* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
| Change 28.120 -Variable QBACPID, Buffer Pool ID, is removed from dataset
VMACDB2 DB2ACCTP where it never belonged; packages can use many Jun 3, 2010 buffer pools. The other QBACxxxx variables exist, but are only non-missing (i.e., populated) if DB2 Accounting Class 10 is enabled. -DB2ACCTB - QLACCNVR didn't belong in KEEP list. -DB2ACCTG - QLACCNVR didn't belong in KEEP list. Thanks to Chuck Hopf, Independent Consultant, USA. Change 28.119 Support for Performance Sentry NTSMF Version 3.1.4.4. EXNTGENE -Existing objects were modified: EXNTIPA4 Dataset - Object Name EXNTIPA6 IPSECDRI - IPsec Driver - fields removed EXNTIPK4 VMGUESTA - VMWARE.Guest.Aggregate - fields added EXNTIPK6 VMGUUCPU - VMWARE.Guest.CPU - field added IMACNTSM VMGUUMEM - VMWARE.Guest.Memory - fields added VMACNTSM VMHOSTAG - VMWARE.Host.Aggregate - fields added VMXGINIT VMWHOCPU - VMWARE.Host.CPU fields added JUN 7, 2010 VMWHOMEM - VMWARE.Host.Memory fields added VMWHONET - VMWARE.Host.Net fields added VMWHOSYS - VMWARE.Host.SYS fields added -New objects supported:
NTIPA4 IPAUTHV4 IPSEC AUTHIPV4 NTIPA6 IPAUTHV6 IPSEC AUTHIPV6 NTIPK4 IPSECIK4 IPSEC IKEV4 NTIPK6 IPSECIK6 IPSEC IKEV6 -These xxxxINST variables were increased to $64: MLLOINST QLWGINST SAALINST VWHCINST VWHDINST
EXMVAOAC datasets: EXMVAOAL EXMVAOEV dddddd Dataset Description EXMVAOEX MVAOAC MVAOACTN MV AUTO OPERATOR ACTN EXMVAORE MVAOAL MVAOALRT MV AUTO OPERATOR ALRT EXMVAORU MVAOEV MVAOEVNT MV AUTO OPERATOR EVNT EXMVAORA MVAOEX MVAOEXEC MV AUTO OPERATOR EXEC EXMVAORS MVAORE MVAORES MV AUTO OPERATOR RES EXMVAOTA MVAORU MVAORULE MV AUTO OPERATOR RULE IMACMVAO MVAORA MVAORUAU MV AUTO OPERATOR RULESET AUTO TYPEMVAO MVAORS MVAORUSE MV AUTO OPERATOR RULESET SET TYPSMVAO MVAOTA MVAOTAPE MV AUTO OPERATOR TAPE VMACMVAO Aug 5: TAKETOTL removed from KEEP=, second "HANDLED" was VMXGINIT removed from LABEL, and all %VMXGDEL(DDDDDD= corrected. May 31, 2010 Aug 5, 2010 Thanks to Christa Neven, KBC Global Services, BELGIUM. Change 28.117 Change 28.039 changed R7451RID to be the first byte due VMAC74 to values observed in SMF 74 Subtype 5 records from IBM Jun 1, 2010 sites, but those observed values are not consistent in other records from other sites, which have the RID in the second byte and a zero in the first byte so this change sets R7451RID to the second byte of that 2-byte field, if if the first byte is zero. Additionally, fields listed in notes 2 and 3 of the SMF manual are set missing based on the value in R7451FLG. Thanks to Melanie Hitchings, BT, ENGLAND. Change 28.116 Support for SMF 102 IFCID 263 now decodes the unique data VMAC102 fields; previously, only Header variables were output May 27, 2010 Jun 11, 2010 Thanks to Christa Neven, KBC Global Services, BELGIUM.
ANAL42DS an _NULL_ is needed after the DATA statement. May 26, 2010 Thanks to Sam Bass, McLane Co., USA. ====== Changes thru 28.114 were in MXG 28.03 dated May 25, 2010======== Change 28.114 Additional diagnostics to show which service classes and UTILRMFI which reporting classes have missing data using SMFINTRV May 25, 2010 dataset.
ANALDB2B including a buffer hit ratio calculation. May 25, 2010 Thanks to Santosh Kandi, JC Penney, USA. Change 28.112 UTILRMFI failed because macro variable VWDUPES has been VMXGINIT truncated to WDUPES in the %GLOBAL statement in VMXGINIT. May 25, 2010 Thanks to David Young, University of California Office of Pres, USA. Change 28.111 ThruPut Manager field REGIONSZ and $JXDBSPR/WW/UU fields VMACTPMX were reported as UNKNOWN due to typo's in text tests, May 24, 2010 and $JXDBSUN is now created. Thanks to Paul Volpi, UHC, USA. Change 28.110 The last "response" bucket created by VMXGRESP utility ASUMDB2 (number specified plus 1) was never right, because every VMXGRESP count that was GT was already put in the last bucket. May 22, 2010 -Enhancement to include bucket values in labels, and a new UNITS parameter describes the units (seconds/jobs/mounts) that are being measured. -ASUMDB2 implements the new UNITS= parameter. Thanks to Wayne Bell, UniGroup, Inc, USA. Change 28.109 -WPS %MACRO Language Compiler Error in READDB2, the error ANALINIT cites a problem with %INCLUDE, but the statement that is EMAIL flagged is the percent sign in %STR(MACRO _Edddddd .... READDB2 (The _Edddddd generates %%INCLUDE SOURCLIB(EXDDDDDD);.) UTILCONT A guess, to replace %STR() with %QUOTE() in statements VFMT102 with that syntax DID circumvent the WPS compiler error! May 21, 2010 -This same error can occur with ANALDB2R if (PDB=SMF,...) is used, since ANALDB2R calls READDB2 to "READ" DB2 SMF. -But, since READDB2 generates MACRO _Edddddd references only when one of the optional/rare dataset sub-tokens is used to select the observations, for example /DB2ACCT in %READDB2(IFCIDS=ACCOUNT,DB2ACCT=YES/IF QWHSSSID='XYZ';); so it was safer to replace %%INCLUDE SOURCLIB(EXDDDDDD); statements with OUTPUT _WDDDDDD; to both circumvent the Compiler error, and because the EXDDDDDD should not have been created in READDB2 in the first place. -An unrelated error when the dataset sub-token was used is also corrected. -The WANTONLY subarguments were further validated. -In validating this READDB2 change, additional exposures were corrected: -VFMT102: If T102S105 and T102S107 did not exist, logic to build OBIDS/DBIDS dataset failed: the &VGETOBS NE 0 test was changed to VGETDSN=datasetname, as it is the existence of, and not the number of observations in, that is the needed conditional test. But this caused investigation of other &VGETOBS NE 0: -ANALINIT: VGETOBS test changed to VGETDSN. -EMAIL: VGETOBS NE 0 changed to GE 0. -UTILCONT: VGETOBS NE 0 change to GT 0 Change 28.108 Support for CleverView for TCP/IP TN3270 Performance Data EXCTCPTN in new SUBTYPE=21 SMF User Record, creates new CTCPTN32 IMACCTCP dataset with the existing TYPECTCP/TYPSCTCP programs. VMACCTCP
VMXGINIT May 20, 2010 Thanks to John Howard, Florida Northwood Shared Center, USA.
VMAC80A in TOKDANAM; MXG coded for only a 32-byte LDAP Host Name, May 19, 2010 but length was 35. Now protected for any length. -In records with LDAPHOST, new TOKDANAM BINDDN and BINDPW were found and decoded into variables TOKBINDD TOKBINPW. Yes, "PW" is a Password, but, no, it doesn't appear to be an exposure, as the value is populated with asterisks. -TOKDANAM='APPLNAM ' text is now decoded into TOKAPLNM. -TOKDANAM='UTYPE', 'JPNUM', numerics in TOKUTYPE/TOKJPNUM. Thanks to Phil Grasser, Norfolk Southern, USA. Change 28.106 The offset SMF92MPF to the Path name length SMF92PPL was VMAC92 incorrect, causing SMF92PPN, Path Name to be blank and May 19, 2010 SMF92PPL contained a large and wrong value. Thanks to David Young, University of California Office of Pres, USA. Change 28.105 -Optional TRENDing enhancement lets you tell VMXGSUM how VMXGSUM many KEEPWEEK=/KEEPMNTH=/KEEPYEAR=/KEEPDAYS='s you want May 19, 2010 kept in TREND output, or KEEPAFTR=01JAN2010 can be used to output only observations GE that DATE. -The KEEPxxxx options are only valid when DATETIME= is used, and the value of the DATETIME= variable is tested for each observation if it is to be output. -A KEEPxxxx argument forces an INCODE= to be created if one was not present; all MXG TRNDxxxx members already have an INCODE= argument, and INCODE normally VIEWs, so this should have minimal impact. -Messages are not written that obs were not output. Thanks to Diane Eppestine, IBM Global Services, USA.
VMACNDM New MACRO _NDMADD defaults to 0, but you can use MACKEEP May 17, 2010 to change _NDMADD to 8 to skip 8 bytes after SYSTEM (and to set the SMF TYPE of your NDM-CD record in _NDMID): %LET MACKEEP= MACRO _NDMADD 8 % MACRO _NDMID 0FAX % ; %INCLUDE SOURCLIB(TYPSNDM/BUILDPDB); Eight bytes can't just be added to OFFSMF in MACFILE for NDM-CD SMF because VMACSMF RETAINs OFFSMF, so new OFFNDM replaced OFFSMF in VMACSMF after the first INPUT. Thanks to Fred Buerger, Visa, Inc, USA. Change 28.103 Support for TOKDANAM='SHARED' token, output in TYPE80TK. VMAC80A
May 12, 2010 Thanks to Jacky Kwoka, SNCF, FRANCE. Change 28.102 Support for user-created ECPAUDIT optional CICS variable. IMACAAAA
IMACICUK UTILEXCL
VMAC110 May 12, 2010 Thanks to Matt Feeney, DoD, AUSTRALIA.
VMACIMS subtype test should have been TPCPSSTY, and not TPCPSBCD, VMACIMSA and the subtype test is for only one byte. May 11, 2010 -Subtype 56x SSTY 11x INPUT STATEMENT EXCEEDED exposed MXG May 18, 2010 mis-alignment; the +(-16) before TPIMSID does not exist, and the +36 segment after the "NID" is conditionally read as it doesn't exist in SSTY=0FAx records. -In validating the printed output, the SYNCTIME in TYPE59x datasets was not converted to local time. -Multiple separate code stubs to calculate IMSSTCK were replaced with LINK IMSSTCK; statements. Thanks to Lars-Olof Thellenberg, Handelsbanken, SWEDEN. Change 28.100 MXG 27.07 removed 13 duplicates in TYPE30_1 for an STC BUILDPDB that were (correctly!) NOT removed by MXG 28.02. Variable May 10, 2010 ASID has been added to TYPE30_1, and 'pseudo' duplicates had different ASIDs and thus were not duplicates. Oct 2010: See MXG Newsletter 57, DUPLICATE JESNR. Thanks to Paul Volpi, UHC, USA. Change 28.099 Variable CPULHKTM, CPU Time When Promoted due to Lock VMAC7072 HOLD, is now INPUT and kept in TYPE72GO; it was added by May 10, 2010 z/OS 1.11 as IBM field R723LPDP. Thanks to Bruce Widlund, Merrill Consultants, USA. Change 28.098 The %VMXGSET macro is enhanced with DSETOPT= argument so VMXGSET SAS Data Set Options (DROP/KEEP/RENAME/etc) can be added May 4, 2010 to each data set that will be "SET". This example syntax %VMXGSET(DATASET=JOBS,DSETOPT=KEEP=JOB READTIME JESNR CPUTM); will keep only those four variables when xxx.JOBS dataset is read. Surprisingly, using a KEEP= on a SET statement has minimal impact on CPU time, saving 7 out of 86 secs. But using DSETOPT to RENAME variables could be useful. Thanks to George Pandzik, USAA, USA. Thanks to Brian A. Harvey, HCL America, USA. Change 28.097 Type 74 subtype 8 R748LERT,R748LEWT,R748LPST variables VMAC74 are correct in MXG; they are documented as accumulated May 4, 2010 values in the SMF manual, but actual data values are not. Variable R748LAID, the Interface ID is added at the end of the _BTY748 BY list to ensure NODUP works when sorted. To verify non-accumulation, it was necessary to insert R748LAID before STARTIME, to group the observations to see if the values were always-increasing, and that is where it really belongs in the BY list, but inserting a variable in an existing BY list is LETHAL for someone, somewhere, who's already (accidentally?) combining days and weeks PDBs, so it is added at the end for NODUP, as that doesn't create a not-sorted exposure. Thanks to Chuck Hopf, Independent Consultant, USA. Change 28.096 When the LIB is in sequential or xport format, that type VGETOBS was identified, but only now is the dataset name printed. Apr 30, 2010 Thanks to Chuck Hopf, Independent Consultant, USA. Change 28.095 Elegant logic using the FINDW() function to parse the VMACAIX AIX record's changed format for the HOSTNAME field May 21, 2010 UPCRECRD=UPCASE(RECORD); LOCHOST=FINDW(UPCRECRD,'HOSTNAME:',' ','E') +1 ; HOSTNAME=SCAN(RECORD,LOCHOST,' '); HOSTNAME=TRANSLATE(HOSTNAME,' ','"'); cannot be used with SAS 9.1.3 nor with WPS 2.4.0.1, because the FINDW() function only exists in SAS V9.2. The old SCAN() function was used with less elegance.
VMACXAM record units are 16 microseconds, so they are now INPUT Apr 29, 2010 by &RB4.6 and multiplied by 16, so they are seconds and fractions of a second, and are FORMATed with TIME16.6, an MXG convention so you know these are duration variables, and so full resolution is printed, so 416 microseconds is printed as 0:00:00.000416. Variable SRMTSLIC is in TOD duration record units, so it is now INPUT by &PIB.8.6 and divided by 4096 into seconds and fractions and also FORMATed with TIME16.6. Variable SRBABSDL is a percentage, but scaled by 65536, so it is now divided by 65535. Thanks to Dave Sadler, United Health Group, USA. Thanks to Rob van der Heij, Velocity Software, THE NETHERLANDS. Change 28.093 Circumvention for SMF 42 Subtype 15 INVALID OFFSETS. The VMAC42 record has OFF42S3=216 and OFF42S4=4056, but the data are Apr 28, 2010 at 108 and 3948. MXG forces S3=108 and calculates S4. Thanks to Michael Friske, FMR, USA. Change 28.092 This ancient program to create a z/OS PDS from the MXG IEBUPDTE IEBUPDTE-format source file had defined values for the Apr 28, 2010 input and output, which required you to EDIT and SAVE and then %INCLUDE SOURCLIB(IEBUDTE); to execute, but those three macro variables are no longer set to any value, so you can set them to your choice and then run. Thanks to Bernhard Babblok, Allianz ASIC SE, GERMANY.
VMACIMS INPUT STATEMENT EXCEEDED RECORD LENGTH; support for 55x Apr 26, 2010 was added, but only tested with IMS 10 and 11. The 55x record is for External Subsystems. The 01x record isn't described in the DSECT, so it is now deleted. -The 35X record with ENQFLAG=2F and FLAG2=00X or 08X is now output in IMS35P dataset. Thanks to David Schumann, Blue Cross Blue Shield of Minnesota, USA. Change 28.090 Variable FIXEDSTO was incorrectly input as &RB.4. It is VMACXAM a &PIB.4. INFORMAT now. Variable MONEVNTS is an address Apr 25, 2010 so it is now formatted HEX8. Thanks to Paul Volpi, UHC, USA. Change 28.089 The PDB.SMFRECNT dataset and its report only counted SMF BUIL3001 records; this enhancement provides both the record count BUILD001 and the byte count for each SMF record and Subtype for BUILDPD3 each SYSTEM, in both PDB.SMFRECNT and the report. BUILDPDB If you want to revert to the count-only report, use TYPSID %LET MACKEEP= MACRO _RPDBID _RPDBIDO % ; VMACID Variable SMFBYTES was added to the TYPEID dataset and is Apr 24, 2010 the bytes variable in PDB.SMFRECNT summary dataset. Thanks to Diane Eppestine, IBM Global Services, USA.
VMACRMFV Change 28.048 inserted temp skip of +192 for z/OS 1.11 to Apr 25, 2010 align, but should have been skipped only if CPUHOLEN=672; earlier z/OS have CPUHOLEN=480. Now, SKIP=CPUHOLEN-480. Thanks to Stephen Hoar, Lloyds TSB, ENGLAND.
VMAC110 because SKIP=SKIP-580 should have been SKIP=SKIP-612. No Apr 23, 2010 fields were skipped and dataset CICRLR was just fine. -Use of internal SAS decompression code on z/OS instead of the ASM code in the INFILE exit in member EXITCICS caused CPU time to increase from .5 to 6 HOURS with 45 GB of SMF records, so the "MXGNOTE" that the SAS code was used is now elevated to a repeated "ERROR:" message under z/OS. Under ASCII, a single note that compressed data was found is printed. Change 27.260 documented a smaller increase! Thanks to Joachim Sarkoschits, DATAEV eG, GERMANY.
ADOCWPMO log file creates 123 new datasets, each named WPMOddd, to EXWPMddd correspond with the DDDDDD=WPMddd token for each dataset. IMACWPMO The DataSet Label is the Object Name, and member IMACWPMO MAKEWPMO has the full list of DDDDDD, DATASET, and Object Name and TYPEWPMO will be updated as new Objects are supported. TYPSWPMO The PERFMON data files can be concatenated, and can have VMACWPMO different sets of fields. The ADOCWPMO member documents VMXGINIT how Glenn has set up his PERFMON data management. May 22, 2010 With all possible objects were enabled, with all threads and instances captured, on a Small Business Server, the PERFMON descriptor record was 861,481 bytes long, way too long to be read on z/OS with its 32760 maximum LRECL, so MXG PERFMON support may require execution on ASCII, where SAS V9.1.3 allows LRECL=1024000, and SAS V9.2, LRECL=4GB. This design is also another significant MXG enhancement: The new MXG code that you execute to process PERFMON data was generated by MXG's MAKEWPMO program, which reads the header record field descriptions, parses/translates them to create open-systems-friendly, 32-byte variable names, with underscore-connectors, builds variable labels from those variable names, changing underscores to asterisks, uses the Object Name for the Dataset Label, and writes out the dozen text files that are then manually cut/paste into VMACWPMO or the other MXG code members. While still a two-step process, this automation makes it easy for MXG to support new objects. MAKExxxx programs have been used for some time to create chunks of MXG code for product XXXX, but only for the static statements that defined or referenced the XXXX product suffix, the DDDDDD dataset suffix token, or that contained the DATASET name. Programmatically creating variable names, labels, and the INPUT code is a BIG step forward in minimizing MY effort to support future "open systems" data or sources that have similar data structures. Since these data don't have a construct of a short field name, but only the long description, using the descriptions as variable name does make sense for this audience, and REALLY saves me TIME!!
"\\SERVER\OBJECT(INSTANCE)\METRIC" and there can be tens of thousands of items. SERVER is variable SERVER, the server name, OBJECT maps to the MXG dataset, and the Object Name (PROCESS) is the Label of the MXG Dataset Name (WPMOPRO), INSTANCE is the instance (e.g., process EXPLORER.EXE), and each separate METRIC item (e.g. HANDLE_COUNT) is a variable in the WPMOOBJ dataset.
in a character variable, with SAS's 32K limit, so the record is brute-force read, byte-by-byte, and parsed for the delimiters to create each "item". Microsoft claims to create comma-delimited data, but a single comma cannot be safely used to parse because the "items" can contain embedded commas! And, while each item is bounded by double quotes, the double quote also cannot be used as a delimiter, as "items" can also contain embedded double quotes! So the MXG default "comma-separated-value" delimiter is the three-character-string "," set in VMXGINIT with %LET DLMWPMO='","'; If you have "tab-delimited" PERFMON files, then you need to use %LET DLMWPMO='09'x; before your %INCLUDE. Once an "item" is parsed, normal SAS character functions are used to create the SERVER, OBJECT, INSTANCE and the METRIC values, and output the WINPERFMON dataset that is subsequently read to output all of the WPMOddd datasets. If there are obs in OBJUNKNOWN or METUNKNOWN datasets, please send the PERFMON file to support@mxg.com, so MXG can be updated to support the new data. Some characters in the descriptions have to be converted because blanks, back-slashes, dashes, parens, periods, pound-signs, etc., can't be used in SAS variable names; underscores are used in the variable name to connect the pieces. Thanks to Glenn Bowman, Wakefern, USA. Change 28.085 Obscure. VMXGSUM is protected if you had parenthesis in VMXGSUM the OUTDATA= argument (for a LABEL= or SORTEDBY=). If a Apr 20, 2010 last data step wasn't needed, parens caused an even more obscure ABEND. While now protected, using the DSNLABEL= or SORTEDBY= argument in the VMXGSUM invocation is safer. Thanks to Chuck Hopf, Independent Consultant, USA. Change 28.084 BMC IMF product's INPUTCLS and LASTCLAS variables are now VMACCIMS restored to one-byte length with $HEX2. format, and new Apr 20, 2010 variables INPUTCL2 and LASTCLA2 are two-bytes with $HEX4. While IMS 9.1 and later define Input Class as two bytes, it appears that no one is using that length, and MXG's changes (27.230,26.058) that expanded the length of those original INPUTCLS/LASTCLAS variables caused existing reports to be incorrect. Thanks to Jim Dammeyer, State Farm Mutual Insurance, USA. Thanks to Mike Stogsdill, State Farm Mutual Insurance, USA. Change 28.083 MXG 28.02-03. READB2/ANALDB2R didn't honor %LET MACDB2H ANALDB2R nor %LET MACKEEP tailoring: Change 28.055 added %CLEARDB2 READDB2 (to reset old-style macros for multiple executions, used Apr 17, 2010 previously only in the MXG QA), but CLEARDB2 also clears Apr 21, 2010 both MACDB2H and MACKEEP. The %CLEARDB2; statement was Apr 22, 2010 removed from both members by this change. Apr 24, 2010 Circumvention with 28.02: Create a member named CLEARDB2 Apr 25, 2010 in your "USERID.SOURCLIB" with a blank line. Apr 26, 2010 No one has asked to execute READDB2/ANALDB2R twice, Jun 17, 2010 but it can still be done, by invoking %CLEARDB2; externally, between the multiple executions: %READDB2 ( WHATEVER ); %LET MYKEEP=%BQUOTE(&MACKEEP); %LET MYDB2H=%BQUOTE(&MACDB2H); %CLEARDB2; %LET MACKEEP=%BQUOTE(&MYKEEP); %LET MACDB2H=%BQUOTE(&MACDB2H); %READDB2 (WHATEVER DIFF); -MXG 28.01-28.02 only. Short Trace report failed with Yüklə 28,67 Mb. Dostları ilə paylaş:
|