* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
Thanks to Graham Harris, Royal Bank of Scotland, UK.
EXCTCP30 are added to the existing CleverTCP User SMF record code: EXCTCP31 DDDDDD MXG MXG EXCTCP32 DATASET DATASET DATASET EXCTCP33 SUFFIX NAME LABEL EXCTCP34
EXCTCP35 CTCP30 CTCP30 CTCP CRITICAL RESOURCE EXCTCP36 CTCP31 CTCP31 CTCP PORT MONITOR EXCTCP37 CTCP32 CTCP32 CTCP LINK VIEW EXCTCP38 CTCP33 CTCP33 CTCP PROCESS VIEW EXCTCP39 CTCP34 CTCP34 CTCP ICMP STATISTICS EXCTCP40 CTCP35 CTCP35 CTCP IP STATISTICS FORMATS CTCP36 CTCP36 CTCP TCP STATISTICS IMACCTCP CTCP37 CTCP37 CTCP UDP STATISTICS VMACCTCP CTCP38 CTCP38 CTCP OSA CHANNEL VMXGINIT CTCP39 CTCP39 CTCP OSA ETHERNET Jul 31, 2014 CTCP40 CTCP40 CTCP OSA LPAR
IMACUOW When there are no CSMI transactions, there is no EXECAPPL Jul 27, 2014 that IMACUOW expected, so it is located by using byte 3 of the PATH(x) variable: If it is T or R, and not DB2, then that CICSTRAN instance is used for EXECAPPL, and for TRANNAME. Thanks to Tom MacCabe, Dominion Resource Services, USA.
VMXGDSNL VMXGDSNL could not resolve the lower level; the message Jul 27, 2014 now simply says a NULLFILE DUMMY was found.
VMACTMD2 records when executed on ASCII, or on z/OS without the Jul 25, 2014 EXITMON6 z/OS-only INFILE exit. Now, the internal SAS algorithm is invoked, but that is VERY CPU intensive and should NOT be used on z/OS. -Change 31.133 added the second iteration support for V5 but it did not note that the 'DB' Thread Detail dataset and dddddd token names were changed from TMDBDB2/TMDDB2 to TMD2DB/TMD2DB. Thanks to Ernest E. Amador, UC Davis, USA. Thanks to Mark A. Turner, UC Davis, USA.
ANALID -A Data Step VIEW is used for the (potentially large) ID FORMATS dataset when %ANALID with READSMF=YES specified, or when TYPEID %INCLUDE SOURCLIB(TYPEID); is used, or when BUILD001 or TYPEID or BUILDPDB or BUILDPD3 is executed, saving LOTS of disk TYPSID space. (Only SAS supports Data Step Views.) VMACID -The DB2 Subsystem is now ALWAYS captured for 100-102s. VMACSMF Originally, the DB2 Subsystem field was read from the VMXGINIT (compressed) Product Section, but all DB2 records have Jul 29, 2014 Subsystem in the uncompressed header, which is now used. BUILDPDB -The COMPRESS and ACCUMAC flags now print 'C' and 'A'. BUILDPD3 -On z/OS, when SMFEXIT=CICS is used (for compressed DB2 or BUILD001 CICS SMF records), records have been uncompressed by that INFILE exit, before the INPUT, so the COMPRESS=Y flag can't be set, but the DB2 IFCID value is INPUT from the product section, so reports show type 102 IFCIDs. - -Instead, when the internal decompression code is used (i.e., either on z/OS without SMFEXIT=CICS, or on ASCII), the DB2 IFCID is not available, because IBM does not populate the 102 subtype field in the SMF header, and the decompress occurs after SMF header processing; while the COMPRESS=Y flag is set, all 102s will be reported as type "102.000: UNKNOWN IFCID COMPRESSED". REVISED JUL 5, 2015: SEE CHANGE 33.159, MXGDECOMP=DB2 will now decompress in the SMF header processing. -Note that if you use IMACFILE/MACFILE (or it is used FOR you, like with READDB2) to select which SMF records are to be processed, those skipped records will NOT be counted in the ANALID reports. -BMC APPTUNE 102 records printed as 134.772-134.780 but this change revises to print 102.8004-102.800B. Thanks to Wayne Montefiore, CSC, AUSTRALIA. Thanks to MP Welch, Bank of America, USA. Change 32.174 DB2 V11 added new QX variables, but MXG only added them VMACDB2 to DB2ACCT. These variables are now added to DB2STATS: Jul 25, 2014 QXALTMP QXCREMP QXCRTSV QXDEGAT QXDRPMP QXDRPSV QXHJINCS QXHJINCT QXMAXESTIDG QXMAXPLANDG QXN1093A QXN1093B QXPAROPT QXPFMAXUG QXPFMAXUM QXPFSENUM QXPFSENUMG QXPFSLNUM QXRSMIAP QXSISTOR QXSIWF QXSTARRAY_EXPANSIONS QXSTODGNGRP QXSTOREDGRP QXWFRIDS QXWFRIDT Thanks to Steve R. Wood, DST Systems, USA. Thanks to Ramu Nalluri, DST Systems, USA.
EXWECRAU record (default 180) creates new dataset: IMACWECR DDDDDD DATASET DESCRIPTION VMACWECR WECRAU WEBSCRAU WEBSPHERE MQ CRYPTO AUDIT VMXGINIT This code has NEVER been tested with actual SMF records; Jul 25, 2014 please send records if they exist at your site. Change 32.172 Support for Websphere MQ Version 8.0 CHANNEL/CHANNEL INIT EXTY115E new subtypes of the 115 (subtype 231) and 116 (subty 10). EXTY116A create these new datasets: FORMATS DDDDDD DATASET DESCRIPTION IMAC115 TY115E MQCHIN MQM CHAN/CHANINIT STATISTICS IMAC116 TY116A MQCHININ MQM CHAN/CHANNEL INIT ACCOUNTING VMAC115 These new subtype have NOT been tested with data. Please VMAC116 send SMF data if you have these new SMF subtypes. VMXGINIT Jun 24, 2015: See Change 33.151, MXG 33.07, which updated Jul 24, 2014 VMAC115 and VMAC116 and tested with data. Change 32.171 Support for Websphere Liberty z/CONNECT SMF 120 subtype EXT12011 11 creates new dataset TYP12011. IMAC120 DDDDDD DATASET DESCRIPTION VMAC120 T12011 TYP10211 WEBSPHERE 11 LIBERTY z/CONNECT VMXGINIT Jul 22, 2014 ====== Changes thru 32.170 were in MXG 32.06 dated Jul 21, 2014========= Change 32.170 Support for CA SYSVIEW 14.0 IMS Records updates (COMPAT). VMACSVIE -New variables added to SV34TRAN dataset: Jul 19, 2014 IMTR_TRN_FPFLAG IMTR_TRN_FPFLAG2 IMTR_TRN_ENQPCB IMTR_TRN_CPUTIME IMTR_TRN_TPTDBIO IMTR_TRN_TPTDBPL IMTR_TRN_SYNCFAIL IMTR_TRN_FLIMRTCD IMTR_TRN_FLIMBQCT IMTR_TRN_FLIMIQTM IMTR_CLK_CNT_ENQ IMTR_CLK_MXG_END IMTR_CLK_CNT_GU IMTR_CLK_UOW_START IMTR_CLK_UOW_END IMTR_CLK_FLIMIQTM IMTR_CLK_FLOMPRTM IMTR_CLK_SYNCPRTM IMTR_CLK_FLDQOTIM IMTR_CLK_SYNCDATE IMTR_CLK_SYNCTIME IMTR_CLK_SAVE IMTR_CLK_IFP5901L IMTR_CLK_IFPMSGWAIT However, no V14 records have yet been available, and there may be other changes in the TIMER records that are under investigation. This text will be revised.
VMACDB2 were wrong with DB2 V10. Added by DB2 V11, the input test Jul 19, 2014 was for QDSTLEN GE 96 instead of GE 114, so they were input when they should have been missing values. -DB2STATS variables QDSTMARD and QDSTNARD were wrongly deaccumulated, causing very large, or zero, values. Thanks to Wayne Bell, UNIGROUP, USA.
VMACNDM The invalid record has NDMRECLN=1014, which should then Jul 18, 2014 have LENGTH=1028, but the record LENGTH is only 1020. Jul 23, 2014 The NDMLENPA length of NDMPACCT field in bytes 1017-1018 contains 10, but there are only 2 bytes left in the SMF record. Tests added to detect the truncated record and to print a message on the log for each defective record, and to only input as many bytes as exist. -IBM APAR PM77776/PTF UK83894 for Direct Connect V 5.x corrects the truncated record. Thanks to Norbert Wagner, Deutsche-Boerse, GERMANY.
ANALDUPE file had typos in S02OF02. The DSN syntax should be Jul 18, 2014 //FMTDAT DSN=&&KEEPFMT,DISP=(OLD,PASS) and the correct syntax is OPTIONS FMTSEARCH=(FMTDAT); Thanks to Richard Schwartz, IBM Global Services, USA.
ASUM70PR was incorrectly/unintentionally changed to QTRHOUR back SAGANAL in MXG 31.09, but the HOUR default is now (AGAIN!) set. Jul 18, 2014 %VMXG70PR (PDB=PDB,INTERVAL=QTRHOUR,CECINTRV=HOUR); Jul 23, 2014 The INTERVAL parameter controls the summarization of the two per-SYSTEM datasets, ASUM70PR and ASUM70LP, and the INTERVAL=QTRHOUR works for data at 5, 10, or 15 min for each individual SYSTEM. The CECINTRV parameter controls the summarization of the two per-CEC datasets, ASUMCEC and ASUMCELP, and the HOUR default is used because it is safer: only if ALL systems in the CEC have 15 minute interval data can QTRHOUR be used for CECINTRV, and so using HOUR protects sites with multiple/different intervals, and my intention was to always create these CEC-level datasets hourly. And, most sites have copied ASUM70PR into their tailoring library, to set their own values for INTERVAL/CECINTRV, so this incorrect change in default was not observed. -However, SAGANAL did require the CECINTRV=HOUR and that was not previously noted in its comments, hence this discovery. Now, instead of INCLUDEing ASUM70PR, the _READ70 macro uses %VMXG70PR with CECINTRV=HOUR. -Jul 23: ERROR PDB.TYPE70PR not found was corrected. Thanks to Ian Porter, Nissan, ENGLAND. Change 32.165 New parameter GROUPBY added to let you find the rankings ANALRANK by some variable. So for example, you wanted to find the Jul 17, 2014 top 20 JOBS by SYSTEM for CPUTM and EXCPTOTL, you would code:
%ANALRANK(DATASET=PDB.JOBS,GROUPBY=SYSTEM,IDBY=JOB, VARS=CPUTM EXCPTOTL,HOWMANY=20); Along the way logic was cleaned up and simplified Thanks to Tom MacCabe, Dominion Resource Services, USA. Change 32.164 FREQ created variables with lower case names. Later code VMXGSUM that went looking for the variable could fail if it Jul 17, 2014 looked for an upper case name. FREQ is now upcased as are all the other variable names. Change 32.163 Unused Change Number. Change 32.162 See Change 33.014. Required for z13. Change 32.161 Support for NDM-CDI M2 record, which is output in NDMMC VMACNDM dataset. These M2-only variables are added to NDMMC: Jul 17, 2014 NDMCFFLB NDMCRTYP NDMCSRVR NDMCTSLB NDMCTTYP NDMMCDSN NDMMCSEQ NDMNBLKS NDMNBYTS NDMNRECS Thanks to Michael Oujesky, DTCC, USA.
ASMTAPEE check for an I/O configuration change during the device Jul 17, 2014 scan loop, which should reduce the chances of logrec entries, which occurred at one site when a new IODF was activated during MXGTMNT's device scan. Other than the 2,000 logrec entries for recovered 0C4 ABENDS for the load module MXGTMNT, and the loss of data for that one interval, MXGTMNT did not fail, and the MXGTMNT job log did report the event was detected: TMNT060I I/O configuration change detected, MXGTMNT suspended pending restart TMNT061I I/O configuration restart complete, MXGTMNT processing resumed Thanks to Ed Brociek, FMR, USA. Change 32.159 -A possible error in ANALCAPD caused it to ignore the PDB= ANALCAPD parameter, so &PDBMXG was always used. Since &PDBMXG Jul 17, 2014 defaults to PDB, an error was unlikely, but the exposure is removed, and a changed PDB= argument will now be used. -Added SGPLOT invocation if you are on SAS 9.3 or higher. Otherwise, if you have SAS/GRAPH, it is used. Otherwise, the ancient PROC PLOT is used, and comments were added to document the meaning of the characters printed when PROC PLOT is used. Thanks to Andrew Woods, Interactive Data, ENGLAND. Change 32.158 Updates for Mobile Work including CSV-generating program. MOBMWRT -%MOBMWRT creates the CSV file for submission to IBM. MOBWRK05 -Changes to MOBWRKnn members now consistently have unique MOBWRK06 &MOBxxx LIBNAMEs, and both SMF and PDB processing has MOBILWRK been revised and tested. VMXGINIT -MOBWRK05 and MOBWRK06 process all five products data if Jul 18, 2014 they exist, so they can be used for one or all products.
GRAFCEC tailor the appearance of graphs. Most of these look GRAFWRKX better as landscape so the defaults are WIDTH=10in and Jul 13, 2014 HEIGHT=8in. The 'in' is required. The footnote parameter will let you add a footnote to the graphs produced that you could use to add the job name that created the graph. For example - to add a left justified footnote with a height of .5 and in red you would specify: FOOTNOTE=JUSTIFY=L COLOR=RED 'Job name' Thanks to Tom MacCabe, Dominion Resource Services, USA.
ASUMCICX response time in all transactions in the BY group. It TRNDCICX probably should have always been the average value but it Jul 11, 2014 is left as the sum and a new variable RESPAVG is created which is the average value.
EXSTCV31 records. IMACSTC -New dataset STCVSM31 created from subtype 31 VMACSTC -New variables added to datasets created from Subtypes VMXGINIT STCnnTPX added to 13,14,15,16,17,18,19,20,25,26,27,28 Jul 8, 2014 29,30 Jul 16, 2014 STCnnTYP added to 16,17,18,19 Subtype 16: STC16LOC STC16MVC Subtype 17: STC17LOC Subtype 18: STC18LOC STC18VPT STC18TND STC18ND Subtype 19: STC19LOC STC19VPT STC189ND STC19ND subtype 26: STC26VPT -All but subtypes 27,and 31 have been data tested. -These (archaic) variables are always missing values: STC13FLG STC13HID STC13SEQ STC13VTI STC14FLG STC14HID STC14SEQ STC14VTI STC18FLG STC18HID STC18SEQ STC18VTI STC19FLG STC19HID STC19SEQ STC19VTI -One question is open: STC14DSN contains a TODSTAMP and not a 44-character DSNAME. Thanks to Richard Stuchell, VISA, USA. Thanks to Bruce MacKay, Oracle, USA. Thanks to Merle Sadler, Oracle, USA.
ANALID "If you want to produce the ANALID report using UTILBLDP, UTILBLDP with BUILDPDB=NO specified, then the ID token must be the Jul 8, 2014 FIRST token in the USERADD= parameter:" Aug 15, 2014 is no longer true; Change 32.192 removed that restriction and ID can be anywhere in the USERADD= list. -While only the SELECTED SMF record's datasets will be created, ALL SMF records in the INFILE will be reported by ANALID, UNLESS you also used MACFILE/IMACFILE to delete SMF records; those deleted records will NOT be counted/reported by ANALID. -If you want to see the code that was created by UTILBLDP, it can be printed on the LOG by specifying either the new ECHO=YES (or ECHO=Y) argument, or with MXGEXIMSG=YES. -Cosmetic change in ANALID to print 0 for small percentage values to eliminate "Note: At least one W.D format was too small for the number to be printed...." Thanks to MP Welch, Bank of America, USA. Change 32.153 Support for Optional CICS User ADP fields in IMACICVH. EXUTILEX The EXUTILEX member is listed here ONLY to note that it IMACICVH is not the correct way to add USER CICS fields to MXG. VMAC110 Instead, send your CICS Dictionary SMF records to support UTILEXCL and MXG will be enhanced to support your optional data, Jul 5, 2014 with a new IMACICxx member just for you! Thanks to Patricia Hansen, ADP, USA.
VMACCTLT way to know what was the GMT Offset when those data were Jul 5, 2014 created.
VMAC74 and TYPE747C were insufficient to remove duplicates. Jul 5, 2014 All of those _BTY74xx macros are extended for removal. -Variable R742PLIN='LIST*NUMBER*WITHIN*STRUCTURE' is now INPUT; it was added to _BTY74PA. However, there are duplicate observations in TYPE74PA. -Variable R744FNAM was added to _BTY74ST for NODUP sort. -Variable R744FNAM was added to _BTY74DU for NODUP sort. -Variable R744HPCP was added to _BTY74HO for NODUP sort. There are duplicate observations created in TYPE74HO. -Dataset TYPE74ID will always have many more duplicates removed than output observations, by design. -If you use TYPE74PA or TYPE74HO, or TYPE74TD, and want to open a PMR with IBM Support to examine why there are duplicates, please send your data to support@mxg.com to verify and to document for your PMR submission. Thanks to Giuseppe Giacomodonato, EPVTECH, ITALY.
VMACPOEX records, updated after the original preliminary support Jul 3, 2014 in Change 29.134. These issues will be reported and this text updated when resolved: CHANGES REQUIRED: 1. In POEXDB2, POEXROWS are accumulated, but POEXSTMT=1 POEXSQLC=0 POEXUPDT=0 POEXDELT=0 POEXINSR=0 in all fifteen DB2 segments so it is unclear if they too are accumulated. BUT: ANY ACCUMULATED FIELDS means that data is lost; the first instance for EVERY JOB must be deleted since it is impossible to know which is the FIRST interval record. Interval records MUST contain interval values. 2. POEXUNDO (28-byte field before POEXCPUG CPU Time in General Section: Undocumented field contains IP address for IPV4 addresses: (161.236.233.152) but IPV6 addresses are truncated at 15 bytes (::ffff:165.37.5) 3. CRITICAL: Subtype 3 Interval records POEXENDT is always unpopulated (missing value) so the actual interval duration can never be known. 4. CRITICAL: Subtype 3 Interval records POEXSTRT is always the START TIME OF THE JOB and is NOT the START TIME OF THE INTERVAL. SO: the interval duration can NOT be known except in the first interval. 5. POEXSECN - Count of sections is ALWAYS ONE in all records, even though there ARE multiple sections in many records. OTHER ISSUES: 6. Issues with UNDOCUMENTED DSN1 & DSN2 in File Segments: a. Length 46 rather than 44 in DSN2, both set to 64 to be safe for open system path names. b. Contains single quotes around DSN in SOME POEXACME=NRDB2 records: 'EDWT.ISG.COMBINED.MO.PRTY.INF' ) c. Contains DSN2=BLANK, DSN1=CONNECTION d. Contains DSN2=BLANK, DSN1=TS01295.SHR.S9S.D140626 e. Contains double quotes at start and interval in some records, i.e., "ZA1P".za1racf1_RACF_RECORD 7. Variables ADDL CIPC NODE REAS RTRN SESS SSI are always blank.
8. Bytes Send and Received Count in CLIENT segment is ALWAYS 256 bytes. 9. POEXCLIE (Client): A JOB is identified by SYSTEM POEXJOB POEXTPID, but POEXSTRT is CONSTANT for each interval. In each INTERVAL record, POEXCPUG (General) is Accumulated, while POEXCPUC (Client) is the interval CPU Time. But in each END record, POEXCPUG and POEXCPUC are EQUAL and are the TOTAL for that JOB. And POEXCPUG (General) is ACCUMULATED while the POEXCPUC (Client) is the DELTA UNLESS: In POEXCLIE jobs where there is a DB2 Section: a. The POEXTPID is always zero, so it is NOT possible to group interval/end records for each job. b. The CPUG and CPUC are accumulated and interval as for POEXCLIE jobs that have FILE sections, but the CPUD from the DB2 section is ALSO ACCUMULATED in the interval records, and is the TOTAL in END record. 10. There are four CPU metrics in four segments: CPUG - General, CPUD - DB2, CPUL - Listener and CPUC - Client, but no documentation of what is or is not included in those fields. Observing values, it appears: POEXLIST - LISTENER - POEXCPUG (General) equal to POEXCPUL (Listener) POEXCLIE - CLIENT - See Preceding Item 19. 11. In the FILE segment, field POEXAMTY is not documented. The values of 01, 0Ax and 19x in POEXAMTY have POEXACME, Access Method, with , DB2, SEQ, and NRDB2, respectively. 12. POEXSTRC - Character datetime value does not contain fractions of a second, while POEXSTRT TODSTAMP does have full resolution. 13. POEXENDC - Character datetime value does not contain fractions of a second, while POEXENDT TODSTAMP does have full resolution. 14. Client Segment TODSTAMPS ENDX STRX are always missing values, but start/end from General Section are valid and kept. 15. No GMT OFFSET value in any record, but the character start time in POEXSTRC is on local while POEXSTRT is on GMT so the offset value GMTOFFPOEX is calculated and used to convert GMT datetimes to local. 16. Records with lots of nulls (SMF record 43, LENGTH=4928 (RDW=4932), but data ends in byte 847, 4181 bytes of nulls). Thanks to Eileen F. Van Etten, Bank of America, USA. Thanks to Christopher D. Carnes, Bank of America, USA. Change 32.149 SMF record TYPE can be 0-127 for IBM records or 128-255 VMACXXXX for USER SMF records. Because the TYPE number of a USER Jun 26, 2014 record is set by the site's product installer, you must Sep 28, 2014 tell MXG the record number that was chosen for USER SMF processing for SMF TYPES 128-255. MXG sets the default TYPE number to a missing value, a period, in MXG 32.10. Previously, a value of 512 was the MACRO _IDxxxx default.
the defining MACRO in the IMACKEEP member in your USERID Tailoring Library/Directory: MACRO _IDxxxx nnn % where the xxxx is the VMACxxxx suffix for the product, (documented in member IMACAAAA), and nnn is the site's chosen record number. This way, any processing of xxxx will use that definition for that product's SMF record. If you are using %UTILBLDP to create your SYSIN program, the RECOMMENDED tool to process multiple SMF records, and especially to add other SMF records, either IBM or USER, with or without executing BUILDPDB, you supply the SMF Record TYPE number in the syntax: %UTILBLDP(USERADD=xxxx/nnn yyyy/mmm . . . .); Alternatively, you can supply the _IDxxxx value in the input in the job that processes the USER SMF record: //SYSIN DD * %LET MACKEEP= MACRO _IDxxxx nnn % ; %INCLUDE SOURCLIB(TYPSxxxx);
SEE CHANGE 32.234. Thanks to MP Welch, Bank of America, USA.
READDB2 which is inserted between the QWHSTYP=1 & 2 segments in VMACDB2H BMC records, but is after the other segments in IBM data. Yüklə 28,67 Mb. Dostları ilə paylaş:
|