* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
The logical AND results in more restrictive filtering because 2 conditions must be met for an OPD entry to be selected.
OPDJOBNAME=ABC* OPDUSERNAME=SAM* selects Address Spaces in the RMF Monitor III OPD table that have a Job Name starting with 'ABC' OR a User Name beginning with 'SAM'. If the Address Space does not match either selection it is filtered and will not appear in the result MXG PDB. The logical OR results in less restrictive filtering than Example 1 above because any of the 2 conditions results in data selection of an OPD entry. -The JOBNAME= (alias JOB=) keyword for multi-table selection is expanded to include job names from the OPD table as well as the ASI and CSR tables. This is a convenience feature to avoid having to code the Job Name parameter three times when the same job names from all three tables are of interest. -The ASI, CSR, and OPD tables must all be selected for the JOBNAME= multi-table selection keyword to function completely. Otherwise only entries from selected tables are filtered. Note that most RMF III tables do not contain common character data fields, but in this case the ASI, CSR, and OPD tables all do contain a Job Name. JOBNAME= Examples: JOBNAME=ABC88888 selects only address spaces with a Job Name of 'ABC88888' in either ASI, CSR, or OPD tables and is equivalent to coding: ASIJOBNAME=ABC88888 CSRJOBNAME=ABC88888 OPDJOBNAME=ABC88888 JOBNAME=ABC:ABC88888 selects only address spaces with a Job Name that is GE 'ABC' and LE 'ABC88888' in either ASI, CSR, or OPD tables and is equivalent to coding: ASIJOBNAME=ABC:ABC88888 CSRJOBNAME=ABC:ABC88888 OPDJOBNAME=ABC:ABC88888 JOBNAME=.* selects only address spaces with a Job Name that begins with a national character in either ASI, CSR, or OPD tables and id equivalent to coding: CSRJOBNAME=.* OPDJOBNAME=.* -Some RMFV001I Execution Environment messages have been reformatted to include DFSMS/MVS version, CPC Name, LPAR Name (if not a VM Guest), or VM Userid (if a VM Guest). TIOT statistics are now grouped on the same message line. -The Creation date was not valid when non-VSAM data set was incorrectly provided as a RMF III data set. CRDATE is removed from the RMFV008I message in this case. -The MXG00 record version is raised to x'07' from x'06'. New fields added to the MXG00 record are: IPL timestamp in LOCAL and GMT time IPL volume serial TIOT size in K and bytes TIOT maximum and used entries CPC Name, LPAR Name, and VM UserId DFSMS/MVS level -Several documentation Sections are updated to support the above changes: Section 5 "Input Data Selection Parameters" Section 12 "Messages" Section 13 "Filtered Records" Section 25 "Ranges and Patterns" Section 30 "Summary" -VMACRMFV was updated to add new variables to ZRBASM: ASMDFLVL='EXECUTION*DFSMS/MVS*LEVEL' ASMOPNRX='MAXIMUM*OPDPROCNAME*RANGES' ASMOPNPX='MAXIMUM*OPDPROCNAME*PATTERNS' ASMOJNRX='MAXIMUM*OPDJOBNAME*RANGES' ASMOJNPX='MAXIMUM*OPDJOBNAME*PATTERNS' ASMOUNRX='MAXIMUM*OPDUSERNAME*RANGES' ASMOUNPX='MAXIMUM*OPDUSERNAME*PATTERNS' ASMOJNRS='SIZE*OPDJOBNAME*RANGE*TABLE' ASMOJNPS='SIZE*OPDJOBNAME*PATTERN*TABLE' ASMOPNRS='SIZE*OPDPROCNAME*RANGE*TABLE' ASMOPNPS='SIZE*OPDPROCNAME*PATTERN*TABLE' ASMSPGAO='SPG*MULTI*FILTER*LOGIC*I/O?' ASMSINDD='SYSIN/SYSINA*DCB*DDNAME' ASMVFREE='FREE=CLOSE*OPTION?' ASMSINMG='SYSIN*MEMBER*OR*GENERATION' ASMOUNRS='SIZE*OPDUSERNAME*RANGE*TABLE' ASMOUNPS='SIZE*OPDUSERNAME*PATTERN*TABLE' ASMIPLTL='LAST IPL*LOCAL*TIME' ASMIPLTG='LAST IPL*GMT*TIME' ASMTIOTB='MAXIMUM*TIOT SIZE*BYTES' ASMTIOTK='MAXIMUM*TIOT SIZE*IN K' ASMTIOTX='MAXIMUM*TIOT*ENTRIES' ASMTIOTU='CURRENT*TIOT*ENTRIES*IN USE' ASMCPCNM='CPC*NAME' ASMLPARN='LPAR*NAME' ASMVMUID='VM*USERID' ASMIPLVL='IPL*VOLUME*SERIAL*NUMBER'
VMXGRMFI to SYSOTHER detection, since if the CPUTM is 0 it cannot Feb 28, 2017 impact totals, but workload names and descriptions are identified so you can find the culprit, since nothing should ever fall thru to Service Class SYSOTHER.
TRNDVMXA the new VMXGSUM correctly supported the old syntax with Feb 28, 2017 no error.
VMAC120 enhancements, adds these variables to TYP12012 dataset: Feb 28, 2017 SM120CDO='REFERENCE*TYPE' MG120CD format decodes: 1='1:READER' 2='2:PROCESSOR' 3='3:WRITER' 4='4:CHECKPOINT' 5='5:BATCHLET' 6='6:PARTITION_MAPPER' 7='7:PARTITION_REDUCER' 8='8:PARTITION_COLLECTOR' 9='9:PARTITION_ANALYZER' 10='10:DECIDER' SM120CDU='PHYSICAL*CPU*ADJUSTMENT*RCTPCPUA' SM120SU_SEC='CPU*RATE*ADJUSTMENT*RMCTADJC' SM120CDW='REPOSITORY*TYPE*JPA* OR MEM' SM120CDX='JOB*STORE*REF*ID' SM120CDY='SM120CDY*FLAGS' Next six variables are only valid in step end record SM120CDZ='STEP*START*LIMIT' SM120CEA='CHUNK*STEP*CHECKPOINT*POLICY' decodes: 0='0:ITEM' 1='1:CUSTOM' SM120CEB='CHUNK*STEP*ITEM*COUNT' SM120CEC='CHUNK*STEP*TIME*LIMIT' SM120CED='CHUNK*STEP*SKIP*LIMIT' SM120CEE='CHUNK*STEP*RETRY*LIMIT'
VMXG70PR to correctly populate all variables. Each SMF 70 record Feb 27, 2017 contains a "This System" segment that populates TYPE70, and an "LPAR Segment" for each LPAR, for TYPE70PR, so the LPAR data can be reported & summarized from a SMF 70 from only one system, but then all the "This System" variables are wrong, notably, SMF70LAC, the IBM 4HR AVG MSU, which will contain ONLY the MSU from the one "This System". This change compares TYPE70 and TYPE70PR to detect if there are missing TYPE70 or TYPE70PR data, printing a a PROC FREQ with missing systems identified, and printing a log message that SMF70LAC will be wrong. Change 35.049 Support for MAINVIEW FOR IMS 5.3 a/k/a IMF or CIMS which VMACCIMS COMPATIBLY added these variables to the CIMSTRAN dataset: Mar 12, 2017 TRNMSYS ='MQ*REMOTE*SYSTEM*NAME' TRNDSYS ='DB2*REMOTE*SYSTEM*NAME' TRNFLTRD='SOME*CALLS*NOT TRACED*FILTERS?' TRNPRELD='PGM*WAS*PRELOADED' TRNINFL ='TRN WAS*CAUGHT*INFLIGHT' Was reserved change number until Sep 23, 2018 Change 35.048 Support for IWS Version 9.3, a/k/a TWS and was OPC, which VMACOPC replaces subtype 23 with new subtype 66 with the original Feb 24, 2017 variables plus these four new variables TRLDURS23='DURATION' TRLOID23 ='OPERATION*ID' TRLOLDST23='OLD*STATUS' TRLREADY23='START*DATETIME*WAIT*OPR' The new subtype 66 record is output in the OPC23 dataset so your reports won't have to be changed. Thanks to Teuvo Virsu, TIETO, Change 35.047 Support for IFCID 316 ACCESS CONTROL AUTH EXIT PARMS. VMAC102
Feb 24, 2017 Change 35.046 Support for IFCID 125 Truncated Package Collection and FORMATS Package Name fields, and new variables for Runtime VMAC102 Adaptive Index in T102S125 dataset: Feb 20, 2017 QW0125TI='INDEX*PROBING*RIDS IN*INDEX' QW0125QI='INDEX*PROBING*RIDS*IN*KEYRANGE' QW0125_TRSN='REASON*LEG*WAS*TERMINATED?' QW0125_PRSN='REASON*LEG*NOT*PROBED?' QW0125_ORSN='REASON*LEG*WAS*REORDERED?' QW0125_FRSN='REASON*LEG*WAS*MARKED FULL?'
ANALDB2R 4.1 to print full 10.1 Release value in reports. Feb 20, 2017
VMACRMFV missing due to an invalid MXG test for LENLEFT. Feb 20, 2017 -New variables found in the Dec 2016 Programmers Guide: Dataset ZRBCPU: CPC_CECNAME='CPC*CEC*NAME' LPARHWGR='LPAR*HW*GROUP*NAME' Dataset ZRBLCP: LCPUHWLW='HW*GROUP*CAP*LIMIT' LPARHWGR='LPAR*HW*GROUP*NAME' -APAR OA58688 adds these new fields. Thanks to MP Welch, Bank of America, USA.
VMAC74 bytes that don't "print pretty". Variable SMF74NIDTWO Feb 20, 2017 keeps those two bytes, formatted $HEX4. for printing. Thanks to Pierre Pascal Joulin, Societe Generale, FRANCE. Change 35.042 Sample code that creates charts of resource group CPU GRAFCAPS usage and capping. Feb 17, 2017
FORMATS map that value in variable DSGFTYPE. Feb 17, 2017 Thanks to J. Alan Gray, CareFirstBlueCrossBlue Shield, USA. Thanks to Stanley M. Helms, CareFirstBlueCrossBlue Shield, USA.
VMAC1415 the UCB segment causing VOLSER to be truncated to two Feb 17, 2017 characters. No code change to support the corrections.
VMACVMXA records which have a new BEGINMTR value for each (hour) VMXGINIT period, which caused the first interval of each (hour) to Feb 15, 2017 be lost, because normal MONWRITE records have a single BEGINMTR value for each file. But the ZWRITE records are contiguous in spite of changed BEGINMTR, so this support is enabled with %LET MXGZWRITE=YES; in SYSIN which will set BEGINMTR only from the first instance so only the very first interval is lost per day. // EXEC MXGSASV9 //VMINPUT DD DSN=YOUR.ZWRITE.MONWRITE.DATA,DISP=SHR //PDB DD DSN=YOUR.ZWRITE.PDB,DISP=OLD //SYSIN DD * %LET MXGZWRITE=YES; %INCLUDE SOURCLIB(VMACVMXA,IMACKEEP); _TESTVM /*READS VMINPUT */ Thanks to Scott Barry, SBBWorks Inc., USA.
FORMATS VALUE MG116QT VMAC116 1='1:LOCAL' Feb 15, 2017 2='2:MODEL' 3='3:ALIAS' 6='6:REMOTE' 7='7:CLUSTER' -GMTOFF116 calculation revised correcting occasional missing values in WQCLOSTI, WQOPENTI, and WQTTTIME. (WQCLOSTI can be validly missing). -APAR PI70580 corrects invalid WQBASENA variable values, which seem to occur in every second segment in each SMF record, but other segment's values are valid. Thanks to Raymond Smith, Optum TECH, USA. Thanks to Pietro Rosella, Canadian National Rails, CANADA
CICINTRV had been created, intending to delete ONLY CICS Stats VMXGCICI datasets to free up //WORK space for subsequent use, but Feb 14, 2017 if CICINTRV was intentionally left in //WORK, it was then unintentionally deleted by that colon modifier, and ITRM expected it to be left in WORK as it had been previously. Since no one had actually asked for this cleanup, it has been removed from VMXGCIC. But, added at the bottom of the CICINTRV member, inside a comment block, is the code to delete all of those CICS Stats, if you do wish to. Thanks to Don Barnard, North Carolina State Government, USA. Thanks to Chris Weston, SAS Institute ITRM, USA. Change 35.037 ASUMDB2P expected variables QPACDBRM/QPACPACK would be ANALDB2R populated, but those bits were removed in DB2 V10, so now ASUMDB2P PACKTYPE is blank. Feb 14, 2017 -ANALD2R was not correctly rolling up the control break totals in the Accounting SHORT report.
VMXGSUM use CLASS NWAY and why it cannot. There are two cases: Feb 11, 2017 - Use of DESCENDING in the SUMBY - same dataset name for input and output and no OUTCODE specified It will also now display the final setting of SUMBYCLASS. Thanks to Tim Hare, Southwood Shared Resource Center, USA.
VMACTPX TPX07LEN=93 but only 44 bytes remain in the record. MXG Feb 10, 2017 silently deleted the record, because of prior invalid 07x Feb 14, 2017 causing zero obs in TPXAPLON Logon dataset. The first 44 are now INPUT, and the remainder conditionally input. -Feb 14: Correction for undocumented 8 byte insert in '06' and a blank in TPXSNAME. -Feb 14: Each pair of subtype 01 TPXSTART records have the same SMF time, but the second record is a continuation of the first record, which is not supportable; a problem report will be opened with TPX Support. -TPX PTFS R088919 and R085818 correct some errors, while CA Fix TR95030 corrects the bad subtype 1 records which turned out to subtype 2 records with wrong subtype. Thanks to Scott Wiig, USBank, USA. Thanks to Paul Volpi, UHC, USA. Change 35.034 Support for the BBMQ large segment record structure that VMACBBMQ are created by the BMC BBM9MD73 utility program that Feb 10, 2017 extracts the records from the history file for TYPEBBMQ to then process. Both old and new format records are supported with this change and there were no changes to the MXG datasets. Change 35.033 Logic to determine the begin/end of month was robusted VMXGALOC and non-zero length MNTHKEEP will always display the Feb 9, 2017 MONTH Libname value.
VMACDCOL and STEP and the TIME of the Creating JOB for DISP=NEW Feb 9, 2017 datasets, but the EATTR option must be specified either in the DATACLASS definition or with a DD statement. Thanks to Paul Newton, IBM RDP Dallas, USA.
VMACDB2 to dataset TYPE42DS, having been overlooked. Feb 9, 2017 Thanks to Stephen Hoar, Lloyds Banking, ENGLAND Change 35.030 DB2 statistics dataset DB2STAT4 QW0225_LMWRITE_REAL and VMACDB2 _QW0225_LMCTYRL_REAL were incorrectly very large due to Feb 7, 2017 a 4-byte misalignment. Thanks to Giuseppe Giacomodonato, EPVTECH, ITALY. Change 35.029 RACF SMF70DTP/RACFTYPE=6 segment was increased in length VMAC80A from 124 to 136 but had not been protected for a change. Feb 7, 2017 Three additional keyword variables ADDLKEY1-ADDLKEY3 are added to TYPE8010 and TYPE8013 datasets. Thanks to Coen Wessels, GTS Infrastructure, SWITZERLAND.
ASMRMFV that were added by z/OS 2.1 but not captured by ASMRMFV. VMACRMFV Variables EDEPCKG EDEPROC EDEUSER EDETRXN ECEACCT were Feb 6, 2017 increased in length and these new variables are kept: EDESCHEDENV ='SCHEDULING*ENVIRONMENT*NAME' EDESCHEDENVLN ='S E NAME LENGTH' EDESUBSYSCOLLECT ='SUBSYSTEM*COLLECTION*NAME' EDEPCKGLN ='PACKAGE*NAME*LENGTH' EDEPROCLN ='PROCEDURE*NAME*LENGTH' EDECLIENTIPADDR ='CLIENT*IP*ADDRESS' EDECLIENTIPADDRLN ='CLIENT*IP*ADDRESS*LENGTH' EDEUSERLN ='CLIENT*USERID*LENGTH' EDETRXNLN ='CLIENT*TRANSACTION*NAME*LENGTH' EDECLIENTWRKSTATION ='CLIENT*WORKSTATION' EDECLIENTWRKSTALN ='CLIENT*WORKSTATION*LENGTH' EDEACCTLN ='CLIENT*ACCOUNT*LENGTH' Thanks to Scott Barry, SBBWorks Inc., USA.
IMACDBNZ Q8ACINSC='INSERT*STATEMENTS*SENT TO IDAA*FROM DB2' Feb 6, 2017 Q8ACUPDC='UPDATE*STATEMENTS*SEND TO IDAA*FROM DB2' Q8ACDELC='DELETE*STATEMENTS*SEND TO IDAA*FROM DB2' Q8ACDRPC='DROP*STATEMENTS*SEND TO IDAA*FROM DB2' Q8ACCRTC='CREATE*STATEMENTS*SEND TO IDAA*FROM DB2' Q8ACCMTC='COMMIT*STATEMENTS*SEND TO IDAA*FROM DB2' Q8ACRBKC='ROLLBACK*STATEMENTS*SEND TO IDAA* FROM DB2' Q8ACOPNC='OPEN*STATEMENTS*SEND TO IDAA*FROM DB2' Q8ACROWI='ROWS*INSERTED*TO IDAA*BY DB2' Q8ACROWU='ROWS*UPDATED*ON IDAA*BY DB2' Q8ACROWD='ROWS*DELETED*ON IDAA*BY DB2' Q8ACROWC='ROWS*RETURNED*BY IDAA*TO DB2' These variables are output in DB2ACCT, but they appear to be defective, as they are supposed to be ACCUMULATED but the 2012 and 2013 test data I have has breaks in the expected monotonic increase, so if you are interested in these fields, please send current SMF 101 data so I can investigate if the accumulation is now valid.
UTILRMFI are printed when SMF 72 records are processed. SYSOTHER Feb 1, 2017 should never happen; it is the fall thru service class when WLM can't classify work and runs at the lowest DPRTY in MTTW mode, and thus should NOT ever happen! Now, when UTILRMFI is run to examine the problem, it will also read the PDB.SMFINTRV or PDB.TYPE30_4 dataset to find what tasks were classified into SYSOTHER, reporting JOB name, READTIME, JESNR, and SRVCLASS and RPTCLASS. If there are type 30 records they will be reported but there may not be any, if no tasks actually went to the service classes, or the workload is one where there is no type 30 record (e.g., DDF). All workloads in your WLM classification rules should have a default service class SPECIFIED: -Unclassified work will default to one of two places - Started Tasks default to SYSSTC - All other work defaults to SYSOTHER Neither is a good choice. SYSSTC runs at very high DP and SYSOTHER runs at very LOW DP. While very low may be appropriate for workloads you do not know, very high is almost certainly not. -Reports 1 thru 3 already exist. -Report 4 is added to give you the job names, read times jes numbers, service and report class where the service class is SYSOTHER, from 30_4 and SMFINTRV. -Report 5 is added to show you any DB2ACCT records that may have landed in SYSOTHER as they may not be in a type 30 record. -Report 6 is added to show you all tasks falling into SYSSTC, from 30_4, SMFINTRV and 30_6. -Report 7 is added as a table of CPU consumption by service class and system, from TYPE72GO. -Report 8 is added as a table of CPU consumption by report class and system from TYPE72GO. Change 35.025 Using the _VMINPUT macro to read VB z/VM MONWRITE data VMACVMXA incorrectly set the length of VMDUSER to only one byte. Feb 1, 2017 Thanks to Scott Barry, SBBWorks Inc., USA. Change 35.024 SMF 120 Subtype 9 variables SM1209EV/EW/SM1209FI are not VMAC120 kept. In TYP1209U detail dataset, they are output either Feb 3, 2017 in new variables SM1209xxEJBDET or SM1209xxWEBDET, and in datasets TYP1209R and TYP1209N they are summed and output in SM1209xxEJB and SM1209xxWIB variables. The TIME format was removed from the EW count variables. Thanks to Giuseppe Giacomodonato, EPVTECH, ITALY.
UTILEXCL only the NEW dictionary records read from SMF are used Jan 29, 2017 to create the IMACEXCL, and old dictionary records are lost; only the new records are output in PDB.CICSDICT. (The step DATA _LCICDIC; SET _WCICDIC; to add the LABEL was changed to SET _LCICDIC to correct this error.) Thanks to Matthew Chappell, QLD Dept Transport Main Roads, AUSTRALIA
VMXGSUM invocation of VMXGSUM (pre MXG Version 21 example!) can Jan 29, 2017 cause VMXGSUM, which is used EXTENSIVELY internally in MANY MXG members, to fail, sometimes with only a message WARNING: VARIABLE QWACBSC ALREADY EXISTS ON WORK.MXGSUM3 or it can ABEND with ERROR FUNCTION COMPBL TOO FEW ARGS. Only three sites reported the error with MXG 35.01/34.34. Primary exposure was this syntax, SUMBY= . . .DATETIME . . . , ID= . . . QWACBSC . . . , DATETIME= QWACBSC, which caused the output dataset variables QWACBSC and DATETIME have missing values. The correct syntax replaces DATETIME in the SUMBY= : with the DATETIME= variable, and removes the DATETIME= variable from the ID= argument, This error was exposed in MXG 34.05 in Change 34.151 for the CLASSNWAY update that is also corrected. But, even though unlikely, this change detects the old syntax with DATETIME in SUMBY= argument, changes to correct syntax, and tells you what was done for you in a log note. If you had DATETIME in the SUMBY= list and also did NOT explicitly use the DROPDT=NO option, then DATETIME variable is kept in the output dataset. -Unrelated, this change adds the FLORCEIL parameter to VMXGSUM so you can create interval start or end times as you can do in VMXGDUR. Setting FLORCEIL=CDIL sets the END time or FLORCEIL=FLOOR sets the START time, and the label indicates START or END. -Note: this internal MXG Change could be INCOMPATIBLE with programs that worked perfectly previously; send your code VMXGSUM invocation and we will update your code. Thanks to Paul Volpi, UHC, USA.
VMAC78 and R782GFFRxxx are incorrect; R782LSMOMIN should have Jan 27, 2017 been INPUT before R782LSMONTME, but statement was lost causing R782LSMOMIM UNINIT message. Thanks to Paul Naddio, FISERV, USA.
VMXGSUM note has no impact; the VMXGDUM in 35.01 is correct, but Jan 25, 2017 the VMXGVERS call was not updated with '35.01' text.
VMACXAM that caused INVALID SEGMENT record, XAMSYT dataset. Jan 26, 2017 -Support for new SYTLC3 segment in XAMSYS records Mar 2, 2017 was added on March 2, adding these new variables: CALLCKID='CALL*CHECKID' INDEX ='TO MATCH*HISTORY DATA' SECONDS ='SECONDS' CALXSCNT='TOTAL*SPIN*TIMES*EXCLUSIVE' CALXTIME='TOTAL*SPIN*TIME' CALSSCNT='SPIN*TIMES*SHARED' CALSTIME='SPIN*TIME*SHARED*MODE' CALCADSH='CAD*INSTRUCTIONS*OBTAIN*LOCK' CALCADEX='CAD*INSTRUCT*OBTAIN*EXCL LOCK' -Support for new HSTME2 segment in XAMTCP was added on March 2, changing only the length of DESCR to 60 bytes. Thanks to Patricia Hansen, ADP, USA.
GRAFWRKX WARNING but did not cause an error, but ZIPTM, IFATM, and Jan 26, 2017 ZIETM were not being properly summed, causing the ZIP ZAP Yüklə 28,67 Mb. Dostları ilə paylaş:
|