* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
Change 14.348 Variable JOBCLASS was increased to LENGTH $8 (from $1) VMAC30 in VMAC30 and VMAC26J3 because JES3 needs all eight VMAC26J3 bytes for JES3 Main Class. Only one byte is input in Feb 19, 1997 VMAC26J3, but by increasing the length in TYPE26J3 will May 21, 1997 cause the kept size in BUILDPD3 to be eight bytes. What is really slick is that the length of JOBCLASS in the JES2 PDB.JOBS and PDB.STEPS datasets will still be only one byte, so there is no increase in the size of the JES2 PDB as a result of this change to type 30. However, the TYPE30_x datasets will have JOBCLASS with length 8 to support either JES2 or JES3 job class names. Thanks to Jack Mintz, Hudson Williams, USA.
ANALRMFR -ANALRMFR now detects and reads PDB on tape or disk. Feb 19, 1997 -Workload Activity Report-Compatability Mode, REPORT=WKLD, RPTOPT parameter is now the way to make report selections. If none are selected no reports are created. RPTOUT=PERIOD Performance group period within performance group. GROUP Summarizes data for all performance group periods within the performance group. DOMAIN Summarizes data by domain for the entire system. STM Summarizes data for the entire system. -Workload Activity Report-Goal Mode, REPORT=WLMGL RPTOPT=WGPER: Update Sevice Policy Page column Resource Groups allowing for all service classes to be output. -Workload Activity Report-Goal Mode, REPORT=WLMGL RPTOPT=RCLASS, report classes defined in a service policy, is added. -Workload Activity Report-Goal Mode, REPORT=WLMGL RPTOPT=SCLASS, summary of data for all service class periods defined for a service class, is added. There was no data to confirm values. -Coupling Facility Activity, REPORT=CF Reports created "Coupling Facility Usage Summary"/ "Structure Summary"/"Storage Summary"/"Processor Summary" and "Coupling Facility Structure Activity" Unconfirmed values a ouput on the report as "??". -LCU summary does not include the first device. Remove LCUIORAT=0 ....... from FIRST.DATE OR FIRST.TIME OR FIRST.DEVCLASS. Move LCUIORAT=SUM(LCUIORAT,IORATE) . To after FIRST.LCU THEN DO; Thanks to David Childress, Lowe, USA. Thanks to Alan M. Sherkow, Management Strategies LTD. Change 14.346 Variables NRBINDS and NRLIMITS should have been spellec VMACTPM NBINDS and NLIMITS, and variable JCTJOBID was added to Feb 17, 1997 the KEEP= list for dataset TYPETPMF. Thanks to Brian Sanga, Eagle Star Group Services Ltd, ENGLAND. Change 14.345 The FMXGUCBL function (to allocate dynamically all DASD FMXGUCBL devices for the archaic VMACVTOC) did not support four Feb 17, 1997 digit UCBs, but now does, thanks to this contribution. Thanks to Sue Yarker, Midland Bank, ENGLAND. Change 14.344 "IHDR" exit for Boole and Babbage IMF records was added, IHDRCIMS after newsletter text was sent to the printer. See text VMACCIMS of Change 14.342. Feb 17, 1997 ===Changes thru 14.343 were printed in MXG Newsletter THIRTY-ONE======= Change 14.343 New ASUMUOW summarizes CICSTRAN and DB2ACCT by "Unit of IMACUOW Work UOW" to create PDB.ASUMUOW with total CICS and DB2 ASUMUOW resources in a single observation, keeping the original Feb 19, 1997 TRANNAME and USERID of the real UOW. Member IMACUOW will May 4, 1997 enable "SPINing" of incomplete CICS transactions (e.g., long running LU 6.2 events) to ensure completeness. This member is a replacement/enhancement of ANALDB2C that is renamed to an ASUMxxxx because it is now designed to be executed as a data-set-builder (PDB.ASUMUOW) rather than an ANALysis example (and UOW makes more sense than DB2C)! The main assumption is that the earliest transaction for any UOW must be the first transaction; during the merge, if the TRANNAME of the first transaction is CSMI (i.e., an MRO mirror) or is blank (DB2), that transaction is not complete, and the current records are SPUN (written to the SPIN library to be held until the next run, if you updated macro _SPINUOW in member IMACUOW to non-zero).
are needed to create the ASUMCICS interval summary from detail CICS transactions (plus the DB2 variables for CICS transactions calling DB2), so the ASUMUOW dataset could be used as input to ASUMCICS (you would probably want to make your own enhanced copy that also keeps the DB2 variables) so as to create interval UOW statistics. In a test of 2.6 million tran obs in CICSTRAN, ASUMUOW had only 1 million obs. ASUMUOW can be used even if only CICSTRAN data exists, as it will combine all MRO events into one observation per UOW. Text revised May, and November, 1997. Thanks to Chuck Hopf, MBNA, USA.
IHDRTMON been read in) are added for Landmark's Monitor for IHDRTMDB CICS/ESA and for Landmark's Monitor for DB2. These IMACMONI "IHDR" exits are similar to the "IMACFILE" exit for SMF IMACTMDB (which should be named IHDRSMF by my new naming TYPETMON convention). VMACTMDB Feb 15, 1997 Change 14.341 Using EXTMSTMS and macro _KTMSTMS in IMACTMS5 to create TYPETMS5 new variables did not work as expected. The new Feb 15, 1997 variables were added to the initial TMSTMS dataset, but post processing in TYPETMS5 did not include the _KTMSTMS macro reference, so the new variable was lost. The _KTMSTMS and _KTMSDSN references were added. Thanks to Andy Chandler, Eagle Star, ENGLAND.
ANALDB2R DB2 4.1. Individual Buffer Pool and Individual Package ASUMDB2A Execution data is now reported (See DB2 Technical Note in ASUMDB2P Newsletter THIRTY-ONE). The Account Detail and Long TRNDDB2A Trace reports now use the detail buffer and package data TRNDDB2P (DB2ACCTB and DB2ACCTP) if they exeist. The default Feb 15, 1997 assumption is that all of your DB2 datasets are in the PDB libref, but if you have used IMACDB2 to send them to different librefs, you can tell ANALDB2R where to find them with the new DB2ACCT=, DB2ACCTB=, and DB2ACCTP= arguments in your ANALDB2R invocation.
ANALDB2R report in that only a single PDB will be used to generate the report. (Previously, you could specify PDB=PDB1 PDB2 PDB3, and the libraries would have been "concatenated". This restructure lost that capability but you can easily combine multiple PDB datasets first and then invoke ANALDB2R.)
added; in particular, the individual buffer and package sections are complete. Thanks to Chuck Hopf, MBNA, USA. Change 14.339 User enhancements to the ASMIMSLG/ASMIMSL5 programs for ASMIMSLG IMS log processing protect for a short error message, ASMIMSL5 and in case of MSC transactions, since the ARRVTIME may Feb 14, 1997 be wrong (out of sync), now, the MSC timestamp is used for ARRVTIME with good results. Thanks to Harry Olschewski, dvg Hannover, GERMANY. Change 14.338 Variable HIUICAV (TYPE71) was added to RMFINTRV dataset RMFINTRV (the maximum of the individual values is kept) and it is TRNDRMFI added to TRNDRMFI dataset (the average of the maximum is Feb 15, 1997 kept). Thanks to Manfred Thomas, BHF-Bank, GERMANY.
ANALEPMV and ENQ data is now rolled into the EPMVEP dataset. The VMACEPMV derivation of results to match Epilog screen reports that Feb 14, 1997 doing straight-forward arithmetic with sample counts did not resemble their (proprietary) calculations but diligent work by Simon is now available in the ANALEPMV member to do those calculations for reports. mon P. Thanks to SiMundy, Telstra, AUSTRALIA. Change 14.336 Support for Landmark's The Monitor for CICS/ESA 1.5 is TYPETMON already in MXG, as the only change in the records is the Feb 14, 1997 version (LMRKVREL) contains 0Fx rather than 0Ax. Their Version 2 product is due out this year, but that will require changes to MXG to support. Stay tuned.
VMAC28 is now decoded and output in the existing NPMSUMRY data Feb 14, 1997 set. Subtype 2Ax allows the collection and reporting of session data summarized by LU Group; the value of SCDSPLUG (LU Group summarized) is stored in existing MXG variable SCDSPNAM, rather than creating a new variable. Change 14.334 Segmented TSO/MON records containing only DRU segments VMACTSOM have missing STRTTIME in TSOMDRU dataset because MXG did Feb 13, 1997 not protect for DRU-only TSO/MON records, but now does. Thanks to Rick Ireland, Imperial Oil Limited, CANADA. Change 14.333 NTSMF datasets ICMP, TCP, and UDP have been validated ADOCNTSM with data records and the table in ADOCNTSM updated. Feb 13, 1997 Thanks to Richard Clary, ENTERGY, USA. Change 14.332 New member PROCSRCE is "Proc Source" for ASCII SAS. It PROCSRCE creates a single IEBUPDTE-format file from individual Feb 13, 1997 files, so that only a single file upload or download is required to move MXG Source library between ASCII and OS/390 platforms. See instructions in member PROCSRCE. The basic technique is to pipe the DIR command's output DIR *.SAS >> NAMEFILE.MXG /B /ON to create a list of file names in NAMEFILE.MXG, then start SAS and submit %INCLUDE SOURCLIB(PROCSRCE); which reads NAMEFILE and copies each file into IEBUPDTE.MXG, inserting the ./ ADD NAME=xxxxxxxx statements.
master file that is copied to 3480 distribution tapes. The actual file on 3480 tape is about 75MB (because each record is exactly 80 bytes long) but on ASCII takes only about 40MB (variable length records with trailing blanks removed), and it PKZIPs to less than 9MB. To upload MXG for OS/390 validation, the 9MB IEBUPDTE.ZIP is sent via IND$FILE across SDLC at 19.2KB, and on MVS is unzipped by PKZIP for MVS (from ASI, 513-885-2031). With PKZIPing, transmission takes 90 minutes. Unzipped, transmission takes 7 hours!
PROCSRCE program, as it reads PROCSRCE's output file to create each member as a separate *.SAS file.
VMACNSPY the +15 and +29 after the INPUT of NSPSTAFL should be Feb 13, 1997 +16. No error was directly reported, but this might be why one European site claimed they had to change the DO interation's maximum value (reduced by one) to avoid an ABEND (but I never got a record dump for diagnosis!). Thanks to Carl Downing, BlueCross BlueShield of Minnesota, USA.
VMACAPAF data records. Changed test for SUBTYPE=31 or =32 to be Feb 13, 1997 =49 or =50 (Amdahl listed subtypes in hex not decimal!), and corrected algorithm for counting bits on in a field (see SAS Technical Note in Newsletter THIRTY-ONE). Thanks to Bob Gauthier, American Stores Company, USA. Change 14.329 SYNCSORT's product COBOL Advantage populates one byte in VMACSYNC the SYNCSORT user SMF record. New variable SOOPCAT in Feb 11, 1997 dataset TYPESYNC contains that byte, formatted in hex. Thanks to Jim Ray, Branch Banking & Trust, USA. Change 14.328 TYPE74CF (Coupling Facility) QSIZnn variables are the VMAC74 structure size as specified in the CFRM in units of 4096 Feb 10, 1997 bytes per unit, while TYPE74ST (Structures) R744SSIZ (and SMAS/SMIS) variables are actual allocated size of the structure, in units of 4000 bytes per unit, but IBM did not document the use of 4000 vice 4096. To properly size the allocations, MXG now multiplies SSIZ/SMAS/SMIS by 4000 instead of 4096 (and SSIZ will then be smaller than QSIZ values). Thanks to Don Deese, (CPExpert), Computer Management Sciences, USA. Change 14.327 Variable I was incorrectly used as the index variable VMAC102 for two nested loops, which caused VMAC102 to loop. Feb 7, 1997 There are two places, in the QWHSIID=21 and QWHSIID=44 processing that contain DO I= 1 TO 19 BY 3; In that line, in the two following lines (...PUT(SUBSTR... and in the line IF I=19 AND J=7 THEN J=20, change the "I" to "L" to correct the error. Thanks to Paul Hill, Midland Bank, ENGLAND. Change 14.326 MXG 14.11 only. Data set TYPE42S2 had incorrect values VMAC42 because "NEW" before SMF42FB2/SMF42FB3 was not removed. Feb 6, 1997 Labels for variables SMFJOD01-SMFJOD16 were corrected. Thanks to Freddie Arie, Lone Star Gas, TEXAS. Change 14.325 MXG 14.11 only. Array CFNUM should only have sixteen VMAC74 elements CFNUM01-CFNUM16 but was accidentally changed to Feb 6, 1997 CFNUM01-CFNUM64. Had no actual impact. Thanks to Bruce Widlund, Merrill Consultants, USA. ===Changes thru 14.324 were included in MXG 14.11 dated Feb 4, 1997=== Change 14.324 INVALID RECORD OBJECT=NETWORK INTERFACE error message VMACNTSM because the test ... OFFDATA NE 20 ... should be NE 21 Feb 4, 1997 (this test is right after the test for IF UPCASE(OBJECT)='NETWORK INTERFACE' THEN DO; Thanks to Richard Clary, ENTERGY, USA.
VMAC102 ERROR: NO MATCHING IF-THEN CLAUSE because the two lines Feb 4, 1997 QW0100SB=QW0100SB+GMTOFFDB; QW0100SA=QW0100SA+GMTOFFDB; were mis-located. They must be immediately before the OUTPUT T102S100; statement. Fortunately, the 100 subtype is not used in ANALDB2R or any other MXG program, so only specific use of T102S100 exposed the error (my QA caught it but I overlooked it). Thanks to Freddie Arie, Lone Star Gas, TEXAS.
ASMTAPES This revision eliminates the use of an SRB to capture ZSMTAPES job-related info, and instead does all of the cross Feb 3, 1997 memory data collection in AR (access register) mode. Running in AR mode will cause more recorded CPU time in the TMNT address space, but what is really happening is that all of our collection overhead is now captured in the (calling) TMNT address space, whereas previously some of the CPU time of the SRB's execution was recorded in the SMF records of the allocating (monitored) address space. The previous revision ML9 is in member ZSMTAPES.
support, UCB common extension access changes, and configuration change detection using ENF, and will be available later this year. Change 14.321 Don Deese's continued research into measurement of the CF VMAC74 caused me to create new variables R744QSIZ and R744QFLG Feb 3, 1997 from the QO segment into the SO per-structure segment so QSIZ/QFLG are added to TYPE74ST (while we investigate why QSIZ is often very difference thatn R744SSIZ!). The structures names in QO and SO are not in the same order, so MXG scans across the QO entries to find the matching SO entry. Thanks to Don Deese, (CPExpert), Computer Management Sciences, USA.
DOC first MXG Version (14.11) to be built at our office on Jan 31, 1997 my new Overland 3480/3480IDRC/3490/3490E tape drive. The EBCDIC to ASCII translation with IND$FILE is not the same as the Overland default, and the Overland default is also wrong (especially trashing the long-vertical-bar and the not-sign), so I had to build my own translate table to create EBCDIC tapes from my ASCII master file: MXG IND$FILE MXG WRONG MXG OUTPUT ON OUTPUT ON ON EBCDIC W/ MVS ON ASCII ASCII OVERLAND My Table cent-sign 4A 5B same AD l. Sq. Bracket not-sign 5F 5E AA caret long vertial 4F 5D B3 BD r. sq. bracket split vert 6A 7C same 4F split vert bar
typed for a not-sign on ASCII, but it is really 'AA'x). I found I also had to change Overland's (DataTools for Windows) default ASCII-to-EBCDIC table three errors 5D-->4F and not BD 5B-->4A and not AD 7C-->6A and not 4F plus my two additions AA-->5F B3-->4F To correct symbols on your ASCII platform after you use IND$FILE to download, change all '5D'x to 'B3'x and all '5E'x to 'AA'x and your comments with vertical bars and non-signs will look correct. Of course, before you could upload that ASCII MXG to EBCDIC with IND$FILE, you must reverse that change on ASCII before the upload. I took this opportunity to rid "not-signs" from all MXG code, using NE or NOT instead, and replaced all of the remaining instances of double long-vertical-bars (for concatenation) with double exclamation points, to minimize any future exposure if MXG source is moved from ASCII to EBCDIC, and removed all of the '4A'x extraneous characters in text members.
bars that were inside comments and too pretty to lose, and a few instances of "not sign" (suggesting not to use it) that were left. These members were revised: ACHAP08 ACHAP09 ACHAP12 ACHAP13 ACHAP16 ACHAP16 ACHAP17 ACHAP18 ACHAP19 ACHAP19 ACHAP21 ACHAP22 ACHAP23 ACHAP26 ACHAP30 ACHAP32 ACHAP33 ACHAP34 ACHAP37 ACHAP38 ACHAP39 ADOCEREP ADOCVMXA ADOC1415 ADOC7072 ANALCACH ANALCTLD ANALIPAC ANALRACF ANALRRTM ANALSNAP ANALVARY ANALVMOS ANALVVDS IDMSBLT IDMSJANL IDMSJRNL JCLCRAYC LOADINFO REXXCOPY REXXEMAC REXXPDB REXXTEST SYSLOGJE TTXPDS TYPEIMS UTILFMTX UTILPRNL UTILSPAC UTILXRF1 VMACIPAC VMACRRTM VMAC102 VMXGHSM XIBMFDP XLOGREC XNALCACH XNPMSESS XTYPEVS1 ZNALDB2R ZRBIPSJ
73 thru 80 are blank in all records on the MXG tape. I have not used line numbers for references for several years, and eliminating line numbers makes the source library much smaller on ASCII.
ASUM70PR which partitions are Capped, and variable LPnSHARE which Jan 31, 1997 contains the partitions percentage of the sum of the LP share values (i.e., if the partition is Capped, then LPnSHARE is the capping percentage). Thanks to Ian Porter, Nissan European Data Center, SCOTLAND. Change 14.318 Support for OS/390 Release 3 (Compatible) adds major new VMAC7072 measures to the TYPE72 and TYPE72GO datasets; the first Jan 30, 1997 five variables were added to both (i.e., COMPATIBILITY MODE in TYPE72 and WORKLOAD MANAGER in TYPE72GO): R723CICT='NON-PAGING*DASD I/O/CONNECT*TIME' R723CIDT='NON-PAGING*DASD*DISCONNECT*TIME' R723CIWT='NON-PAGING*DASD I/O*WAIT*TIME' R732CIRC='NON-PAGING*DASD I/O*SSCH COUNT' R723VELI='EXECUTION*VELOCITY*INCLUDES*I/O DELAY?' and only in WORKLOAD MANAGER do we find: PCTDLIOD='DASD*DELAY*SAMPLES' PCTDLNDI='NON-DASD*I/O USIN/DELAY*SAMPLES' PCTDLQ ='QUEUE*DELAY*SAMPLES*WAIT FOR*SERVER' PCTDLSHS='SERVER*HIPERSPACE*PAGING*DELAY' PCTDLSMP='SERVER*MPL*DELAY*SAMPLES' PCTDLSPV='SERVER*PRIVATE AREA*PAGING DELAY' PCTDLSSW='SERVER*SWAP-IN*DELAY*SAMPLES' PCTDLSVI='SERVER*SPACE VIO*PAGING*DELAY' PCTUSIOU='TOTAL*USING I/O*SAMPLES' PCTUSTOU='TOTAL*USING*SAMPLES' The calculation of VELOCITY is also changed in Release 3, as PCTUSTOU (R723CTOU) replaces PCTUSCUS (R723CCUS) in the numerator and variable R723VELI='Y to indicate that the exection velocity now includes I/O delays. Change 14.317 Dataset DB2ACCTP (Package Accounting) has new variables VMACDB2 QWACBSC and QWACESC (start/stop of plan) added to KEEP= Jan 30, 1997 list because they appear on DB2PM reports, and because Oct 13, 2003 variable QPACSCB (start of package) cannot be used, as it is null in the 2nd and subsequent segments of the type 101 subtype 0 (accounting record, holds first ten package statistics), and QPACSCB is nulls in ALL of the type 101 subtype 1 (package only) records, and finally because the subtype 1 record is written first (so I can not retain from the subtype 0!). Update: October 13, 2003: QPACSCB and QPACSCE appear to always exist, now, in DB2ACCTP observations, both from Subtype=0 (first 10 packages), and from Subtype=1 (more than 10 packages). However, QWACBSC and QWACESC will always have missing value in DB2ACCTP observations from subtype=1 because that record does not contain the QWAC segment! This 2003 note is only for documentation. Thanks to Chuck Hopf, MBNA, USA.
VMACHMF STATEMENT EXCEEDED error. Although undocumented, this Jan 29, 1997 record had only the bus segment; there was no memory segment present. To correct, insert after the @; that is after the INPUT of HMF5BUS: IF HMF5SEGS EQ 2 THEN DO; and after the following IF SKIP GT 0 THEN INPUT +SKIP @; insert END; (so that only the bus stats are read when HMFSEGS=1). Thanks to Ann Knapik, ISSC Akron, USA. Change 14.315 DB2 utility programs set QWHCATYP=41 as their attachment FORMATS type value, an undocumented value that IBM does not set. Jan 28, 1997 The site is pursuing IBM for a better answer. Thanks to Roman Jost, Gjensidigo, NORWAY. Change 14.314 Dataset MEMOACCT new variables ENTRCALE/ENTRFORM are now EXTYMEMO INPUT from previously reserved fields. In addition, the VMACMEMO MEMOACCT dataset contains variable CURRMODE with values Jan 28, 1997 of T - terminate session or C - continuation, the latter having accumulated values, so in most cases you only want the type T, so in member EXTYMEMO I have added an example of the code IF CURRMODE='T' THEN OUTPUT _LTYMEMO; that would cause only type T obs to be output, but that sample code in EXTYMEMO is commented out by default, so MXG will create both C and T observations until you change it. Thanks to Roman Jost, Gjensidigo, NORWAY.
Yüklə 28,67 Mb. Dostları ilə paylaş:
|