* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
WARNING: Dimension crossing has multiple format modifiers. message that was eliminated by the addition of a separate FORMAT statement for ID and SUBTYPE variables, thanks to SAS Technical Support. Thanks to Alf-Terje Thomassen, Ergo Group, NORWAY. Change 26.257 Corrections/enhancements for Omegamon CICS user SMF data. VMAC112 Previously, three subtypes, 200 (INTR), 201 (SYSR) and VMACOMCI 203 (ONDV) were written as user SMF records, supported Nov 16, 2008 by MXG TYPEOMCI code, in Versions V550 and earlier, but now, in Version V560, only subtype 203 is created, and it is written as SMF ID=112 (now that Candle was absorbed into IBM, they can use "IBM" SMF record numbers!). MXG has supported the SMF 112 for some time, but in this revisions, errors in the DataCom and MQ count/clocks were found and corrected, so this change is now required. -Only TYPEOMCI supports all three (200,201,203) subtypes, but only for Version V550 and earlier. -TYPE112 supports only subtype 203, because V560 only creates that subtype, and only as SMF ID=112 records. However, with this change, the TYPE112 code reads both the new V560 and old V550 format subtype 203 records. This change in VMAC112 tests ID=112 OR ID=_IDOMCI, so if you have defined MACRO _IDOMCI nnn % in IMACKEEP for your current OMCI User SMF record number, those records and new SMF 112s will be read automatically. -Or, you can set _IDOMCI "instream" using: //SYSIN DD * %LET MACKEEP= MACRO _IDMOCI nnn % ; %INCLUDE SOURCLIB(TYPE112); -The dataset names created for TYPE112 Subtype 203 are different in TYPE112, starting with T112xxxx, instead of the old OMCIxxxx dataset names, but all old variable names are still created in the new-named datasets. This text replaces Change 26.160, as it is/was incorrect. -The Datacom counts/clocks were wrong in TYPEOMCI support but were correct in TYPE112; four pairs of variables, DTCNT/DTCLK/DTFCNT/DTFCLK 02,07,09,14 did not exist in the records, but the old OMCI code input them, causing other variables to have incorrect values. -The 24 MQ counts did not exist; there are 8 pairs of MQ clocks and counters now correctly input, and the MQ filename was truncated. Thanks to Richard Schwartz, State Street Bank, USA. Change 26.256 MXG 26.09 only. %ANALDB2R(PDB=SMF); i.e., reading SMF to ANALDB2R create DB2 reports, caused many errors, starting with TESTANAL ERROR: OLD-STYLE MACRO NAME % MUST CONTAIN ONLY LETTERS, Nov 13, 2008 DIGITS, AND UNDERSCORES. The error has existed for some time, caused by recursive, mis-located include of VMACDB2 when SMF is read, and it can be circumvented by using %READDB2 to read SMF, and then use %ANALDBR only to create reports. The default reports with %ANALDB2R(PDB=SMF); can be created using %READDB2(PDB=SMF,PDBOUT=WORK,IFCIDS=ACCOUNT 106); %ANALDB2R(PDB=WORK); RUN;
until you install this corrected ANALDB2R. The MXG QA stream tested only %ANALDB2R(PDB=PDB,...), but now a new test of %ANALDB2R(PDB=SMF,...) is added. Thanks to Dan Almagro, Automobile Club of South California, USA. Change 26.255 -ASMVTOC updated to prevent ABEND when it scanned a device ASMVTOC that was pending offline. VMACCTLG -VMACCTLG syntax error for missing "B" in two bit tests VMXGINIT for IF GDGATTR='1.......'. EXCTLGE4 -User coded enhancements to Catalog Processing: Nov 11, 2008 (a) Write observation to CTLGDSN for each Generation Dec 23, 2008 Dataset (GDS) so this dataset includes an observation for each non-vsam dataset. (b) Write observation to CTLGVSAM for each AIX dataset. Since AIX datasets exist in a Cluster Sphere, the cluster gets written to the CTLGVSAM with the last AIX data and index component names. (c) When writing out the observation for CTLGVSAM, write out the volumes used for the index separate from the volumes used for the data component. (d) Added support for Connector Records (a user catalog connected to the master catalog). A Connector Record is composed of a Name Cell c'U', an Owner Cell x'01', 0 or more Association Cells x'03', and a Volume Cell x'04'.
(e) Use of the CATRECNR to merge the observations together in the different datasets is problematic when AIX datasets are involved. This change tracks the sphere cluster name, the aix cluster name, and the component name; these are output into the Vsam, Association, and Volume datasets. Also, the Generation Dataset (GDS) Name is output when writing the Volume dataset, updating the tracking of volumes, so that the volumes are associated with the specific Generation Dataset. -Additional protection/corrections made Dec 23, 2008. Thanks to Ken Sharpe, Oklahoma Dept of Human Services, USA. Change 26.254 Support for MAINVIEW for CICS 6.1 CMRDETL (INCOMPATIBLE). EXMVCIC1 and support for 6.2 new data. EXMVCIC2 The new 'F7'x records contain expanded 12-byte time data, EXMVCICC so this change is required to process the new release. EXMVCICE The support for the File segments is enhanced, with new EXMVCICO datasets for CONNECT/SESSION, DB2, DBCTL, MQ and OTHER, EXMVCICQ with their unique variables, so the existing CMRFILE IMACMVCI dataset now contains only the CICS FILE segment data. VMACMVCI The complete set of datasets now created are: VMXGINIT dddddd Dataset Description Nov 9, 2008 MVCICS CMRCICS MAINVIEW CICS CMRDETL DATA Nov 10, 2008 MVCICC CMRFCON MAINVIEW CICS CMRDETL FILE CONNECT MVCICE CMRFSES MAINVIEW CICS CMRDETL FILE SESSION MVCICF CMRFILE MAINVIEW CICS CMRDETL FILE CICS MVCICO CMRFCON MAINVIEW CICS CMRDETL FILE OTHER MVCICQ CMRFCON MAINVIEW CICS CMRDETL FILE MQ MVCIC1 CMRFCON MAINVIEW CICS CMRDETL FILE DBCTL MVCIC2 CMRFCON MAINVIEW CICS CMRDETL FILE DB2 Thanks to Henk van der Veur, Fortis, THE NETHERLANDS.
VMACVMXA as the support for that record was added for z/VM 5.3 Nov 7, 2008 and expected an 88-byte, versus 84 byte, record segment. Note that this was VERY archaic VM/ESA data being tested as part of a QA job, and is NOT a current z/VM error.
ANALACTM A character operand was found in the %EVAL function GRAFWORK or %IF condition where a numeric operand is required. GRAFWRKX when the SAS WORK option was WORK=HFS:/usr/lpp/.../work. GRAFWORX Those forward slashes were seen as divide-by operations READDB2 by the macro compiler. By replacing IF %UPCASE(&xxx) VMXGDEL syntax with %IF %QUPCASE(&xxx) function, that %QUPCASE VMXGGETM function "quotes" the text strings, preventing the macro Nov 7, 2008 compiler from seeing the slashes as an operator. Only VMXGINIT VMXGDEL caused an error, but all IF %UPCASE(&xxx) Nov 10, 2008 statements, where &xxx could contain a libref text, were identified and corrected in the other listed members. However, in addition, VMXGINIT was revised, to add a test for a forward slash for unix (like the existing test for a back slash for NT) when it sets the "WORK" librefs. Thanks to Brian Watts, Dept of Education, Employment, AUSTRALIA. Thanks to Chris Weston, SAS ITRM Development, USA.
FORMATS had 'E2C4'X='SD:BY ADI WITH DATA' but that is changed to Oct 31, 2008 have 'E2C4'X='SD:BY ATI WITH DATA'. Thanks to Clayton Buck, UniGroup, Inc, USA. Change 26.250 -The four variable sets G1vvvvvv-G4vvvvvv in BVIR33 have VMACBVIR the data for the first four clusters, but documentation Oct 31, 2008 of this Hnode Grid Historical Record says there can be a Nov 13, 2008 maximum of eight clusters, so new sets G5vvvvvv-G8vvvvvv are now created, labeled, and kept. -Variable DVMXMT00 was not kept due a typo in the KEEP=. Thanks to Harald Seifert, HUK-COPBURG, GERMANY. Thanks to Jens Mohring, HUK-COPBURG, GERMANY. Thanks to Francois Vancoppenolle, Rainbow ICT, BELGIUM.
VMACHSM _SHSMWWF and _SHSMWWV still contained DATA steps rather Oct 30, 2008 than PROC SORTs, which could cause NOT SORTED errors. The DATA step is coded when new datasets are created, as the PROC SORT NODUP can't be added until I have test data to know what BY variables are required for duplicate removal. I assume I just forgot to go back when I finally got ABARS test data to validate their _Bdddddd macros. Thanks to Christine Wong, MMSA, USA.
ANALSMF MACRO _MYCISIZ is defined, so &MACKEEP can be used to Oct 29, 2008 change that definition. Last update to ANALSMF was 1996! Thanks to Stephen Hoar, Lloyds TSB, ENGLAND. Change 26.247 Support for SMF 113. Incomplete, do not use. VMAC113 Test SMF data is needed to finish writing the support Oct 29, 2008 for this new monitor record. See Change 27.002.
VMACRMFV wrong, by a factor of 1000, because their informat was Oct 28, 2008 incorrect. Now, all three are correctly input as ASIPHTMA &PIB.4.3 /*PREEMTEABLE*CLASS*SRB*TIME*/ ASIPHTZA &PIB.4.3 /*PREEMPTABLE*SRB*FOR ZAAPS*/ ASIPHTZI &PIB.4.3 /*PREEMPTABLE*SRB*FOR ZIIPS*/ ASIPHTMA was informat PIB4., and is now TIME12.2 format, ASIPHTZA and ASIPHTZI were informat PIB4.6. Thanks to Graham Harris, Royal Bank of Scotland, ENGLAND. Thanks to Matt Ellis, Royal Bank of Scotland, ENGLAND. Thanks to Rob D'Andrea, Royal Bank of Scotland, ENGLAND.
VMACTPMX in dataset TYPETPMX, but labeled them as UNKNOWN. They Oct 27, 2008 now renamed and correctly labeled: JXSLMNGD='JXSLM*MANAGED' JXSLSRVC='JXSLM*SERVICE' Using an old MXG Version (prior to 26.08) to read SMF records with either field caused this MXG message: ***ERROR.VMACTPMX. VARNAME=$JXSLMJ_ NOT FOUND, TOKFIELD=$JXSLMJ_ WAS NOT INPUT. TOKENID=53613 TOKENID=D16D TOKNAME=$JXSLM FLENGTH=1 Thanks to ???, BOA, USA. Thanks to Nancy DiFilippo, MVS Solutions Inc, USA. Change 26.244 Variable MIPS was not being calculated for WORKLOAD=0, GRAFWRKX the uncaptured work, causing confusing graphs when the Oct 27, 2008 MSU and MIPS were compared. Thanks to Jorge Fong, DOITT NYC, USA. Change 26.243 Support for OA21140 - High Performance FICON (zHPF) and FORMATS RMF HiperDispatch enhancements. VMAC7072 -TYPE70 existing SMF70Q00-SMF70Q12 counts of IN AND READY VMAC73 users previously were based on the number of processors VMAC74 that were online; now, with this APAR installed and with VMAC79 HiperDispatch Active, the count is based on the number of Oct 27, 2008 processors being ONLINE AND NOT PARKED when the sample Nov 2, 2008 was taken. Nov 10, 2008 -TYPE70 new variable SMF70NRM='zIIP*NORMALIZATION*FACTOR*/ -TYPE70PR field SMF70POF creates two new variables: POLARITY='POLARIZATION' 0='0:Horizontally Polarized or not indicated' 1='1:Vertically Polarized Low Entitlement' 2='2:Vertically Polarized Medium Entitlement' 3='3:Vertically Polarized High Entitlement' Variable POLARITY is decoded by new MG070PO format. POLARCHG='POLARITY*CHANGED?' value of Y if polarization was changed during interval. -TYPE73 dataset has these new zHPF variables: CHFACTV ='FICON*OPERATIONS*CONCURRENTLY*ACTIVE' CHFDFER ='FICON*OPERATIONS*DEFERRED*PER SEC' CHFRATE ='FICON*OPERATIONS*PER SEC' CHFXACTV='ZHPF*OPERATIONS*CONCURRENTLY*ACTIVE' CHFXDFER='ZHPF*OPERATIONS*DEFERRED*PER SEC' CHFXRATE='ZHPF*OPERATIONS*PER SEC' SMF73ECP='CHANNEL*PATH*IDENTIFICATION' SMF73EIX='INDEX TO*EXTENDED*CHANNEL*MEASUREMENTS' SMF73EOC='FICON*COMMAND MODE*OPERATIONS*ATTEMPTS' SMF73EOD='FICON*COMMAND MODE*OPERATIONS*NOT INIT' SMF73EOS='SUM COUNT OF*COMMAND MODE*OPERATIONS' SMF73ETC='FICON*TRANSPORT MODE*OPERATIONS*ATTEMPTS' SMF73ETD='FICON*TRANSPORT MODE*OPERATIONS*NOT INIT' SMF73ETS='SUM COUNT OF*TRANSPORT MODE*OPERATIONS' -TYPE74 dataset has this new variable: R744FLPN='PARTITION*IDENTIFIER*OF CF' -TYPE79C dataset for subtype 12 has the same new zHPF fields as those added to TYPE73, above, but the SMF73 is changed to R79 to be consistent with existing TYPE79s. Thanks to Dr. H. Pat Artis, Performance Associates, USA. Thanks to Brian Currah, Independent Consultant, CANADA. Change 26.242 Enhancements to internal utility functions/programs. ANAL115 -VMXGOPTR is used to store, set, and restore SAS options, ANALCISH but only supported one option. Now, multiple options can ANALRMFR be handled, but multiple new values must be separated by ANALTCP exclamation points, because values can contain blanks: VGETOBS %VMXGOPTR(OPTNAME=OP1 OP2 OP3,NEWVALUE=V1!V2!V3); VMXGOPTR -VGETOBS could fail if no DDNAME was specified. Oct 27, 2008 -ANALCISH, ANALRMFR, ANALTCP, and ANAL115 all now use the VGETOBS macro instead of the archaic internal OBSCHEK.
VMACVMXA BROKEN CONTROL RECORD error message. The CONTROL RECORD Oct 23, 2008 was fine: any MXG out-of-alignment in z/VM MONWRITE data Oct 30, 2008 surfaces with that error. The MXG circumvention to skip the hundreds of extra bytes in the record based on NRCPUS was defective and those extra bytes are now unilaterally skipped, to eliminate the circumvention and exposure. This problem was only observed in Release 3.5 data; the data from z/VM 5.4 does not have any extra bytes. -Divide by zero in daccumulation of VXAPLTCP dataset due to insufficient BY list - variable CLUSNAME must be used. Thanks to Nick Altieri, Wachovia Bank, USA. Thanks to David Schumann, Blue Cross Blue Shield of Minnesota, USA.
VMACCIMS value, even though it's input as TODSTAMP8, so it is now Oct 20, 2008 TIMEPARTed and formatted TIME12.2. Thanks to Kenneth D. Jones, Bell Aliant, CANADA Thanks to Bruce Perry, Bell Aliant, CANADA
IMACUOW VMXGUOW, is now in comments in IMACUOW, and all three of Oct 17, 2008 the examples are documented completely in both members. Thanks to Chris Weston, SAS ITRM Development, USA. Change 26.238 Utility to PROC PRINT with the LABEL and VARIABLE NAME as UPRINDOC the header is enhanced so you can select how many obs are Oct 17, 2008 printed and change the Line Size if desired. This is NOT an elegant implementation, but it works, and is very good for investigation of new data sources so you can see both the name and the label. Every dataset in the PDB library is PRINTed, and a PROC MEANS N MEAN MIN MAX; is run with all observations in each dataset, to show the statistics of each numeric variable. Previously, it was hard-coded to print only the first 9 obs, (for the ADOC members) with fixed line size.
ASUMTAPE is now corrected, and the ASUMTAPE dataset now is output Oct 17, 2008 with the _LSUTAPE macro, defined as &PSUTAPE..ASUMTAPE, to be more consistent with MXG naming conventions. The previous output was just to &PDBMXG..ASUMTAPE, but as the default for PDBMXG and PSUTAPE are both //PDB, and it is unlikely that you would have changed, this change SHOULD be transparent. Thanks to Scott Barry, SBBWorks, Inc, USA.
VMAC7072 CPUPATTM could be missing in PDB.TYPE70, causing PCTMVSBY Oct 17, 2008 to be too large and CPUMVSTM to be too SMALL, with wrong values also in variables SHORTCPS & PLCPRDYQ, if a system had offline CP engines (see Changes 26.197 and 26.192). Those engines have IFARRAY=CP, but their parked time is a missing value. The real culprit was this add statement: IF IFARRAY(LCPUADDR+1)=0 THEN CPUPATTM=CPUPATTM+PATWAIT(LCPUADDR+1); which sets CPUPATTM to missing when PATWAIT is missing. I should have added a test for PATWAIT non-missing, but I instead now use the SUM() function in this revision: IF IFARRAY(LCPUADDR+1)=0 THEN CPUPATTM=SUM(CPUPATTM,PATWAIT(LCPUADDR+1)); to preserves the CPUPATTM value when PATWAIT is missing.
re-reading the SMF data, using this program: DATA NEW.TYPE70; SET OLD.TYPE70; IF CPUPATTM=. THEN DO; CPUPATTM=SUM(OF CPUPATM0-CPUPATM9 CPUPATMA CPUPATMB CPUPATMC CPUPATMD CPUPATME CPUPATMF CPUPATMG CPUPATMH CPUPATMI CPUPATMJ CPUPATMK CPUPATML CPUPATMN CPUPATMO CPUPATMP CPUPATMQ CPUPATMR CPUPATMS CPUPATMT CPUPATMU CPUPATMV CPUPATMW CPUPATMX CPUPATMY CPUPATMZ CPUPATZA CPUPATZB CPUPATZC CPUPATZD CPUPATZE CPUPATZF CPUPATZG CPUPATZH CPUPATZI CPUPATZJ CPUPATZK CPUPATZL CPUPATZM CPUPATZN CPUPATZO CPUPATZP CPUPATZQ CPUPATZR CPUPATZS CPUPATZT CPUPATZU CPUPATZV CPUPATZW CPUPATZX CPUPATZY CPUPATZZ CPUPATYA CPUPATYB CPUPATYC ); IF CPUPATTM GT 0 THEN CPUMVSTM=CPUUPTM-MVSWAITM-CPUPATTM;/*-SMF70PAT*/ ELSE CPUMVSTM=CPUUPTM-MVSWAITM; IF CPUUPTM GT 0 THEN DO; PCTCPUBY=100*CPUACTTM/CPUUPTM; PCTCPUEF=100*CPUEFFTM/CPUUPTM; IF CPUPATTM GT 0 THEN /*PER OA24074*/ PCTMVSBY=100*CPUMVSTM/(CPUUPTM-CPUPATTM); ELSE PCTMVSBY=100*CPUMVSTM/CPUUPTM; END;
IF CPUACTTM=. AND CPUPDTTM=. AND PCTCPUBY=. AND PCTMVSBY GT 0 AND CPUMVSTM GT 0 THEN DO; PCTCPUBY=PCTMVSBY; CPUACTTM=CPUMVSTM; END; IF PCTMVSBY GT 0 AND PCTCPUBY GT 0 THEN DO; SHORTCPS=PCTMVSBY/PCTCPUBY; PLCPRDYQ=100*(PCTMVSBY-PCTCPUBY)/PCTMVSBY; IF . LT PLCPRDYQ LT 0 THEN DO; SHORTCPS=1; PLCPRDYQ=0; END;
END; ELSE DO;
SHORTCPS=.; PLCPRDYQ=.; END; END;
RUN; Thanks to Frank De Bree, DEXIA, BELGIUM. Thanks to Christine De Clercq, DEXIA, BELGIUM.
IEBUPDTE that was created on Windows caused characters '0D'x (CR Oct 15, 2008 or Carriage Return) to be treated as a data byte. This is because when unix files are written, only a '0D'x CR is written to terminate each line, while Windows files are terminated with a '0D0A'x CRLF (CR plus Line Feed). So SAS under unix only looks for an LF line terminator. To get SAS under unix/linux to read a Windows file and not store the '0D'x as data, the TERMSTR=CRLF option must be specified on the INFILE statement. This id documented in the SAS Companion for unix INFILE note at: http://support.sas.com/documentation/cdl/en/ hostunx/59542/HTML/default/chifoptfmain.htm Fortunately, TERMSTR=CRLF works under Windows, so it can be added unconditionally to the INFILE statement and now that program will run on all ascii platforms. Thanks to Steve Clark, DHL IT Services Americas, USA. Thanks to Jan Squillace, SAS Technical Support, USA. Change 26.234 Strange ORACLE SMF records, with none of the offsets that VMACORAC are expected, but with an offset in a formerly reserved Oct 15, 2008 field, followed by variable length text data, are now detected and the first 500 instance printed for tests. Thanks to Diane Eppestine, IBM Global Services, USA.
READDB2 the text of Change 25.090; now, READDB2 is enhanced to Oct 14, 2008 create both datasets T102S225 and DB2STAT4 when IFCID=225 has been requested; observations from DB2 V8 or V7 will be in T102S225 and from V9 or later in DB2STAT4, but the variable names are the same. Jan 2009: See Change 26.311 which actually made the change correctly in READDB2.
FORMATS but the Installation Exits manual discussion of MEMLIMIT VMAC30 under IEFUSI (how's that for obscure SMF documentation) Oct 10, 2008 notes that that value is set when MEMLIMIT is NOLIMIT. Unfortunately, there's no easy way out, to print NOLIM for that value, because MEMLIMIT is FORMATted with the standard MGBYTES decoding format, used for all byte vars. So, new MG030ME format is created MEMLIMIT, and the above hex value sets MEMLIMIT=. so NOLIM is printed. Thanks to Dan Case, Mayo Clinic, USA. Change 26.230 Variable QPACPAC was incorrectly set in DB2ACCTP whenever VMACDB2 variable QPACCLS7='Y'. QPACPAC was set from the same bit Oct 7, 2008 in QPACFLGS after QPACCLS7 had been set. Thanks to Don Cleveland, Wellpoint, USA. Change 26.229 MQ variables QWHCPST and QWHCPSB for IMS access were VMAC116 only correct for WTIDATYP=3; the MXG test for IMS should Oct 4, 2008 have input those fields for 3 or 4, but code had 2 and 3. Thanks to Scott Barry, SBBWorks, Inc, USA. Change 26.228 Example to report DASD Storage Group summarization using ASUMSTGP MXG's DCOLLECT data. GRAFSTGP - ASUMSTGP creates PDB.ASUMSTGP summary by storage group TRNDSTGP and DSNAME. Oct 3, 2008 - TRNDSTGP creates PDB.TRNDSTGP trending summary - GRAFSTGP produces graph of allocated and used space from the TREND data. Larry Douty, ExxonMobile, USA. Change 26.227 Previously, VMXGSUME protected for variables that did not VMXGSUME exist in the incoming datasets, but changes to VMXGSUM Oct 3, 2008 now provide that same protection, so there is no need for a separate VMXGSUME member. So now, VMXGSUME will just bring in the standard VMXGSUM member, and any references in your code to use %VMXGSUME can be changed to %VMXGSUM, or any %INCLUDE SOURCLIB(VMXGSUME); can be removed, but those statements do not HAVE to be changed, as long as you do NOT have your own VMXGSUME member in tailoring libraries. Change 26.226 Reserved Change Number. Oct 2, 2008 Change 26.225 Variable QPACAAFG was still wrong after Change 26.080, as VMACDB2 it is INPUT in two places, but only the first was fixed. Oct 2, 2008 Thanks to Glenn Bowman, Wakefern, USA. Yüklə 28,67 Mb. Dostları ilə paylaş:
|