and contains "Real Time into Transaction". - Added swap 'SR'='IN REAL' to FORMAT $MG079SR - New R791MLIM variable ASID Memory Limit added. Yüklə 28,67 Mb. Dostları ilə paylaş:
|