* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
be %INCLUDEd, while the code to use in the alternative EXPDBINC member is one or more %INCLUDE statements. -UTILBLDP sets &EPDBINC if USERADD=xxxx was specified when it created its output SYSIN file, so this error could cause a perfectly-running "BUILDPDB" that was built by %UTILBLDP to fail when 27.10 is 'dropped in'. Copy the EXPDBINC member into your tailoring library and remove the &EPDBINC at the bottom, and your job will then run. (Fortunately, the error is a syntax error at the start of the BUILDPDB run, so it is fully restartable.) Thanks to Tom Kelman, Commerce Bank of Kansas City, USA.
ANALRMFR most fields; averages of averages were printed, and the Dec 20, 2009 CPU Busy included Specialty Engines when it shouldn't. Dec 31, 2009 Thanks to Lisa L. Lawver, Land's End, USA.
ANALJOBE chosen jobs and prints the sequence of events in the life Dec 18, 2009 of a job. The MACJBCK selection example tested JESNR, but JESNR doesn't exist in many job-related records that are read. Comments were revised for selection choices. -The logic was restructured to eliminate uninitialized and missing values, and durations of DSENQTM, LOADTM, and the SELAPSTM are correctly printed with associated events. Thanks to Douglas C. Walter, Citigroup, USA.
VMAC80A protection for unknown TOKDANAM revised to prevent the Dec 16, 2009 STOPOVER ABEND. Thanks to David Schumann, Blue Cross Blue Shield of Minnesota, USA. Change 27.330 -Variable SMF82PDK now input as VARYING4096, vice 256 and VMAC82 longer length is protected if SMF82PLL is GT 4096. Dec 16, 2009 -SMF82PTA is input as $EBCDIC8. vice $CHAR8. and no longer VMAC80A formatted $HEX. Jan 17, 2009 -SMF82RKN is input as $CHAR64. vice $CHAR16 and format is updated.
Thanks to Matthew T. Chappell, Queensland Dept. Transport, AUSTRALIA. Change 27.329 MXG 27.10 only, and ONLY if _S110ST is used to only sort VMAC110 the statistics dataset. ERROR: CICIDNNR NOT FOUND because Dec 15, 2009 that variable was not kept in the new CICSIDND data set. But the two new Identity datasets, CICSIDND and CICSIDNT should not have been in the _S110ST "sort statistics" macro; they are from subtype 1, not subtype 2, records, and they are NOT sorted in _S110, as they can be large volume datasets. Four other non-statistics dataset from subtype 1 were also removed from _S110ST sort macro. These subtype 1 datasets are not sorted by MXG: /*_SCICRDS - CICSRDS IS NOT SORTED, CAN BE HIGH VOLUME*/ /*_SCICRDS - CICSRDS IS NOT SORTED, CAN BE HIGH VOLUME*/ /*_SCICRDF - CICSRDFI IS NOT SORTED, CAN BE HIGH VOLUME*/ /*_SCICRDQ - CICSRDQU IS NOT SORTED, CAN BE HIGH VOLUME*/ /*_SCICIDN - CICIDNTY IS NOT SORTED, CAN BE HIGH VOLUME*/ /*_SCICIDD - CICIDNDD IS NOT SORTED, CAN BE HIGH VOLUME*/ unless you choose to add their _Sdddddd invocation in the EXPDBOUT (for BUILDPDB), or after your TYPE110 include. The Pdddddd/Wdddddd resets were also removed from the _CICSTAT and _CICSTAS macro definitions for consistency. Thanks to Glenn Bowman, Wakefern, USA.
ASUMMIPS interval of your summarization, e.g. 1 Hour, so it does Dec 6, 2009 not report if there were less than a full hour input (i.e., the last interval). New variable DUR70 created in PDB.RMFMSUSE contain the actual duration of the TYPE70 records that were summarized, so you can delete those short intervals with DUR70 LT DURATM, or use DUR70 to identify incomplete (or possibly outage) intervals. -ASUMMIPS now works with &KEEPALL=NO; it created output obs only if the MXG default &KEEPALL=YES was in effect, but that was an MXG oversight/error now corrected. Thanks to Willy Iven, BNP Paribas Fortis, BELGIUM. Change 27.326 Variable QDSTQCIT should not have been deaccumulated in VMACDB2 PDB.DB2STATS. Dec 5, 2009 Thanks to Terry L. Berman, DST Systems, USA. Change 27.325 z/OS 1.11 ONLY: INVALID TYPE 0 RECORD DETECTED, DELETED VMAC0 with LENGTH=60 is an MXG error; that new record length is Dec 3, 2009 correct, and should have been output to TYPE0/IPLS. MXG Change 27.229 decoded the new z/OS 1.11 field, but I had forgotten about this (VERY OLD) test for invalid length. This test exists because it is easy for a SYSPROG to accidentally create an ID=0 SMF record when trying to write a "user" SMF record, and there is a very high cost (or embarrassment) if MXG reports that there was an IPL when one had not happened! Thanks to Douglas C. Walter, Citigroup, USA.
EXCDCLOG product's user SMF record creates these new datasets: EXCDCSRC dddddd Dataset Description Contains Segments EXCDCSYS CDCLOG CDCLOGCA LOG CACHE DLR DCW EXCDCTGT CDCSRC CDCSOURC SOURCE SCT SDT DSL CDI CDO FORMATS CDCTGT CDCTARGT TARGET TCT DTC CDI CDO IMACCDC CDCSYS CDCSYSTM SYSTEM OSC CIT PAL MAA CMO TYPECDC CLS CVF DSC DCM DLP TYPSCDC DAL PAA CCI CCO VMACCDC
VMXGINIT Dec 3, 2009 Thanks to Jerry Urbaniak, Acxiom, USA.
VMXG70PR is eliminated. Dec 2, 2009 Thanks to Kenneth D. Jones, Bell Aliant, CANADA. Change 27.322 -New "STATS" value for IFCIDS= in READDB2 creates only the READDB2 PDB.DB2STATS Interval Statistics Dataset in PDBOUT=PDB, Dec 3, 2009 an alternative to the value "STATISTICS" that creates ALL DB2 Statistics Datasets in the PDBOUT. PDB.DB2STATS is is created from DB2STAT0/DB2STAT1/DB2STAT4 and summarized DB2STATB, all of which are left in //WORK with "STATS". The PDB.DB2STATS contains all of DB2STAT0/STAT1/DB2STAT4. -NOTE APR 2012: SEE CHANGE 30.077, IFCIDS=STATS CHANGED to create PDB.DB2STATS and PDB.DB2STATB. -The "ACCOUNT" value creates all DB2ACCTx datasets, but you can select a subset using the WANTONLY= argument. You can now just name the wanted DB2ACCTx dataset in your IFCIDS= argument, in place of ACCOUNT, and only those are created. Both IFCIDS= and WANTONLY= values can now be an MXG dataset name (e.g., DB2ACCTP, DB2ACCT), although the original unique-to-READDB2-token-names (DB2ACTP,DB2ACCT) that are listed in READDB2 comments are still valid. So if you only want the three common datasets created, %READDB2(IFCIDS=DB2ACCT DB2ACCTP STATS,PDBOUT=PDB) will create PDB.DB2ACCT, PDB.DB2ACCTP, and PDB.DB2STATS from SMF 100 and 101 records. -DB2 V8 only: The "DB2STAT4" is an SMF 102 IFCID=225 trace record; if you have those data, this syntax %READDB2(IFCIDS=DB2ACCT DB2ACCTP STATS 225,PDBOUT=PDB); creates T102S225 and use it also to create PDB.DB2STATS.
FORMATS Package that was executed and decoded by $MGDB2PK format MRGDB2 with new-in-DB2-V9.1 value of '04:NATIVE SQL' added. Dec 2, 2009 VALUE $MGDB2PK ' '='BLANK:NOT DEFINED' '01'='01:STORED PROCEDURE' '02'='02:USER DEFINED FUNCTION' '03'='03:TRIGGER EXECUTING' '04'='04:NATIVE SQL' -New MRGDB2 member will be renamed or become an example. -It merges variables from PDB.DB2ACCT with PDB.DB2ACCTP to create PDB.ENHACCTP for Enhanced Package Accounting. In this example, variable DB2PARTY is added, so ACCUMACC Package ROLLUP RECORDS (QWACRINV=1,2,3 DB2PARTY='R') obs can be identified; ROLLUPs don't have valid values in these Package (DB2ACCTP) variables, so this example sets all these variables in PDB.ENHACCTP to blank/missing for the ROLLUP Package records: IF DB2PARTY='R' THEN DO; /* DOCUMENTED INVALID IN PK67870: */ QPACLOCN=' '; QPACCOLN=' '; QPACPKID=' ';
QPACAANM=' '; QPACARNA=.; QPACASCH=' '; QPACCANM=.; QPACCAST=.; QPACCONT=' '; QPACEJST=.; QPACLOCN=' '; /*YES, IN BOTH*/ QPACSCB=.; QPACSCE=.; QPACSPNS=.; QPACSQLC=.; QPACUDNU=.; QPACUDST=.; END;
-The sort order of the new example PDB.ENHACCTP dataset is BY QWHSSSID QWHCAID QWHCOPID QWHCPLAN QWHSACE QWHSLUUV DESCENDING QWHSSTCK; Thanks to Scott Barry, SBBWorks, Inc, USA. Change 27.320 CICS User fields SRCE,SRCETYPE,SRCEPROG,MSGLEN are IMACAAAA supported in listed new IMACICxx CICS exit members. IMACICUF IMACICUF - SRCE IMACICUG IMACICUG - SRCETYPE IMACICUH IMACICUH - SRCEPROG IMACICUI IMACICUI - MSGLEN UTILEXCL VMAC110
Nov 26, 2009 Change 27.319 -DB2ACCT obs for Java Universal Type 4 JDBC Driver (JCC) FORMATS (QMDAPTYP='JCC') did not populate these QMDASQLI fields: VMACDB2 QMDAPLAT QMDAAPPL QMDAATID QMDASFLN ACCOUNTn Nov 25, 2009 -Cosmetic: Format $MGDB2PN added JCC for QMDAPTYP. 'ARI'='ARI:DB2 FOR VM,VSE' 'DSN'='DSN:DB2 FOR Z/OS' 'JCC'='JCC:UNIVERSAL JDBC DRIVER' 'QSQ'='QSQ:DB2 FOR I/SERIES' 'SQL'='SQL:DB2 FOR LINUX/UNIX/WIN' Thanks to Hugh Lapham, Royal Canadian Mounted Police, CANADA. Change 27.318 Cosmetic; no change if COMPRESS=YES default is unchanged. VMAC102 Several DB2 text-containing variables were $32000 LENGTH Nov 24, 2009 because old DSECTS had no clue of their maximum lengths, but some of these SMF 102 fields are shortened to their (newly documented, probably unchanged) stored lengths: QW0059CN $128 QW0061CN $128 QW0063ST $5000 QW0145TX $4000 QW0203PA $128 QW0206MS $256 QW0206MR $256 QW0206HR $64 QW0208MS $256 QW0208MR $256 QW0236MS $256 QW0236MR $256 These text fields are still $32000 stored length: QW0004MS QW0005MS QW0062ON QW0064CN QW0065CN QW0066CN QW0090CT QW0092P1 QW0097P1 QW01242T QW0140TX QW0141TX QW0142TX QW0168ST QW0180DS QW0194DS QW0204TH QW0350SP
FORMATS -Variables R7023CT and R7024CT, Crypto Processor Type, are VMAC7072 now decoded by $MGRMFCX format. Nov 20, 2009 -Dataset TYPE70Y2 new variables created by the APAR: R702CDLV='ICSF*DATA*LEVEL' R702AESC='AES*ENCIPHER*CALLS*SENT' R702AESB='AES*ENCIPHER*BYTES*PROCESSED' R702AESI='AES*ENCIPHER*OPERATIONS' R702AEDC='AES*DECIPHER*CALLS*SENT' R702AEDB='AES*DECIPHER*BYTES*PROCESSED' R702AEDI='AES*DECIPHER*OPERATIONS'
JCL40GIG streams, to reduce run times, or to run parts of MXG more Nov 19, 2009 than once per day (like multiple CICSTRANs), etc, and is the example for the DOC40GIG parallelization document. Thanks to Chuck Hopf, Independent Consultant, USA.
VMACTPMX from $HSM_MH and $SAF_US field names in the ThruPut Mgr's Nov 18, 2009 user SMF record. -Doc: ThruPut Manager delays: DCS - Dataset Contention Delays; when TPM detects that a job needs a dataset that is being held, it holds the new job. DBS - Device Busy Delays. Thanks to Betty Wong, BOA, USA.
VMAC82 GMT, but now SMF82XTD/XTN/XTW are shifted by GMTOFF82 to Nov 17, 2009 local time. -New variable SMF82ELP, Elapsed Duration, is now created in TYPE8219 and TYPE8220 datasets. -Label for SMF82SSI corrected. Thanks to Cesar Cocco, Citigroup, USA.
ANALINIT dataset only if it exists, eliminating NOT FOUND error. Nov 15, 2009 Thanks to Chuck Hopf, Independent Consultant, USA. Change 27.312 MXG 27.09 only. Change 27.254 was incorrect for NTSMF VMACNTSM processing on ASCII; that change was intended only for Nov 15, 2009 VBS format data. The RECFM=V is restored for ASCII. Thanks to Jim Quigley, ConEd, USA. Change 27.311 Support for CTG 8.0 (COMPATIBLE) fields added to SMF 111 VMAC111 record for TY111GD dataset: Nov 14, 2009 CTGIXACO='INTERVAL*XA*TRANS*COMPLETED*HA GROUP' CTGLXACO='LIFETIME*XA*TRANS*COMPLETED*HA GROUP' Change 27.310 New parameter DATEJUL= creates DSNAMEs with the date in VGETDDS Julian format, YYYYDDD, as an alternative to DATEBASE= Nov 17, 2009 which creates the DATE in the DSNAME in DDMONYY format. The START and END values can be specified with MDY values or JULIAN values, or even a numeric string. Thanks to Brian Harvey, HCL America, USA. Change 27.309 -z/VM dataset VXSYTEPM variable's values were carried from FORMATS from prior CMG segments (so ESCON variables could be VMACVMXA populated in FICON observations). Now, the "other" CMG's Nov 11, 2009 variables are set to missing values. -New variable CHPIDTYP identifies the type of channel and is formatted by new $MGVXACH format, which identifies the type of OSA adapter, and it should be used in your sorts. -New variables PCTCPCBY and PCTLPABY are calculated from Work Units (like TYPE73) for CMG=2 (FICON) records. Thanks to Melanie Hitchings, BT, ENGLAND. Change 27.308 MXG 27.08-27.09. SYNC59=NO again default VMXGDUR,VMXGSUM. VMXGDUR
VMXGSUM Caused minor differences in hourly counts in PDB.CICS Nov 10, 2009 when minute 59 count/resources were shifted to next hour. ASUMCICX Change 27.214 changed the internal default to SYNC59=YES, ASUMCICS in these two internal members that summarize everything, ASUMCICS its text claiming "ALWAYS SAFE"! But that is ONLY true if ASUMCICT the input data being summarized is "interval" data, i.e., ASUMCIMS it already contains the Interval Start/End times, like ASUMDB2A RMF,SMFINTRV,DB2STATS, or if SYNC59=YES was specified in ASUMDB2B the ASUM/TRND/etc that invoked %VMXGSUM. ASUMDB2G
ASUMDB2P Instead, when the input data being summarized is "events" ASUMDBDS or "transactions", like CICSTRAN,DB2ACCT,ASUMUOW, wherein ASUMDBSB VMXGSUM creates the Interval Start/End times, my change ASUMHSM from NO to YES caused differences in hourly counts in the ASUMIDMS PDB.CICS dataset (created from CICSTRAN by ASUMCICX) as ASUMJOBS those minute 59 data were being counted in the next hour. ASUMSTC (There's nothing wrong in using SYNC59=YES in ASUMCICX ASUMVTVM to shift those minute 59 transactions, if that's what you know you want because you have SYNC(59) data, but I should not have changed that global default!). While not required with the restored default value, the listed ASUMxxxx members use input event data and create interval start/end times, so they now all contain the SYNC59=NO option, mostly for documentation. Thanks to Paul Naddeo, FISERV, USA.
VMACDB2 length for the "truncated" QMDALOCN field, so MXG read in Nov 6, 2009 that longer text into QMDALOCN, but examination of those longer text values show they are not valid, and since the DB2 V7.1 DSNDQMDA doesn't show the "truncated" offsets, which appear to have been added in DB2 V8.1, the input of the truncated field is now bypassed for QWHSRELN LE 7.1.
VMACORAC ERROR: ILLEGAL VALUE FOR STARTS messages. This change Nov 6, 2009 recovers all original values, and creates seven ORACUNNn numeric unknown variables and two ORACUNCn character variables observed in the SMF records. These variables will be named when a DSECT is available. Change 27.305 The MSU and MIPS calculations in ASUMMIPS for workloads ASUMMIPS (i.e., the Service Class MSU from TYPE72GO and Address Nov 6, 2009 Space MSU from SMFINTRV) did NOT take into account the Capture Ratio. of each of the engine types; only the Captured CPUTM was used. This change keeps the RMFINTRV Capture Ratio variables (CAPTURAT CAPIFART CAPZIPRT) for each Engine type, and the MSUUSED is divided by (CAPxxxxx/100) to take into account the Interval Capture Ratio for that SYSTEM for that engine type. This will cause the xxxUSED values to be estimates of the hardware utilization, and they will be slightly larger now. If intervals are summarized, the Interval Capture Ratio is the weighted average of the input data. The use of Capture Ratio can be turned off by specifying a 0 instead of 1 for the _USECAPT macro. Thanks to Brian Harvey, HCL America, USA. Change 27.304 The IFA/zAAP and zIIP Capture Ratios are wrong if those VMXGRMFI Specialty Engines are faster than the CP engines, because Nov 6, 2009 the IFATM/ZIPTM are normalized values. Now, the factors Dec 3, 2009 R723NFFI and R723NFFS, respectively, are used in the MXG calculation of CAPIFART and CAPZIPRT Capture Ratios. The CPU times are NOT changed back to raw values. Change 27.303 CICS Total I/O Wait WTTOIOTM & Total Other Wait WTOTIOTM ADOC110 variables now include all 45 WAIT durations included in VMAC110 Suspend Time that are in IBM's CICS Performance Guide for Nov 5, 2009 CICS/TS 4.1; these six wait durations are not separately Nov 10, 2009 in WTOTIOTM because they are included in RMISIOTM: IMSWAITM DB2RDYTM DB2CONTM DB2WAITM DSCHMDTM WMQGETTM -Member ADOC110 is updated with the complete list of which waits are I/O and which are Other waits, and a schematic of how these waits are related. The new equations are: WTTOIOTM=SUM( /*TOTAL I/O WAIT DURATION*/ WTTCIOTM,WTTSIOTM,WTSHIOTM,WTTDIOTM,WTJCIOTM, WTFCIOTM,WTRLIOTM,CFDTWATM,SOIOWTTM,ISIOWTTM, SOOIOWTM,WTIRIOTM,LU61IOTM,LU62IOTM,SZWAIOTM); WTOTIOTM=SUM( /*TOTAL OTHER WAIT DURATION*/ DSPDIOTM,ENQDIOTM,GNQDELTM,WTICIOTM,WTLMIOTM, WTWEIOTM,WTWCIOTM,RUNTRWTM,SRVSYWTM,RQRWAITM, RQPWAITM,SYNCDLTM,MAXOTDTM,MAXJTDTM,MAXSTDTM, MAXXTDTM,RRMSWATM,PTPWAITM,RMISIOTM,JVMSUSTM, DSTCBMTM,DSMMSCTM,WTDWIOTM); WTUNIOTM=MAX( /*UNCAPTURED WAIT DURATIOM(/ 0,(SUSPNDTM-(WTTOIOTM+WTOTIOTM))); Thanks to Tom Kelman, Commerce Bank of Kansas City, USA. Thanks to Dick Arnold, Commerce Bank of Kansas City, USA.
VMACTCP variable FTPUSRMT is now correctly labeled "REMOTE" and Nov 4, 2009 not "LOCAL". Thanks to Marybeth Delphia, Texas Comptroller of Public Accounts, USA Change 27.301 Unused Change Number. Change 27.300 ASMTAPEE ML-45 provides several enhancements. ASMTAPEE -Addition of assembly time stamp to initialization message Nov 1, 2009 TMNT016I. This will help to confirm that the version of TMNT that is active matches the expected assembly. TMNT016I MXG Tape Monitor maintenance level 45 initialization complete (2009/10/31-13.15) -Modification of the TMNT018I initialization message to include the SMF record number being used: TMNT018I MXG Tape Monitor interval set to 0.50 seconds using SMF record number 238 -Added message TMNT011E (in addition to existing TMNT010E) to more clearly identify, in human terms, the reason why SMF recording has failed: TMNT010E SMF write failed - SMFEWTM return code is 00000028 TMNT011E SMF write failed because a buffer shortage caused the data to be lost -Added support for BAM (basic access method) use of XTIOTs (TIOTs resident above the 16mb line). This is only relevant when the XMEM=YES option is specified. -Each ML has small performance enhancements to decrease MXGTMNT overhead; generally this involves taking advantage of macro options that improve performance and/or eliminating unnecessary instructions from performance sensitive code paths. The benefits are small from ML to ML but the difference between ML45 and ML26, for instance, is fairly significant.
ALOCGOVO by only opening the GDG Base Name with relative reference Oct 31, 2009 to get the full "goovoo" dsname, freeing the Base Name, and then allocating the FILENAME SMF with the "goovoo" name for the actual SMF processing, so the duration of the DSNENQUE for the GDB Base Name is very short. Normally, if you use JCL to allocate SMF(0), your job will prevent an SMF Dump from running to create (+1), until your job ends (or, at least until SMF reading is completed, if you use FREE=CLOSE on your //SMF DD). The WAIT=20 option is used to open the relative reference FILENAME SMF "YOUR.SMF(0)" DISP=SHR WAIT=20; so that if the GDG Base is already in use, SAS will enter a Detected Wait state for up to 20 minutes, waking every 15 seconds to see if the enqueue has cleared; without the filename option WAIT=20, SAS would have terminated with ERROR: FILE IS IN USE, YOUR.SMF(0) message). The examples show DDNAME=SMF, but that and the options on the generated FILENAME statement can be changed. -This ALOCGOVO only works if your REMOVE the static DD statements from the JCL; otherwise they will still control the allocation. Thanks to Christian Hodel, Swisscom IT Services, SWITZERLAND. Thanks to MP Welch, SPRINT, USA. Change 27.298 Another Elegant algorithm uncompresses ASG/Landmark data VMACTIMS records in SAS code, so it works on ASCII systems (which VMACTMDB do not support INFILE exits). However, it is EXPENSIVE, VMACTMDC requiring SIX TIMES THE CPU THAT THE EXITMON6 INFILE EXIT VMACTMMQ REQUIRES: USE EXITMON6 ON z/OS. See Change 27.260. VMACTMO2 ASG's compression technique is different than IBM's, but VMACTMTC a similar increase in CPU Time was observed. However, VMACTMVS since ASCII systems do not support INFILE exits, this SAS VMACTMVT code algorithm does support processing of compressed data Nov 5, 2009 on ASCII or with WPS, which might be worth the CPU cost. Thanks to Ian Gibson, CPT Global Ltd @ Bendigo & Adelaid Bank, OZ. Thanks to Peter Turner, CPT Global Ltd @ Bendigo & Adelaid Bank, OZ. Change 27.297 -ANALDB2R - MERGE STATEMENT HAS REPEATED BY VALUES, and ANALDB2R SUBSTR errors in AUDIT reports were corrected. READDB2 -ANALDB2R corrected logic to populate the IFCIDs being Nov 4, 2009 requested with READDB2 is used with PDB=SMF. Yüklə 28,67 Mb. Dostları ilə paylaş:
|