* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
default, and create a new TREND library, instead of reuse (so that your TREND libraries can be GDGs, which is required for CA's 11 job recovery product). -Summarization of DB2ACCTB (Buffer used by Plans) should have had WEEK.DB2ACCTB instead of WEEK.DB2ACCT. Thanks to Chuck Hopf, MBNA, USA.
ASUMTMNT alternatively use the STK VSM STCVSM13 and STCVSM16 data Oct 11, 2004 sets as input. Comment blocks must be removed, and you must provide a way to identify the VSM devices. Change 22.256 This change was replaced by Change 22.265. VMAC30
Oct 11, 2004 Change 22.255 MXG format $MGSASPR, which maps the SAS procedure name to FORMATS SAS Product Name, was updated for alternative spellings Oct 11, 2004 (eg., SUMMERY for SUMMARY). The SMF record contains what name was typed in, but when incorrectly spelled, SAS can often "guess" the correct procedure name, leaving the wrong spelling in the SMF record. Thanks to Dr. Alexander Raeder, Itellium, GERMANY.
VMAC42 message caused only the last entry for that VOLSER to be Oct 11, 2004 not output in TYPE42VT dataset; all other entries were correctly output. Three tests for invalid length were revised: each needed "-4" to be added to each test: IF 0 LT S42VTVDO LT COL OR S42VTVDO+S42VTVDL-4 GT LENGTH 0 LT S42VTVXO LT COL OR S42VTVXO+S42VTVXL-4 GT LENGTH 0 LT S42VTVVO LT COL OR S42VTVVO+S42VTVVL-4 GT LENGTH Thanks to John Parkes, Experian, USA.
BLDNTPDB in your source code to be all lower case, but internal Oct 11, 2004 code in BLDNTPDB was incorrect; using upper case text in your invocation caused errors in the Week and Month selection. Thanks to Terry Heim, ECOLAB, USA. Change 22.252 Variable QW0258TS is revised; it is input TODSTAMP8. and VMAC102 formatted as DATETIME21.2, and the divide by 4096 was Oct 12, 2004 removed. -All raw DB2 datetimestamps are on GMT clock, so each must be converted to local by the addition of GMTOFFDB, and then each must then be in a %VMXGTIME call (in case your site uses %TIMEBILD to convert all times to a different time zone). These variables were missing the GMTOFFDB addition: QW0022TS,QW0022BT,QW0125TS,QW0316TS. Also, extraneous refs for QW0260TS,QW0261TS,QW0262TS were removed. Thanks to Christa Neven, KBC, BELGIUM. Change 22.251 Format $MGSTCMR, for the type of mount, had additional FORMATS values '03'x thru '07'x that are now decoded. Oct 10, 2004 Thanks to Gilles Fontanini, ABS Decision, FRANCE. Change 22.250 Support for Release 3.5 of BETA93 product for subtypes FORMATS 0, 1, 2, 3, and 5 has been revised and 0-3 validated: VMACBETA -Dataset BETA1 new variables: Oct 11, 2004 BETAPCR BETADCR BETAFRM BETAEXT BETAPDEF BETAFDEF BETABTKN BETARBTK BETAREPR BETARMOD BETARUSR - However, the new REPRINT variables BETARPUS, BETAREPR, BETABTKN, BETARMOD, and BETARBTK appear to be trashed in the SMF record, producing missing/strange values. -Dataset BETA3 is now valid; most variables were missing. -Dataset BETA5 new variables: BETAFRM BETAEXT BETARPTN BETAWRPT BETALTKN BETARTKN -Some inconsistent variable names (FRM2,FRMX,EXT2,EXTX) were eliminated. -Other subtypes will be validated when sample records are received; some timestamps are now SMFSTAMP8 format, but only real records will reveal their internal formats. -The subtype 50 record still has not been validly created so it is not yet supported. Thanks to Graham Cornwell, Donovan Data Systems, ENGLAND. Thanks to Klaus Messer, COMLAB GmbH, GERMANY. Change 22.249 Support for TDSLink user SMF record creates 26 datasets: EXTDS01F dddddd dataset description EXTDS01I TDS01T TDSL01TC TCP REJECTED PORTS EXTDS01M TDS01I TDSL01IC ICMP EVENTS EXTDS01T TDS01F TDSL01FT FTP EVENTS EXTDS01V TDS01X TDSL01XO XOT EVENTS EXTDS01X TDS01V TDSL01VT VTAM EVENTS EXTDS04 TDS01M TDSL01MV MVS EVENTS EXTDS06 TDS04 TDSL04 XOT CONNECTIONS EXTDS07 TDS06 TDSL06 TCP CONNECTIONS EXTDS08 TDS07 TDSL07 IP INTERFACES EXTDS09 TDS08 TDSL08 LOCAL APPLICATIONS EXTDS0A TDS09 TDSL09 REMOTE APPLICATIONS EXTDS0B TDS0C TDSL0C TCP/IP STACK EXTDS0C TDS0A TDSL0A IP NETWORKS EXTDS0D TDS0B TDSL0B ENTERPRISE EXTENDER/IP NODES EXTDS0E TDS0D TDSL0D CSM BUFFERS EXTDS0F TDS0E TDSL0E PING OBJECTS EXTDS10G TDS0F TDSL0F MQ SERIES QUEUES EXTDS10L TDS10X TDSL10XC XCA HEADER EXTDS10M TDS10G TDSL10GR GROUP EXTDS10P TDS10M TDSL10MX CROSS DOMAIN RESOURCE MGR EXTDS10R TDS10P TDSL10PU PHYSICAL UNIT EXTDS10S TDS10S TDSL10SK SKLETAL PHYSICALUNIT EXTDS10X TDS10L TDSL10LU LOGICAL UNIT EXTDS11 TDS10R TDSL10RX CROSS DOMAIN RESOURCE EXTDS12 TDS11 TDSL11 CPU AND MEMORY TYPETDSL TDS12 TDSL12 DISK TYPSTDSL VMACTDSL
VMXGINIT Oct 7, 2004 Thanks to Khalid Meskinia, SAS France, FRANCE. Thanks to Alain Delaroache, Atlantica (Credit Agricole), FRANCE Thanks to Jacky Hofbauer, Atlantica (Credit Agricole), FRANCE.
ASUMCACH Previously, zero obs caused $CACHID format to not be Oct 6, 2004 created. Thanks to Jeff Harder, Indiana Farm Bureau, USA. Change 22.247 Default test for Oracle SUBSYSID EQ 'OSDI' replaced the VMACORAC SUBSYSID EQ 'TGW9' value, left after debugging. If you Sep 30, 2004 re-define the _IDORACO macro in your IMACKEEP, it will use your subsystem IDs, rather than the default in the VMACORAC member, but 'OSDI' should have been the default in the MXG code. See also change 20.111 text. Thanks to John Rivest, TDS Corporate, USA.
VMACNTSM -MSEXCHANGEDSACCESS PROCESSES object, dataset MSEXPROC Sep 30, 2004 has only seven variables, down from thirty nine. -EPOXY object, dataset EPOXY, has two new variables. -MSEXCHANGEIS MAILBOX object, dataset MSEXCHPU has six new variables. Change 22.245 -Support for Velocity Software's XAMAP Release 3.4, which EXXAMSYT is INCOMPATIBLE, because new SYTCUP segment with LPARNAME VMACXAM of "Totals:" had an invalid value for the number of LCPUs Sep 30, 2004 causing INPUT STATEMENT EXCEEDED RECORD LENTH error if Release 3.4 records are read without this change. -The "Totals:" record is not output, as it duplicates the CPU time in XAMSYT dataset, and caused 200% CPU busy's! That tailoring code is externalized into the EXXAMSYT exit member, if you ever decide you need it, unlikely. -New variables are added to the XAMSYS dataset: CPUCAPAB CPUCFGCT CPUCFGCT CPUCHAR CPUCOUNT CPUCOUNT CPUDEDCT CPURESVD CPURESVD CPUSHARD CPUSTNBY CPUSTNBY DEDCNT IOPCNT LPARCAF LPARNAME LPNUMBER MONECONH MONECONL MONEVNTH MONEVNTL MONSAMPH MONSAMPL MONSCONH MONSCONL SCPCAPAB SYSMMODL SYSMPOM SYSMSEQC SYSMTYPE VL3CAF VL3CFGCT VL3COUNT VL3CPNAM VL3DBCT VL3MNAME VL3RESVD VL3STNBY -New variables added to the XAMSYT dataset: CALPTIS LCUCWCPL LCUCCAPP LCPTYCP LCPTYICF LCPTYIOT -Additional data from SUBSUM, PRCIOP, and IODVSW will be added later, when someone asks to use the new data, or I get bored/caught up. The preceding changes are all that are needed to keep your current reports valid, and add those new variables. This note will be revised when that happens. Thanks to Tony Curry, BMC, USA. Change 22.244 RACF REMOVE event, dataset TYPE8023, did not decode the VMAC80A Data Type 6 segment; the OWNER/GROUP and bit variables Sep 30, 2004 are now created. Thanks to Robert D. Mitchell, UBS, USA. Change 22.243 INPUT STATEMENT EXCEEDED if old VMACNSPY member (prior to VMACNSPY Change 22.131) read a NETSPY record that is a NETMASTER Sep 30, 2004 subtype. NETSPY and NETMASTER are now combined and write Referenced: a single SMF record with both NSPY and NETM subtypes, and EXNETM50 Change 22.131 added support to create both NETM and NSPY VMXGINIT datasets from the single SMF record. However, messages INVALID NETSPY SUBTYPE were printed with Change 22.131, and NETMASTER datasets all had zero observations, because the more robust test for NETMASTER subtype must be first, and the test for a NETSPY subtype can be executed when it is clear the record is not a NETMASTER subtype. Note that TYPENETM will still process NetMaster subtypes in the NETSPY record, but if you have added both NSPY and NETM logic to your BUILDPDB, either VMACNSPY will ABEND, because both want to create the five NETMxxxx datasets. You should remove NETM tailoring and use only NSPY to create both NSPYxxxx and NETMxxxx datasets. -This VMACNSPY requires MXG 22.02 or later, because 22.02 contained Change 22.037, which created the new EXNETM50 member and updated VMXGINIT to support the new NETM5000 NetMaster dataset. Thanks to Andy Creet, Department of Defence, AUSTRALIA. Change 22.242 Example analysis provides Gartner Group with information: ANALGART Batch Workload (e.g., MVS Batch Percentage CPU) Sep 28, 2004 Interactive Workload (e.g., MVS/TSO Percentage CPU) Online Workload (e.g., CICS,DB2,IDMS Percentage CPU) across all partitions data. The example assumes your Batch Work is in WORKLOAD=JES2, TSO in WORKLOAD=TSO, Enclaves in DDF, CICS in CICS, so you may need to tailor those definitions, and then it reads your PDB library and build a tailored "RMFINTRV" dataset named GARTNER, and print a summary report. Thanks to Atle Mjelde, Ergo, NORWAY.
ASUMCIMS PDB.ASUMIMTR : /*SUMMARY OF CIMSTRAN */ VMXGINIT PDB.ASUMIMDD : /*SUMMARY OF CIMSDBDS */ Sep 28, 2004 PDB.ASUMIMDB : /*SUMMARY OF CIMSDB2 */ PDB.ASUMIMPR : /*SUMMARY OF CIMSPROG */ Thanks to Ray Dorsing, Affiliated Computer Services, USA. Change 22.240 Support for z/VM 4.4 INCOMPAT changes in SYTCUP record. VMACVMXA New LCPTYPE='HARDWARE*CPU*TYPE' variable inserted in the Sep 28, 2004 record is now supported. As no other errors with data from z/VM 4.4 have been reported, I can claim "support" with this change, but there could be new data variables that are not yet decoded. Thanks to Reinhard Lidzba, AGIS Allianz Dresdner, GERMANY.
ERRORASC you incorrectly download SMF data, and the member shows Sep 28, 2004 how to use the %UDUMPEBC utility to determine what is in the downloaded file, and how to resolve the cause of the ERROR: Undetermined I/O failure. ERROR: Unrecoverable I/O error detected... This was originally in the undocumented "ERRORS" member, but it contained unprintable characters that caused other problems when that member was downloaded, so the revision updated the contents and removed the unprintables. Thanks to Dan Squillace, SAS Institute, USA.
SYSLOG -Appends the type 'S' continuation record to create a Sep 30, 2004 single event record -Creates LOGGROUP sequence number for multi-record SYSLOG events. See preliminary comments. Please contribute suggestions for enhancements, as this present implementation is still in first tests; some important LOG events may be create new datasets, etc. -Original example completely replaced. Thanks to Freddie Arie, TXU Energy, USA. Change 22.237 -To monitor STK tape devices with ML-31, you must use the VMACTMNT XMEM=YES,MXIT=NO options when you assemble ASMTAPEE, to Sep 27, 2004 capture the job-level information via Cross Memory calls. Instead, for IBM tape devices, MXIT=YES,XMEM=NO is used to eliminate Cross Memory calls and capture all mounts in the IBM Volume Mount Exit - STK's HSC does not use that exit, so you still have to use XMEM, until we can find a usable HSC exit, research long underway. Using the above options for STK caused DDNAME and STEPNR to be blank/zero, because of a logic error in VMACTMNT: The MXIT architecture captures these new variables: ASID INITTIME JOBCLASS LOCLINFO PGMRNAME RACFGRUP RACFTERM RACFUSER STEPNAME STEPNRX TMNTACTN TMNTDEVC TMNTEDUR TMNTFLAG TMNTRCN and these old variables: DDNAME STEPNR ML-31 SMF records have a new segment with those fields, but they are not populated unless MXIT=YES was used; MXG incorrectly INPUT those variables if the segment was detected, causing the DDNAME/STEPNR fields in the new segment overwrote the valid old . That logic was corrected in this change. -PDB.ASUMTAPE had missing values in RACFxxxx and ASID variables because of the TYPETMNT error; that blank DDNAME in TYPETMNT caused ASUMTAPE's merge of TYPETMNT, TYPETALO, and IBM TYPE21 datasets to fail to correctly propagate old ASID variables from TYPETALO into the output PDB.ASUMTAPE dataset. Fixing TYPETMNT fixed. Thanks to Normand Poitras, IBM Global Services, USA.
ANALFLSH in each 15 minute interval, as lots of parallel flashes FORMATS can degrade performance. New format $MG022ST decodes the VMAC22 copy status, and tests for LENGTH-COL were corrected to Sep 26, 2004 use LENGTH-COL+1 for the bytes remaining in the record. Since multiple records are written from multiple systems, the ANALFLSH report must be tailored to select data from only one system, since TYPE22_A dataset can have pseudo duplicates. The sort order for TYPE22_A was changed to SYSTEM SMFTIME SMF22CUA SMF22VOL to remove acutal dupes when TYPS22 is used to sort TYPE22_A dataset. Thanks to Allan ??, ???, USA. Change 22.235 Support for ASG/Landmark TMON for DB2 Version 4.0. VMACTMDB Only one dataset, TMDB7P appears to have been changed. Sep 26, 2004 Thanks to Martin Legendre, Regie des Rentes du Quebec, CANADA. =Drove to East Coast(NYC, NJ, DEL) then drove thru Hurricane Ivan in GA =enroute to FLA (Wakulla County), then drove thru Hurricane Ivan again =one week later in Shreveport, LA. Rain was 6 inches per hour, for =about an hour both times, but traffic only slowed to 60 mph. Change 22.234 Support for DB2 V8.1 IFCIDs 140,141,142,145 SQL text that VMAC102 was relocated; without this change the QW014nTX variable Sep 9, 2004 was blank even when there was SQL text in the record. Thanks to Ep van der Es, Dow, BELGIUM. Change 22.233 MXG's calculation of NRCPUS when IRD was NOT active could VMAC7072 be very slightly wrong, calculating 2.0020 for 2 engines. Sep 1, 2004 The use of NRCPUS=ROUND(SMF70ONT/DURATM,.001) was the cause, but by replacing that calculation with IF SMF70ONT GT 0 AND NRCPUS GT 0 AND LPARWLMG='Y' THEN NRCPUS=SMF70BDA; the correct IRD calculation (LPARWLMG='Y') is done, and for non-IRD, MXG had correctly counted the integer number of engines, which is unperturbed when LPARWLMG NE 'Y'. Thanks to Al Sherkow, I/S Management Strategies, Ltd., USA. Change 22.232 CMR time is added to the calculation of MPL74 time: VMAC74 MPL74=SUM(DEVCONTM,DEVDISTM,DLYCMRTM)/DURATM; Sep 1, 2004 Originally, errors in the CMR field in the record caused it to be left out, but APARs have corrected the value. This agrees with Cathy Cronin's "FICON and FICON Express Channel Performance Version 2.0" IBM White Paper. Thanks to Ray Waugh, Merrill Lynch, USA.
EXNTDB2A RUNWEEK=WTD caused VARIABLE NOT IN DATASET errors. The EXNTDB2D new PDB.DTSCPULP was build from two datasets by did not FORMATS have a label. IMACNTSM -FORMATS was updated with the $MGDDDDD and $MGDDDSN for VMACNTSM all of the new NTSMF datasets. VMXGINIT -DATABASE object with 170 fields support added. Aug 31, 2004 -MSEXCHANGEIS object with 126 fields support added. Sep 8, 2004 -DB2 DATABASE MANAGER (originally DB2 NT DATABASE MANAGER) object with 28 fields is support added. -New objects DB2 DATABASE and DB2 APPLICATIONS with 94 and 90 fields respectively create NTDB2DAT,NTDB2APL datasets. Thanks to Terry Heim, ECOLAB, USA.
VMAC110 than blanks when there was no pool name. MXG now tests Aug 30, 2004 and sets the variable to blanks instead of hex zeroes. Thanks to Harald Seifert, HUK=COBURG, GERMANY. Thanks to Don Deese, (CPExpert), Computer Management Sciences, USA.
BUILD005 printed the first time BUILDPDB/BUILDPD3 were executed, BUIL3005 with nothing in the //SPIN library, but those messages Aug 26, 2004 had no impact. Compiler fakers suppress the messages. Change 22.228 Second instance of INCLASS in the KEEP list should have VMACTPMX been INCLASSJ. Aug 26, 2004 Thanks to Bruce W. Christopher, Housholde International, USA. ====== Changes thru 22.227 were in MXG 22.09 dated Aug 25, 2004========= Change 22.227 Typo left a start-comment /* as the last line of member VMAC99 VMAC99, which caused problems. Stray text was deleted. Aug 25, 2004 Thanks to Chris Weston, SAS ITSV Development Cary, USA. Change 22.226 Back-level ASG/TMON Version 2.0 records caused INVALID DO VMACTMO2 LOOP CONTROL because I forgot that SAS won't tolerate a Aug 25, 2004 missing value and had added DO _I_=1 TO TAAMQCNT; without testing that TAAMQCNT had been input; it was a new field in ASG/TMON Version 2.3 that was missing when 2.0 records were read. All the DO _I_ statements are now protected. Thanks to Nori Abakah, Progress Energy, USA.
ANALDB2R FOUND (but a $MGD140O did exist in the format library), Aug 25, 2004 because the variable that was PUT with $MGD140O format did not exist, and thus was created as numeric, causing SAS to look for a numeric rather than a character format. The variable did not exist because not all datasets that ANALDB2R expected were in the PDB library. If you use %ANALDB2R(PDB=SMF...), it creates all needed datasets for the requested reprots; if %READDB2(INFILE=SMF,IFCIDS=...) is used to create a PDB, and all IFCIDs are not requested then using %ANALDB2R(PDB=PDB ..) may fail. You can use %READDB2(INFILE=SMF,CLASS=AUDIT...) to create a PDB with all needed datasets for AUDIT reports (and that PDB will have zero obs datasets for the IFCIDs with no records, with all datasets created, all variables exist. Seeing this exposure, a "compiler faker" was added for each of the character variables that are PUT with FORMAT, forcing those variables to be character variables, to circumvent the error when all expected datasets are not present. Thanks to Sharon Nagy, Comerica Bank, USA. Change 22.224 SMF 6 or 57 record with ESS segment, but with SMF57TUL=0, VMAC57 was not expected, caused INPUT STATEMENT EXCEEDED error. VMAC6 Since there was an ESS segment, text was assumed present! Aug 24, 2004 Now, the text length is tested. This text was revised Oct 26, 2004 Oct 26, when same error did occur with SMF 6 record. Thanks to Scott Wiig, US Bank, USA. Thanks to Jurgen Strauch, Zweckverband Kommunale Stuttgart, GERMANY.
VMACNDM The CDI DSECT does not agree with the record; the QTYPE Aug 24, 2004 field is only 1 byte in the record but the DSECT shows a Aug 26, 2004 halfword. The Queue Status contains a valid 'SS', and the SEQ field contains '0000'x, but the error was because MXG always expected that was followed by an 8-byte field with SYSPLEX SERVER NAME, but that field does not exist in all releases of NDM. Code was added to conditionally input SERVER NAME only when it exists for the QE, TP, PI and FA subtypes, and all with SERVER NAME field. -Invalid value for OFFSET to JOB DATA SET NAME protected. Thanks to Gerrit Van Meerbeek, HP/Procter and Gamble Cie, BELGIUM. Change 22.221 Support for z/OS 1.6, but now, Change 22.272 is required. VMAC30 Change 22.180 and prior APARs supportd most z/OS 1.6 new VMAC7072 data and changes in MXG 22.07 and 22.08, but TYPE72GO BUILD005 code (ONLY FOR z/OS 1.6 RECORDS) was wrong, causing many BUIL3005 INVALID DATA messages and incorrect output in TYPE72GO Aug 25, 2004 when z/OS 1.6 records were read (NO errors with 1.5 RMF). The SKIP=SKIP-16 needed to be SKIP=SKIP-28 in SUBTYPE=3 code block for LENSCS GE 532. -The extensive MXG Technical Note on IFAs/zAAPs in MXG Newsletter FORTY-FIVE discussed MXG revisions to some of the existing MXG variables; this change implements those new calculations and new variables for TYPE72GO/TYPE70: -In TYPE72GO: CPUIFATM - IFA CPU Time spent on IFA CPUIFETM - CP CPU Time that was Eligible to run on IFA IFAUNITS - IFA Service Units spent on IFA IFEUNITS - CP Service Units that were Eligible for IFA CPUTCBTM - CP CPU Time (does NOT contain IFA CPU time) CPUUNITS = CP Service Units (NOT contain IFA Units) -It is likely the VELOCITY calculations in TYPE72GO will be changed, but this is just a note that they have NOT been revised to include IFA samples. -TYPE70 dataset was revised to include only CP engines in existing PCTCPUBY, CPUACTTM, etc, and labels now include "CP" to identify. New IFA Capacity Variables added. PCTCPUBY,CPUACTTM,CPUUPTM,CPUWAITM are all "CP-only". PCTIFABY,IFAACTTM,IFAUPTM,IFAWAITM are all "IFA-only". Individual PCTIFBYn and IFAWAITn variables exist for each of the possible 32 IFA engines. IFAACTTM='IFA ENGINE*EXECUTING*(ACTIVE) CPU TIME' IFAUPTM ='IFA ENGINE*AVAILABLE*(UP) TIME' NRIFAS ='NUMBER OF*IFA*ENGINES*AVAILABLE' PCTIFABY='ALL IFAS*PERCENT*BUSY (DISPATCHED)' IFAWAIT0-IFAWAITX - IFA wait for each of 32 engines PCTIFBY0-PCTIFBYX - IFA PCT BUSY for each engine. -This data has been tested with z/OS R 1.6 RMF 72 records, but not with IFAs installed, so please validate the old and new percentages when you install an IFA. Change 22.272 updates have been tested with 30s and 72s. -TYPE30 variable's names were changed to be consistent: Yüklə 28,67 Mb. Dostları ilə paylaş:
|