* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
be written at one minute intervals, MXG now detects a gap of more than 120 second as the start of a new interval, and it is DIF()'d but not output. However, neither BEGTIME nor ENDTIME are on the minute DURATM values of a few seconds up to 100 seconds have been observed, hence the test value of 120 seconds. -DB2STATS Variables QISEDPSL QISEDPSC QISEDPSM QISEDPSF were incorrectly input with DB2 V11; the test GE 32 is corrected to GE 232 for their input. Thanks to Giuseppe Giacomodonato, EPVTECH, ITALY. Change 35.228 Support for up to 20 user character fields in CICSTRAN, IMACICAU USERCHAR1-USERCHAR20, increased from only nine fields. IMACICBU IMACICCU
IMACICDU IMACICEU
IMACICFU IMACICGU
IMACICHU IMACICIU
IMACICJU IMACICKU
IMACICLU UTILEXCL
VMAC110 Oct 11, 2017 Change 35.227 If you specified RMFINTRV=NO many bad things could have UTILBLDP happened. First, if you did not also suppress one or Oct 17, 2017 more of the records used by BUILDPDB, then RMFINTRV=NO was ignored because the code looking for it was inside of the SUPPRESS= logic. If you DID suppress something then you could have hit a problem with a missing % sign on a MACRO. -User ABENDS are replaced by messages on the OUTFILE to tell you there was an error that terminated UTILBLDP. -if you suppress type 26 without specifying J2 or J3 it will now generate an error -if you useradd 102 without specifying the subtype it will now generate an error Thanks to Trevor Holland, ANZ, AUSTRALIA. Change 35.226 Unused Change Number. Change 35.225 WARNING: VARIABLE YSTEM NEVER REFERENCED because SYSTEM VMACTMO2 was missing the first S. Oct 10, 2017 Thanks to Derek Purves, FDIC, USA. Change 35.225 -New UTILBLDP Parameters EXPDBVAR/EXPDBCDE/EXPDBOUT added UTILBLDP to enable more tailoring and specifically to make it easy Oct 8, 2017 to create a new dataset with a subset of variables and obs from an existing MXG dataset, like this example to create new CICSTRIM dataset with a subset of CICSTRAN variables, and output only for selected CICS APPLIDs: USE EXPDBVAR TO BUILD A SUBSET OF CICSTRAN DATA FOR REPORTING PURPOSES %UTILBLDP(OUTFILE=INSTREAM, BUILDPDB=YES, OUTFILE=INSTREAM, EXPDBVAR= CICSTRIM (KEEP=APPLID TRANNAME STRTTIME ENDTIME CPUTM ELAPSTM TASKNR USER ABCODE), EXPDBOUT= PROC SORT DATA=CICSTRIM OUT=PDB.CICSTRIM; BY ENDTIME APPLID TRANNAME TASKNR;, MACKEEPX= MACRO _ECICTRN IF (RTYPE= 'E3'X OR RTYPE = 'T') AND APPLID IN('CICSZFN3','CICSAUD3','CICSDBS') THEN OUTPUT CICSTRIM; OUTTPUT _WCICTRN; % );
VMAC110 sets for tailoring had new datasets added which caused Oct 8, 2017 only one dataset listed per line; that statement now has two datasets per line, half as many lines/bytes. Change 35.223 The extra zVPS/XAM SYTCUP segment with totals was not VMACXAM included in the SYTNLPS count of segments, and there is Oct 7, 2017 no LENDATA value for each subsegment, so MXG assumed 20 for LENDATA and detected the extra subsegment when SEGLEN was NE 20*SYTNLPS, but the 35.09 correction statement SYTNLPS=SYTNLPS+1 was mistyped as SYTNLPS=SYTNLPS=1; so only one obs per LPAR was output in XAMSYT dataset. But other MXG corrections have increased the obs count, depending on past MXG Version (eg. 35.06 to 35.09). There will be one obs for each LCPUADDR in each LPARNAME, and an extra "total" obs with LCPUADDR='60'x (doc '40'x) for each Engine Type in each LPARNAME, but these "total" obs all have zero values in 4303 and 4313 releases. There are also a pair of original LPARNAME='Totals' subsegments at the start of each SYTCUP segment that are not output by MXG. Thanks to Paul Volpi, UHC, USA. Thanks to David A. Sadler, UHC, USA. Change 35.222 Unused Change Number Change 35.221 Many zVM VXPRCPUP dataset variables values were not VMACVMXA divided by 65536, two variables needed deaccumulation. Oct 4, 2017 The segment SSIZE is 96 but only 72 are documented. Thanks to Pat Hansen, ADP, USA. Thanks to Mike Chaves, ADP, USA.
VMAC119 a debugging asterisk left where it shouldn't have been in Oct 4, 2017 line 2942 of VMAC119. Thanks to Paul Volpi, UHC, USA. Change 35.219 MQMACCT variable NETSNAME is created from QWHCTOKN if it VMAC116 is populated, or from QWHCNID in not, but the format of Oct 4, 2017 the raw data is different; heuristics were revised to recognize two formats found in this site's data, with values AAAAAAAA.BBBBBBBB if QWHCTOKN is populated or values CCCCCCCC.L when QWHCTOKN is not populated. NETSNAME is also kept in MQMACCT1 and MQMQUEUE but it should not have been, as it is always blank for those two datasets. And NETSNAME is not populated in MQMACCT observations from BATCH/TSO Attach. Thanks to Jim Poletti, Edward D Jones, USA. Thanks to Art Morelock, Edward D Jones, USA. Change 35.218 XAMSYPUP dataset's INPUT is now correctly aligned once it VMACXAM was explained that PL/1 "3 rsrvd(4) Char(4)" is SAS +16. Oct 3, 2017
ANALID unbalanced quotes that impacted SMF ID=80 descriptions in FORMATS the ANALID SMF Report (default is on in BUILDPDB). Oct 2, 2017 Also option UNIFORM was added to the SUMMARY PROC PRINTs. Thanks to Scott Barry, SBBWorks Inc., USA.
JOINDB2 to the corresponding DB2ACCTP observations. Written Oct 2, 2017 specifically to capture QWACWLME since the DB2ACCTP records do not contain the QWAC segments but will copy as many variables as you like from DB2ACCT to DB2ACCTP. NOTE: QWACWLME only exists in DDF records and that is by IBM's design.
VMXGRMFI INTTRND macro variable as documented. Also the TRNDRMFI TRNDRMFI member was updated to include the TRENDINT parameter so Oct 2, 2017 that you can more easily see how to modify the interval. Thanks to Steve Carlson, UCOP, USA.
UTILROLL it now uses PROC APPEND instead of a data step which had Sep 30, 2017 to read both the input (ROLLTO) and output (ROLLFROM) datasets. Change 35.213 z/VM variable VMDUFACT in dataset PDB.VXBYUSR should not VMACVMXA have been deaccumulated as it is an end of interval count Sep 28, 2017 of frames. Thanks to Graham Harris, RBS, ENGLAND. Change 35.212 Support for SMF 30 User Key CSA Audit Enhancements adds VMAC30 new SMF30_RAXFLAGS to TYPE30_1, TYPE30_V, TYPE30_4 and Sep 28, 2017 the TYPE30_5 datasets. This code change has been in MXG Feb 28, 2018 35.09 and later, but this change text replaced previous "Reserved Change" on Feb 28, 2018. This field was added by APAR OA53355, but will only be needed thru z/OS 2.3, as User Key Common Storage usage support ends there. This is Health Check ZOSMIGV22R3_NEXT_VSM_USERKEYCOMM. These APARs required no additional code changes: OA53434 Corrects ASM DSECT Lengths, no MXG impact OA53289 Corrects value of SMF30HVR from zero to valid. OA45767 APAR that added the extra triplet caused OA53434 Change 35.211 Documentation. Variables QWACWLME QWACBSC QWACESC should VMACDB2 not have been kept in DB2ACCTP after IBM moved the DB2 Sep 27, 2017 Package Data (IFCID 129) to its own ID=101 Subtype=1 record, which does not contain a QWAC segment. Thanks to Glen Bowman, Wakefern, USA.
ASUM113 now has 128 entries). Default LABELs are now for z14. VMAC113 To change the default labels for z13, you would use: Sep 27, 2017 //SYSIN DD * Oct 1, 2017 %LET MACKEEP= MACRO _XLA113 _XLA113D % ; %INCLUDE SOURCLIB(TYPS113,ASUM113); To create correct labels with both z13 and z14 data, you must create separate datasets: //SYSIN DD * %INCLUDE SOURCLIB(VMACF113); DATA PDB.ASUM1131Z13; LABEL _XLA113D ; SET PDB.ASUM1131; IF SM113VN2=4; DATA PDB.ASUM1131Z14; LABEL _XLA113E ; SET PDB.ASUM1131; IF SM113VN2=5; RUN; Many of the RNI and other equations were changed for z14. -You can use OPTIONS OBS=99; _RPT113; RUN; to print those calculated variables values from PDB.ASUM1131, the data set you should use, as it contains interval data. The original ASUM113 data set was accumulated and obs were lost in deaccumulation. Thanks to Elie Sawaya, RBC, CANADA. Change 35.209 -Support for BETA 93 Version 610, altered header fields FORMATS and three new variables are added to dataset BETA0: VMACBETA BETAPABS='MAX*PABS' Sep 29, 2017 BETASPAG='FIRST SPLIT*OVER GT*1 PAB' Nov 22, 2017 BETABMOD='PAB PAGE*BREAK*MODE' -New values added to $MGBETAT format for BETA0. -Support for subtype 51 RDATA1 and RDATA2 fields and all segments are read and output - previously the first segment was repetitively output. See Change 36.074 for Subtype 51 update. Thanks to Andreas Menne, Finanz Infoirmatik, GERMANY. Change 35.208 Nigel's Monitor for AIX and LINUX changed BBBP Endtime VMACNMON was always "HH:MM" text, but new "N MINS" tripped MXG Sep 27, 2017 variables BBBPENDING051, UPHOURS, and BBBPENDING052, UPTIME.
Thanks to Steve McCulloch, TMS/CDS Group, CANADA. Change 35.207 Enhancement for GDG datasets to add FIRSTGEN and LASTGEN VMAC6156 values of GATGEN to know the range of Generation values. Sep 25, 2017 No change, but variables GATEXTNO GATVER GATGEN GATWRAP are from each GAT segment, so they should not have been kept as they contain only the value of the last cell. Thanks to Satish Kodavatiganti, John Deere, USA. Change 35.206 If you need ACCOUNTn fields for long-running STCs in your BUILD005 PDB.SMFINTRV dataset, you had to set SPINCNT in IMACSPIN BUIL3005 longer than the number of days between IPLs; if you had a VMXGINIT smaller value, then only that number of days of the Sep 25, 2017 PDB.SMFINTRV would have the ACCOUNTn fields populated. This change creates optional macro variable &SPINSTC to enable this change and to keep that many day's SPIN30_1 for STCs in the SPIN library for STC accounting, and you can then set a much smaller value for the spining of you other jobs. Use %LET SPINSTC=365; to keep a year's data. A large SPINCNT used to be important when you had lots of held print output, since BUILDPDB waits for the PURGE (26) record to know all SMF records for the job have been created. But now, with output typically sent to a spool handler, almost all jobs PURGE right after they terminate on z/OS, so now, you can set a small SPINCNT (1 or 2) and only SPIN the jobs that were in execution when SMF was dumped, or that run for more than a full day or two. Thanks to Gennady Katsnelson, IBM Global Technology Services, USA. Change 35.205 Documentation of what is counted in SMF 30 EXCP counts, TYPE30 from a posting to IBM-Main: Sep 25, 2017 The EXCP count fields count whatever the IOS Driver decided to pass into SMF in the IFASMFEX exit that accumulates the type 30 EXCP fields. For example: -For SAM, EXCP (an IOS driver) lets SAM do the calls to IEASMFEX. -For non-SAM use of EXCP, EXCP calls IEASMFEX with a count of 1 so for non-SAM use of EXCP, it is the count of EXCPs -And IEWFETCH (an IOS driver which fetches load modules) counts the number of SSCH for non-VIO data sets, and uses EXCP for VIO data sets. So either way, it is the number of channel programs executed. Thanks to Jim Mulder z/OS Diagnosis,Design,Development,Test IBM Corp Change 35.204 Support for DB2 APARs PI71903 and PI84045 adds these VMAC102 new variables to IFCID 376 T102S376 dataset: Sep 28, 2017 QW0376SC='SCHEMA*NAME' Oct 11, 2017 QW0376PR='SPECIFIC*NAME' Dec 7, 2017 QW0376INC='INCOMPAT*PARMS' QW0376SQL='SQL*TEXT'; Dec 7: Offsets for VL/VN corrected, no data for these new fields yet. Thanks to Lori A. Masulis, FMR, USA Thanks to Steve McKee, FMR, USA. Change 35.203 z/VM 6.4.17.1 INCOMPATIBLE, fields inserted in SYTCUP and EXSYTLCX SYTLCK, and new VXSYTLCX data set created. VMACVMXA -In SYTLCK when there are no shared/exclusive lock entries VMXGINIT (CALNMSXE=0) there are 8 bytes inserted where that second Sep 28, 2017 array should not have existed. -In SYTCUP, SKIP logic was not correct with new data. Thanks to Dr. Wolfgang Kueller, IT Solutions, AUSTRIA
VMXGALOC since that section of code was only used to CLEAR any Sep 18, 2017 exising LIBNAMEs before allocating new ones (which in any case would happen when a new LIBNAME statement was issued). Thanks to Steve Bagshaw, Marks & Spencer, ENGLAND. Change 35.201 Modified to limit the number of LIBNAMEs reported where PDBAUDIT the path name is the same. First looks for a PATH where Sep 18, 2017 the LIBNAME is PDB and deletes any other LIBNAMEs with that path then sorts on PATH and LIBNAME eliminating all but the first occurrence of each PATH so that there are not a lot of duplicate lists. Needed for MXG QA. -A new parameter PATHLIST= added with a default value of NO. Change to YES to create report of the LIBNAMES by PATH. Useful if you are running with AUTOALOC=YES.
VMXGALOC monthly and trend "PDB's" to different paths (drives or VGETALOC directories). All default to the BASEDIR if left blank. BLDSMPDB BASEWEEK= sets the location of weekly database Sep 13, 2017 BASEMONTH= sets the location of monthly database BASETREND= sets the location of TREND database If you choose to use these new destinations for your output of BLDSMPDB, be aware and use caution since the old destination's files will NOT be aged off, nor will they be allocated for monthly/weekly processing. You will need to copy old data from the old path to the new path.
VMAC90A to incorrect offset and field length in the GA SMF manual Sep 17, 2017 for SMF90T38_UTOKENUSERID which is 16 not 8 and at offset 112 and not 118. Subtype 38/39 datetimes are now local. Thanks to Bernie Ethridge, Fiserv, USA. Thanks to Paul Naddeo, Fiserv, USA. Change 35.198 z/VM 6.2.11 SYTLCK "BROKEN RECORD" error because SKIP was VMACVMXA not calculated correctly. Sep 11, 2017 Thanks to Kare Martin Torsvik, EVRY, NORWAY. Change 35.197 IMF CIMSTRAN dataset datetime variables all now have VMACCIMS microsecond resolution; MXG had overlooked the MIJUs. Sep 16, 2017 ACTARRV ARRVTIME STRTTIME TRNETIME ENDTIME TRNSTCKE Oct 26, 2017 INPQUETM, SERVICTM and RESPNSTM are now calculated from those datetimes for microsecond resolution. Thanks to Randy Hewitt, DXC Technology, USA. Change 35.196 Support for BETA 97 extended header (INCOMPATIBLE) V 610. EXTYB97D All variables in all datasets are now INPUT and correct. IMACBE97 New dataset BE979751D is created from Subtype 51 with the VMACBE97 database field details. VMXGINIT Sep 15, 2017 Thanks to Andreas Menne, Finanz Infoirmatik, GERMANY.
EXXAMPUP new dataset DDDDDD DATASET DESCRIPTION IMACXAM XAMPUP XAMSYPUP PRCPUP DATA VMACXAM -The test for invalid SYTCUP segment was revised when the VMXGINIT old test incorrectly reported an invalid segment. Sep 15, 2017 Change 35.194 Unrequested log messages containing MXGDEBUG: VMXGOPTR ANAL116 were printed if you %LET MXGDEBUG= to a non-blank value. ANALDB2R The LENGTH(&MXGDEBUG) test was removed from VPUTMSG and ASUMDB2A relocated to each calling member, with revised logic: ASUMDB2R %IF %UPCASE(&MXGDEBUG) NE VMXGSUM1 %THEN %DO; ASUMNTIN %VMXGOPTR(OPTNAME=NOTES,NEWVALUE=NONOTES); TESTTRND %END; TRNDNTIN The MXGDEBUG macro variable is primarily for internal MXG VGETALOC testing and it exists in only a few members; the enable VMACDB2 values are documented in each member's test statements. VMXG70PR With this correction. MXGDEBUG=VMXGSUM1 was used for MXG VMXGALOC QA which exposed these overlooked corrections: VMXGDSN -ASUMDB2A QXHJINT typo was observed and removed. VMXGOPTR -VMXG70PR LPMSUHR was missing. VMXGRMFI -ANAL116 had a superfluous ID=ENDDT argument, removed. VMXGSUM -VGETALOC/VMXGALOC protected for blank MXGDEBUG. VPUTMSG -VMXGSUM many calls to VPUTMSG revised so VMXGSUM1 now Sep 10, 2017 is also enabled if 2/3/4 are requested. -VMACDB2 did not keep SHIFT in DB2ACCTR dataset. -VMXGRMFI had spurious SMF70GMN SMF70GMU MXGDEBUGs. -ASUMNTIN did not keep one variable -TESTTRND builds PDB.CICS from ASUMCICX, clearing the first PDB.CICS created by ASUMCICS which had different kept variables and is not the recommended CICS summary. -ASUMDB2R needs KEEPALL=YES for missing variable notes. -TRNDNTIN needed a variable added to KEEP list. Thanks to Donald Blaszka, WiPro, USA.
VMAC74 variables, and new SMF74SQRRATE/SMF74SQWRATE sync rates. Aug 30, 2017
VMAC116 are missing values in obs created from SMF 116 Subtype 2 Aug 29, 2017 (continuation) records, but now INTS/STRT are populated by WQTTTIME and INTE is populated by SMFTIME. Thanks to Raymond J. Smith, OPTUM, USA.
VMACRMFV validated so the bypass execution tests (-99 EQ 99) are Sep 5, 2017 are now removed. -Variable GEIFLG2 is now INPUT and kept in ZRBGEI dataset. -Unpopulated Extended Length EDE segment overlaid original INPUT of variables EDEPCKG EDEPROC EDEUSER EDETRXN EDEACC with blank values. -Overlooked OPD variables now INPUT and kept: OPDDCTIIP='DELTA*TCB TIME*FOR*ZIIP' OPDCTIIP='PROCESS*SYSTEM*USER COMPUTE*ON ZIIP' Thanks to Kurt Gramling, T-SYS, USA.
VMAC0203 EXCEEDED RECORD LENGTH due to MXG INPUT mis-alignment. VMACSMF -Dataset TYPE0202 now keeps those signature variables, and Aug 28, 2017 dataset TYPE0203 reverts to the way it originally was, keeping only the header variables. -BUT: the type 2 subtype 1/2 records have SYSTEM='DUMY' in the SMF header! Since that is not a real system name, and because it could confuse any SMF Audit or analysis, MXG's handling of the _SMF header in VMACSMF now detects type=2 subtype=1/2 and SYSTEM='DUMY' and changes SYSTEM to the actual SYSTEM (SMF2IRSID/SMF2GRSID) name. Thanks to Daniel Erikols, Svenska Handelsbanken, SWEDEN. Change 35.189 Variables QPAC_PIPE_WAIT and QPAC_PIPE_COUNT are INPUT VMACDB2 and kept in dataset DB2ACCTP. Aug 25, 2017 Thanks to Giuseppe Giacomodonato, EPVTECH, ITALY. Change 35.188 If default was set to "work" rather than "WORK" the case VMXGDEL mismatch caused datasets to be deleted, Aug 25, 2017
VMAC123A the KEEP= list. Aug 23, 2017 Thanks to Patricia Hansen, ADP, USA. ====== Changes thru 35.186 are in this MXG 35.08 dated Aug 24, 2017===== Change 35.186 New variables MKFLAGA/MKRLSOPT/MKLRTIME are created and VMACEDGS kept in EDGSKREC dataset; variable MKSTORE2 is no longer Aug 23, 2017 valid, as it now contains the time part of MKLRTIME. Aug 25, 2017 MKLRTIME='LAST*REFERENCE*DATETIME' MKFLAGA ='FLAG-A' MKRLSOPT='RELEASE OPTIONS' -Aug 25: MKLRTIME was missing value because 2-byte field before the new date value was not documented, but the hex record and doc offset showed 2 bytes were inserted. Thanks to Marybeth Delphia, CPA Texas, USA. Change 35.185 Change 35.167 forced you to have a PDB libname when you BLDSMPDB may not have needed one. There are 4 executions of Aug 22, 2017 VGETSORT within BLDSMPDB with differing needs. If running with RUNWEEK=YES the LIBNAME pointed to by WEEKSTRT is used If running with RUNWEEK=WTD the LIBNAME pointed to by PDB is used If running with RUNMNTH=YES the LIBNAME pointed to by WEEK1 is used If running with RUNMNTH=MTD the LIBNAME pointed to by PDB is used
PDBAUDIT caused termination of the QA test job with 35 libraries, Aug 23, 2017 and PDB.PDBAUDIT and SPIN.SPINAUDIT datasets not created. The original error it was supposed to prevent was found to be unrelated the LIBNAME count.
EX102389 T102S477 have the IFCID-specific variables; the three EX102404 others identify the event, but keep only the thirty-six EX102413 variables from the DB2 Header and Product segments. EX102414 DDDDDD DATASET DESCRIPTION EX102477 102389 T102S389 ALL INDEXES WITH FTPS IMAC102 102404 T102S404 AUTHORIZATION COMPATIBILITY VMAC102 102413 T102S413 BEGIN WAIT FOR PIPE SUSPEND VMXGINIT 102414 T102S414 END WAIT FOR PIPE SUSPEND Aug 22, 2017 102477 T102S477 ALOC/DEALOC FAST TRAVERSE BLOCK Thanks to Harald Seifert, HUK-COBURG, GERMANY.
VMAC74 tested SMF748LL instead of SMF748CL for the INPUT of the Aug 17, 2017 new field, which "worked" when there were link segments, but this record had only the control segment and non-zero R748CRTN Return Code. Thanks to Harald Seifert, HUK-COBURG, GERMANY. Change 35.181 Support for four new SYTSTSCP variables added: VMACXAM
Aug 16, 2017 Change 35.180 MXG 35.07. SMF 92 Subtype 50 INPUT STATEMENT EXCEEDED due VMAC92 to MXG typo of 44 vs 4 in the INPUT, but also the 16-byte Yüklə 28,67 Mb. Dostları ilə paylaş:
|