* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
EXXMVNAP and SYTSXP are added to XAMCPUBY/XAMCPUTO datasets, new EXXMVSFT variables from old segments SUBSUM,SYTRSG,SYTXSG,MTRMEM, EXXMVSYS STORSG and SYTCPU are added to XAMSYS dataset, variables EXXMVUSR from new segments PRCAPC,PRCAPM,SYTSXG,STOSXG are added IMACXAM to XAMSYS dataset, new vars from IODVSW,USRCON,USRACT VMACXAM will be decoded. VMXGINIT May 1, 2007 Thanks to Robert Obee, IMS Health, USA.
VMACNDM dataset (previously output, but with only the header May 1, 2007 variables). Variable NDMTIME is kept but it is not populated in the sample SMF record. Thanks to Trevor Ede, EDS New Zealand, NEW ZEALAND.
VMXGDUR failed to specify a DATETIME= variable that cause strange VMXGSUM syntax errors. Now, MXG issues USER ABEND 1916 when that May 1, 2007 required operand is missing. Thanks to Chuck Hopf, Bank of America, USA.
VMACRMFV ZRBLCP dataset, which now has one observation for each May 1, 2007 LCPUADDR for each LPARNAME that has engines assigned. Thanks to Jerry Urbaniak, Acxiom, USA. Thanks to Sam Knutson, Geico, USA.
INSTALL an ABORT ABEND nnnn; statement, so these programs could May 1, 2007 ABEND with the listed User Abend Code: Feb 1, 2015 USER Member USER Member ABEND ABEND 1 VMACTMV2 1234 VMACTMS5 69 VMACSMF 1911 VMXGRMFI 69 VMACTMV2 1912 VMXGRMFI 69 VMACTMVS 1913 VMXGRMFI 99 VMACCADM 1914 VMACSET 99 VMXGVTOC 1914 VMXGSET 99 VMXGVTOF 1915 VMACSET 111 VMACTNG 1916 VMMXGDUR 666 UTILDUMP 1917 UTILBLDP 666 UTILDUMP 1918 UTILBLDP 666 VGETALOC 1919 UTILBLDP 666 VMACMVCI 1954 BLDSMPDB 666 VMACXAM 2098 VMACTMVT 666 VMXGALOC 2099 VMACTMD8 666 VMXGSUM 2099 VMACTMDB 666 VMXGSUME 2099 VMACTMDC 990 VMACRMFI 2099 VMACTMDQ 991 IHDRTMS5 2099 VMACTMTC 992 IHDRTMS5 8888 VGETOBS 0001 VMACTMV2 8888 VGETOBS 0010 VMACPKSZ MXGABND VMAC110 0069 VMACTMV2 MXGABND VMACSMF 0666 IMACNORM MXGABND VMACSMF 1099 TYPEMOND MXGABND VMACTMS5 1099 TYPETMON MXGABND VMXGRMFI 1099 VMACMON8 1099 VMACTIMS 1099 VMACTMO2 1111 MONTHASC 1111 MONTHBL3 1111 MONTHBLD 1111 MONTHBLS 1111 MONTHDSK
for syntax/usage of &MXGABND USER ABEND macro variable. Change 25.075 DB2 V8.1 incompatibly changed the QBGL Group Buffer Stats VMACDB2 segment, making these nine fields now reserved: Apr 30, 2007 QBGLAD QBGLAN QBGLAR QBGLAZ QBGLMN QBGLRB QBGLRF QBGLSU QBGLXN and adding five new variables to PDB.DB2GBPST dataset: QBGLCM QBGLCR QBGLMW QBGLWP QBGLWS Thanks to Martin Packer, IBM, WORLDWIDE.
VMAC7072 debugging the SPLIT70 logic, is removed. Apr 30, 2007 Thanks to Chuck Hopf, Bank of America, USA. Change 25.073 Enhancement to "Nigel's Monitor for AIX and Linux' adds EXNMONAA variables HOST and INTERVAL to NMONINTV dataset, decodes IMACNMON the IOADAPTR and LPAR data types, and creates new NMONAAA VMACNMON dataset with the "AAA" configuration records information. VMXGINIT Apr 29, 2007 Thanks to Massimo Pugliese, ELSAG Spa, ITALY.
DALYTAPE and the MXGTMNT tape mount and allocation records. Can Apr 26, 2007 be used after a modified BUILDPDB, or with the %UTILBLDP example in the comments, the reports can be created from an SMF data file.
UTILBLDP macros when BUILDPDB=NO was specified due to a typo; the Apr 26, 2007 USERADD= datasets were not sorted into the PDB library. May 7, 2007 Additionally, these products that need deaccumulation and require their _Sxxxx product sort macros to be invoked: 7072 DB2 HSM NTCP ROSC TMDB TPX 103 28 will now have those macros invoked whenever USERADD= is invoked for them. -May 7: Support for SMF ID=102 record, which requires the _HDR102, _END102 macros. With over 300 IFCID/subtypes, you can select which T102Snnn datasets are created using USERADD=102.63 102.106 102.141 102.142 syntax. The 102 tokens must be contiguous in your USERADD= parameter, but do not need to be continuously numbered, nor do they need to be in numeric order in your USERADD= in you %UTILBLDP. Thanks to Chuck Hopf, Bank of America, USA. Change 25.070 -MXG's SYSLOG-reading program failed, INVALID ARGUMENT TO SYSLOG FUNCTION INPUT, because the final tested changed member Apr 25, 2007 was still in my "USERID.SOURCLIB" from Nov, 2005, when it Apr 28, 2007 should have been moved into the MXG source library. The Apr 29, 2007 output dataset TYPETMSG contains an observation for each SYSLOG event (but not for each physical record in SYSLOG: records starting with "S" are continuation records, and their text is appended in the variable TMTGMTXT so that only one output observation is created for these events. -Variable TMTGMTXT was increased to $120 from $80, which caused text for "S" records to be truncated. -In addition to using thid code to read SYSLOG messages, you can also write any SYSLOG message to SMF, using the MXGTMNT Tape Mount Monitor program (in ASMTAPEE), which with then be output in BUILDPDB's TYPESYSL dataset. Thanks to MP Welch, SPRINT, USA. Thanks to Stephen Van Scyoc, SPRINT, USA.
UTILRMFI symbol, if you have similarly named Service or Reporting VMXGRMFI Class names to map to workloads in your WORKnn= argument. Apr 24, 2007 Thanks to Chuck Hopf, Bank of America, USA.
ANALDB2R of text was less that 60 bytes, due to a typo; the line Apr 23, 2007 ELSE TEXT=QW0141TX; is corrected to ELSE TEXT1=QW0141TX; Thanks to Michael Gebbia, Eddie Bauer, USA. Change 25.067 The VMXGUSE macro is updated to add the _STY70 macro if VMXGUSE processing of the SMF 70 records was requested; VMXGUSE UTILBLDP is still supported, but the newer UTILBLDP is recommended Apr 20, 2007 as the better tool, as it offers many more options; it was also updated to ensure _STY70 is invoked when needed. -MXG25.03 also had double percent instead of %VMXGVERS. Thanks to Scott Barry, SBBWorks, Inc, USA.
VMACDB2H of '0CB5'x, 3253 decimal, for QWHCEUID_Off, but with its Apr 20, 2007 LRECL of only 486 bytes, caused INPUT STATEMENT EXCEEDED. The input now validates that the offset is LE LENGTH. Thanks to Robbie A. McCoy, Salt River Project, USA.
UTILBLDP be executed when BUILDPDB=YES is specified. The default Apr 18, 2007 list of members %INCLUDEd after BUILDPDB is not changed, but they are now set in the new MXGINCL= argument: MXGINCL= ASUMUOW ASUMCICX ASUM70PR ASUMCACH ASUMTAPE ASUMTMNT ASUMTALO ASUMDB2A ASUMDB2B, so you can now remove any unwanted ASUMs. Thanks to Tom Kelman, Commerce Bank of Kansas, USA.
VMACDB2 contained accumulated values; now all QISE variables are Apr 17, 2007 DIF()'d except for those that contain counts of pages or statements, in creating the PDB.DB2STATS dataset from the SMF 100 Subtype 1 DB2 Statistics records. Also, QISExxxx labels with "Dataspace" were corrected to "DBD POOLS". Thanks to Lori Masulis, Fidelity Systems, USA.
ADOC71 format and in the ADOC71 member documentation. FORMATS VALUE $MG079SR Apr 17, 2007 '00'X='00X:UNKNOWN' 'AS ='AS:AUXSTOR SHORTAGE (VERY UNLIKELY)' 'AW'='AW:APPC WAIT' 'DW'='DW:DETECTED WAIT' 'EX'='EX:CAP EXCHANGE' 'IC'='IC:IMPROVE CSTORE' 'IP'='IP:IMPROVE PAGING' 'IW'='IW:OMVS INPUT' 'LW'='LW:LONG WAIT' 'MR'='MR:MAKE ROOM' 'NQ'='NQ:CAP ENQUEUE' 'NS'='NS:TRANSITION TO NON-SWAPPABLE' 'OW'='OW:OMVS OUTPUT' 'RQ'='RQ:REQUEST SWAP' 'RS'='RS:REAL STORAGE SHORTAGE' 'SR'='SR:IN-REAL' 'TI'='TI:TERMINAL INPUT' 'TO'='TO:TERMINAL OUTPUT' 'TS'='TS:TRANSITION' 'US'='US:CAP UNILATERAL' 'VR'='VR:VIRTUAL REAL REQUEST' 'WT'='LW:LONG WAIT' 'XS'='XS:AUX STORAGE SHORTAGE' ; Thanks to Warren Hayward, TJX, USA. Change 25.062 Protection for truncates RMF ID=72 SUBTYPE=7 LENGTH=78. VMAC7072 while the site examines logs for x37 ABENDS and/or opens Apr 16, 2007 a problem report with IBM. Thanks to Ravi Kumar, CSC, USA. Change 25.061 Variable SMF12WIC is padded with '00'x instead of blanks; VMAC120 those '00'x are now translated to blanks. Apr 15, 2007 Thanks to Stan Dylnicki, Royal Bank of Canada, CANADA Change 25.060 Using TYPSTPF invoked very old PROC SORTs that still had VMXGTPFI invocations of %VMXGFOR, long ago archaic and no longer VMXGFOR needed, but they caused a 180 Syntax error for "FORCE". Apr 12, 2007 The %VMXGFOR calls were removed from VMXGTPFI, and that member was revised to "FORCE" is only created for SAS V5 and V6 where it was originally required. Change 20.327 in 2003 documented the removal of FORCE. Thanks to Bruce Widlund, Merrill Consultants, USA. Change 25.059 All %VMXGVERS(25.03,MEMBER) invocations were typo'ed as DOC %VMXGVERS(25.03MEMBER);. Wrong syntax, but, fortunately Apr 12, 2007 there was no error; the old macros would still be noted in the messages to the log, which still had the right information, albeit slightly different than expected. Thanks to Dr. Alexander Raeder, ATOSORIGIN, GERMANY. ====== Changes thru 25.058 were in MXG 25.03 dated Apr 12, 2007========= Change 25.058 Variable NDMSNODE appeared twice in the _BNDMCT by list. VMACNDM
Apr 12, 2007 Thanks to Trevor Ede, EDS New Zealand, NEW ZEALAND. ====== Changes thru 25.057 were in MXG 25.03 dated Apr 10, 2007========= Change 25.057 The 9672x CPUTYPE without APAR OW37565 does not populate VMAC7072 SMF70CIN, causing PARTNCPU to be zero. Now, MXG forces Apr 10, 2007 SMF70CIN='CP' when CPUTYPE='9672'X and NRCIXGTO=0. Thanks to Dr. Alexander Raeder, ATOSORIGIN, GERMANY. Change 25.056 DB2STATS SMF 100 Subtype 1 Language Environment QLExxxxx VMACDB2 LE variables have always been wrong, partially due to MXG Apr 10, 2007 code errors (I assumed the 8-byte fields contained both a duration and a count, and deaccumulated some fields that were not accumulated), and partially because IBM's record is incorrectly written and does not agree with the DSECT. The 8-byte header (normally has the 'QLES' eye catcher) doesn't exist, but LENQLES=168, and visual examination of hex dumps showed the extra 8 bytes are inserted between QLELOWIM and QLESTG1M fields. DB2 Versions 1.6 thru 1.8 SMF records all have LENQLES=168 and the inserted field. Thanks to Don Deese, (CPExpert), Computer Management Sciences, USA. Change 25.055 Support for SAMS objects 2151,2226,2229 and 2231 has been EXSAM151 validated and new datasets created. EXSAM226 -Many variables in many of the existing SAMSnnnn datasets EXSAM229 were renamed to avoid conflict with same-named variables EXSAM231 created from other SMF records. IMACSAMS
VMACSAMS VMXGINIT
Apr 8, 2007 Change 25.054 DB2TCBTM in DB2ACCT/ASUMUOW includes QWACSPCP,QWACTRET VMXGUOW CPU times, but with OTE, the DB2TCBTM includes some of Apr 6, 2007 the CPU time already recorded in the CICSTRAN's TASCPUTM; using PDB.ASUMUOW, the correct CPU metric would be the sum of TASCPUTM from CICSTRAN and QWACSPCP and QWACTRET from DB2ACCT. This change SUMs QWACSPCP and QWACTRET into PDB.ASUMUOW so that you can add those DB2 CPU times to the CICS CPU times with no overlap. Change 25.053 Variable PCTCPUBY greater than 100% in PDB.ASUMCEC, but VMXG70PR only in hour intervals when LPARs were started/stopped, Apr 5, 2007 and only if the DURATM of the last LPAR (alphabetically) was less than the full hour. DURATM was intended to be set to CECINTRV, but MXG set it only for FIRST.SMF70GIE; it was then being output from the DURATM of that last LPARNAME. DURATM=CECINTRV is now set unconditionally. Thanks to David Bixler, FISERV, USA. Thanks to Randy Shumate, Lexus-Nexus, USA.
EXTDS16 from subtypes 16x, 17x, 18x and 19x: EXTDS17 dddddd dataset description EXTDS18 TDS16 TDSL16 ZCOST CPC EXTDS19 TDS17 TDSL17 ZCOST LPAR IMACTDSL TDS18 TDSL18 ZCOST SYSPLEX VMACTDSL TDS19 TDSL19 ZCOST SYSPLEX/CPC VMXGINIT
Apr 5, 2007 Change 25.051 Change 24.116 reported NRIFACPU for z990/2084 CPUTYPE was VMAC7072 not countable - SMF70CIN='ICF' in LPARNAME='PHYSICAL' in Apr 6, 2007 RMF 70 records instead of 'IFA' - and created the macro variable CECIFANR so you could set NRIFACPU if needed. Now, Al has found a way using the non-PHYSICAL-LPARNAME segments (that do have SMF70CIN='IFA' for IFAs!) to count and populate variable NRIFACPU for z990s with IFAs. -New field SMF70CTN, the total count of engines of that SMF70CIN type (CP,IFL,etc), in the CEC was incorrectly INPUT, but it was not kept nor used. It still is not used for NRxxxCPU in ASUM70PR, but now it is input correctly and kept in PDB.TYPE70PR dataset. It will have the same value in all obs of that engine type, so a 27-engine CEC will have SMF70CTN=27 in all obs with SMF70CIN='CP', both in both the LPARNAME='PHYSICAL' and LPARNAME='real' obs. Thanks to Al Sherkow, I/S Management Strategies, USA. Change 25.050 Support for CrossSysplexManager user SMF record (Skill EXCSMDDN Software) creates two new datasets: EXCSMJOB CSMDDN CSMDDN CROSS SYSPLEX DDNAME IMACCSM CSMJOB CSMJOB CROSS SYSPLEX JOB TYPECSM TYPSCSM
VMACCSM VMXGINIT
Apr 4, 2007 Thanks to ???, ???, Italy. Change 25.049 Format $MGSASPR, which maps SAS Procedure Name in the SAS FORMATS "USER" SMF record, has new QUANTREG and ROBUSTRE values Apr 2, 2007 added for SAS STAT product. Thanks to Len Rugen, University of Missouri, USA. Change 25.048 Corrections to MXG 25.02 included correction for the BLDSMPDB libname=PDB operand, the CALL SYMPUT('SRTDSN',SORTDBY) DOC spelling, and logic related to VGETSORT invocation. Apr 1, 2007 References to WEEK1 were replaced with WEEK. Apr 17, 2007 New SORTEDBY= option can be set to sortedby=no to bypass the requirement for input weekly/daily datsets to be sorted. 17Apr07: Null input library protected with msg. Thanks to Loren Mitchell, Los Angeles Couty Office of Education,USA. Change 25.047 -Support for APAR OA19502, adds SMF14KET, the Key Exchange FORMATS Time (Encryption Overhead). This time is only recorded VMAC1415 in SMF 15 output records and only for a CLOSE processing Apr 1, 2007 after a non-parallel OPEN writing file sequence one from the loadpoint. Otherwise, SMF14KET is zero. This is the elapsed "wall clock" duration, and is NOT a CPU time metric. A "non-parallel OPEN" is when the parameter list passed to OPEN includes only 1 DCB or ACB. -New $MG014CD format decodes the SMF14CD1 and SMF14CD2 Encoding Mechanisms: L:LABEL - the key label itself is to be stored as part of the EEDK structure on the tape cartridge. H:HASH - "Public Key HASH" - a hash of the public key referenced by the key label is to be stored on the cartridge rather than the key label itself. Storing the hash value rather than the key label itself allows for greater flexibility when tapes are exported to another location, especially if that location may be using a different key label (than the originating site) to refer to the same key. Thanks to Jack Muehl, CitiGroup, USA. Change 25.046 Cosmetic. INVALID DATA FOR MDY message and hex dumps of VMACSTC the first two instances are printed when the LSBTIME date Mar 31, 2007 field contains zeros. Now, LSBTIME is only created when the data exists. These records with no date also have zeros in all other fields, and so the defective records were not output into the STCLMU dataset, either before or after this change. Thanks to David Kaplan, Depository Trust, USA. Change 25.045 Variable TTTOD was not kept in TPFDH, TPFKC, or TPFSB VMACTPF datasets, but were referenced in BY statement in _STPFKC. Mar 30, 2007 Now, that retained variable is kept in those datasets. Thanks to Chris Weston, SAS Institute, USA. Change 25.044 New intervals (obviously for TRENDing, and not tactical!) VMXGDUR of QUARTER, SEMIANN, and ANNUAL will create summary data VMXGSUM when those arguments are used. The code change was in Mar 28, 2007 VMXGDUR; only comments were changed in VMXGSUM. Thanks to Nalini Elkins, Inside Products, USA.
EXIODHPC (Apr 12 changes were moved Apr 17, were not in 25.03.) EXIODHPP EXIODLPT
EXMTRHPP EXPRCAPC
EXPRCAPM EXPRCDIA
EXPRCINS EXSAMSDT
EXSCLSCA EXSTOSCS
EXSTOSXG EXSTOSXP
EXSYTLCK EXSYTSPT
EXSYTSXG EXSYTSXP
EXVNDLSD EXVNDLSU
EXVNDSES IMACVMXA
VMACVMXA VMXGINIT
Mar 28, 2007 Apr 12, 2007 Change 25.042 Enhancement to support HSM records with different SMF IDs VMACHSM perhaps from different systems. HSM writes 2 SMF records Mar 28, 2007 with and ID and ID+1 architecture, and MXG defined two ID macros that you tailored with your SMF TYPE numbers: MACRO _IDHSMDS 224 % MACRO _IDHSMDS1 225 % But I have redefined the second ID macro to now be MACRO _IDSHMDS1 _IDHSMDS+1 % which has no impact on your existing IMACKEEP tailoring, but now, you could use this logic to redefine them: MACRO _IDHSMDS 512 OR (ID=224 AND SYSTEM='SYS1') OR (ID=229 AND SYSTEM='SYS2') % MACRO _IDHSMDS1 512 OR (ID=225 AND SYSTEM='SYS1') OR (ID=230 AND SYSTEM='SYS2') % and MXG will process both pairs of HSM records. This works because the reference syntax in VMACHSM is IF _IDHSMDS THEN DO; IF _IDHSMDS1 THEN DO; Thanks to Chuck Hopf, Bank of America, USA. Change 25.041 Support for CICS/TS 3.2 (INCOMPATIBLE). EXCICDHD -Major Change: All CICS CPU times are expanded to 8 bytes EXCICISR with significant increase in captured TASCPUTM possible. EXCICLDB -CICS Transaction Segments in SMF 110 subtype 1 can be EXITCICS compressed; the new MXG member EXITCICS is the ASM code IMAC110 to create a SAS Infile Exit (named "CICS") that will read UTILEXCL the compressed (or uncompressed) records. To install, VMAC110 you ASM and LinkEdit the EXITCICS code into a load module VMXGINIT that you put in a load library you add to the //STEPLIB, Apr 7, 2006 and tell MXG using %LET SMFEXIT=CICS; in your //SYSIN. Sep 19, 2006 Test records were reduced from 32K to about 6K bytes, but Oct 14, 2006 only the 110-1 records are compressible. Mar 27, 2007 -UTILEXCL updated with new CICSTRAN fields added in 3.2. Apr 12, 2007 -Updates for the changed Statistics Subtypes: Jun 29, 2007 (INCOMPATIBLE except for DST, DSR, and MNG): STID DSNAME New Variables added. - 2 SMS ==> Nothing New in DSECT. - 64 CICDST DSTCIUHI DSTCIULO DSTNIUHI DSTNIULO
==> But 20 bytes added are not doc'd. - 65 CICDSR DSRCIULO DSRCIUHI ==> But 20 bytes added are not doc'd. - 67 CICFCR DFHA17DS DSECT. ==> Nothing New in DSECT - 74 MQG MQ Connection Statistics Global New ==> DFHMQGDS DSECT does not exist - 81 MNG New Compression Variables ==> DFHMNGDS DSECT is not current. -109 ISR IP Connection (Resource) New
==> DFHUSRDS DSECT does not exist -117 SJG
==> Nothing New in DSECT. -117 SJR
==> Nothing New in DSECT. MXG 25.03 contained the critical compatibility changes, but MXG 25.07 contains the full support for compressed ID=110 Subtype=1 CICS/TS 3.2 SMF records. Change 25.040 Support for APAR OA20077, adds UNCOMPRESSED tape device VMAC21 read/write bytes to TYPE21/PDB.TAPES, complementing the ASUMTAPE existing COMPRESSED tape channel read/write bytes that Mar 21, 2007 have always been in SMF type 21 records: Apr 10, 2007 BYTEREAD='COMPRESSED*CHANNEL*BYTES*READ' BYTEWRIT='COMPRESSED*CHANNEL*BYTES*WRITTEN' SMF21DBR='UNCOMPRESSED*DEVICE*BYTES*READ' SMF21DBW='UNCOMPRESSED*DEVICE*BYTES*WRITTEN' -Apr 10: ASUMTAPE updated to keep SMF21DBR/SMF21DBW.
EXAXTCPS datasets: EXAXTCPU dddddd DATASET Description EXAXTDSK AXTCPS AIXTCPUS CPUS EXAXTFS AXTCPU AIXTCPU CPU INTERVAL EXAXTFSS AXTDSK AIXTDSK DISK EXAXTIP AXTFS AIXTFS FS EXAXTIPS AXTFSS AIXTFSS FS-S EXAXTLAN AXTIP AIXTIP IP EXAXTLPA AXTIPS AIXTIPS IPS EXAXTMEM AXTLAN AIXTLAN LAN EXAXTNFS AXTLPA AIXTLPA LPARS EXAXTPGI AXTMEM AIXTMEM MEM EXAXTPGS AXTNFS AIXTNFS NFS EXAXTPRO AXTPGI AIXTPGI PAGSP-S EXAXTSYC AXTPGS AIXTPGS PAGSP EXAXTSYI AXTPRO AIXTPRO PROC EXAXTTCP AXTSYC AIXTSYC SYSCALL EXAXTUDP AXTSYI AIXTSYI SYSIO EXAXTWLM AXTTCP AIXTTCP TCP EXAXTWLS AXTUDP AIXTUDP UDP VMACAIXT AXTWLM AIXTWLM WLM IMACAIXT AXTWLS AIXTWLS WLMS VMXGINIT
TYPEAIXT TYPSAIXT
Mar 16, 2007 Change 25.038 TMON/DB2-created SMF 101 Subtype 1 caused INPUT STATEMENT VMACDB2 ERROR because the record is defective; LENQPAC=390 but Mar 16, 2007 there were only 358 bytes of data in the segment. This INPUT EXCEEDED error is now protected, but the data in second and subsequent observations in DB2ACCTP dataset will be trashed. Thanks to Robert McElhaney, Texas Workers Commission, USA.
Yüklə 28,67 Mb. Dostları ilə paylaş:
|