* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
BUILD606 _CDE macros for performance, but putting _CDE30 ahead of Jan 20, 1998 _CDE26J2 caused variable JOBCLASS to be 8 bytes instead of 1 byte. (JES2 has a one byte JOBCLASS, JES3 has three bytes kept, but the variable is input at $EBCDIC8 in the type 30 logic.) By placing the _CDE26J2 processing prior to the _CDE30 processing for the JES2 BUILDPDB, the JOBCLASS variable will be 1 byte in JES2 and 8 in JES3. Thanks to Tom Parker, CSC Hogan Systems, USA.
VMAC26J2 type 26 ACCOUNTn fields to be not input. The test Jan 20, 1998 (SMF26OAG+SMF26LAG-1 LE LENGTH) should have been (SMF26OAG+SMF26LAG-4 LE LENGTH). Thanks to Tom Parker, CSC Hogan Systems, USA.
VMAC28 Change the test that sets COFTYPE='VCD' to read: Jan 19, 1998 ELSE IF 0D0X LE NPMSUBTY LE 0DBX OR NPMSUBTY EQ 0DDX THEN COFTYPE='VCD'; Thanks to Mel Hitchings, British Telecom, WALES.
VMAC28 the KEEP= list contained _VA28ACD, but that should have Jan 15, 1998 been _VA28NCD, which is the shorter configuration record that is in NPMINFRP. Variable names that were misspelled are corrected: LFRPOBOL is LFRPOBQL, LXLKTSLP/LXLKRSLP is LXLKTLSP/LXLKRLSP and LXPUTSLP/LXPURSLP is ....TLSP/RLSP. Thanks to Ley Teng, QANTAS, AUSTRALIA.
UTILUOW error: the member UTILUOW was added at the last minute, Jan 15, 1998 and I failed to change its "./" IEBUPDTE control cards to "XY", so when IEBUPDTE read the tape to create your source library, those "./" cards in UTILUOW overwrote all of the (previously unloaded successfully!) EXCICxxx members. Replacement tapes with Jan 15, 1998 date (and with XY vice ./) were shipped to the 23 sites getting the bad tape, but you could copy the tape and delete lines 864,653 thru 865,135 inclusive, and then use that copy as input to IEBUPDTE. Thanks to Freddie Arie, Texas Utilities, USA. ======Changes thru 15.338 were in MXG 15.08 dated Jan 14, 1998====== Change 15.338 Further cleanup and enhancements to eliminate unneeded ANALCISH datasets (the INT, EOD, USS, RRT, and REQ datasets are Jan 14, 1998 not used by any ANALCISH report, so they are no longer created when PDB=SMF is specified, for example). You can select from more than one PDB library for input with the PDB=TUE WED THU, syntax, and handling of the many WORK.TRANSxx merges was revised. Thanks to Bruce Widlund, Merrill Consultants, USA.
AIXPDS has been revised and enhanced from the original unix Jan 14, 1998 examples in member VMACRSxx. Those members and new ones, including JCL examples for FTPing the unix data to MVS, are provided in this find enhancement authored and contributed by Joe. See the comments to expand the member into a separate MXG.RX6000.SOURCLIB library. Using the unix commands to measure unix performance is marginal at best, but if you do not have a monitor product for unix, you can get useful information by running the PSSTAT, IOSTAT and/or VMSTAT commands and using this code to turn them into MXG datasets. Thanks to Joseph J. Faska, Depository Trust, USA.
EX028VTT variables to the CSL, FRP, GBL, TRI, VCD, VEN, VGB, and IMAC28 VVR segments, and new dataset NPMVSVTT is created from VMAC28 from new subtype 'DD'x record. Many new variables Jan 14, 1998 provide HPR counts (frames/bytes sent/rcv/discarded). These changes were COMPATIBLY made by IBM, so earlier versions of MXG will not fail with NPM 2.3 records, but this change is required to capture the new fields. Thanks to Melanie Hitchings, British Telecommunications, WALES.
UTILUOW to determine which transaction record contains the real Jan 13, 1998 transaction name for each unit-of-work. With the output of UTILUOW, you can edit IMACUOW to tell MXG which name to keep, so that dataset ASUMUOW will have the "correct" TRANNAME name for the UOW. Read the member's comments. Thanks to Chuck Hopf, MBNA, USA.
VMXGSUM there were no observations in the input dataset, but now Jan 13, 1998 it is always deleted. Thanks to Chuck Hopf, MBNA, USA. Change 15.333 Support for type 79 subtype 15 (IMS Long Lock) record has EXTY79F only been visually tested, as no IMS site has sent any IMAC79 test data. For each Long Lock, you get the IRLM lock VMAC79 structure name and Lock Name, as well as identity of the Jan 11, 1998 IMS subsystem and even CICS task name (if CICS).
ADOCx as of MXG 15.07; this is the output of an extensive piece Jan 9, 1998 of work by Freddie that merge new variables and datasets into the existing ADOC members, while preserving the text descriptions, tutorials, discussions, and the sample PROC PRINT and MEANS output. This iteration makes all of the ADOCs conform to the format described in ADOCxxxx, so that this process can be fully automated and will soon be an automated part of the MXG QA jobstream. Thanks to Freddie Arie, Lone Star Gas, USA. Change 15.331 OMEGAMON FOR VTAM 400, Change 15.296 was not complete, VMACOMVT amd caused MXG to loop on an INPUT statement. All byte Jan 9, 1998 related variables are now formatted with MGBYTES. MXG now creates new interval rate, or "Per Sec", variables for each pair of ACCUM/PREVIOUS counters. These new variables, suffixed with R, should be most useful. The ACCUM/PREVIOUS counters could be dropped if space becomes an issue. However, the OMVTMPCS pair ON08BCT and ON08BCTP for one interval produced a negative value for the "Per Sec" value, ON08PCTR, because the ACCUM of 104,425 was less than the PREVIOUS of 108840! This will be investigated with Candle. Thanks to Joseph E. Darvish, Farmers Insurance Group, USA. Thanks to Robert S. Miller, Farmers Insurance Group, USA. Change 15.330 APAR OW28613 corrects an error in SMF26OAG; without the VMAC26J2 APAR, MXG had INPUT STATEMENT EXCEEDED error condition. Jan 8, 1998 To circumvent the error until you get the APAR installed change the test of SMF26OAG to read as follows: IF SMF26OAG GT 0 AND SMF26LAG GT 0 AND SMF26NAG GT 0 AND (SMF26OAG+SMF26LAG-1 LE LENGTH) THEN DO; (The bad record I saw has LENGTH=417 but SMF26OAG=439!) Only variables ACCOUNTn in dataset TYPE26J2 (which are new additions) are affected by the error, and the will be blank until you install the APAR. Furthermore, MXG uses the type 30 ACCOUNTn fields in BUILDPDB normally, and will use the new type 26 fields only for job's with JCL errors that did not execute. Thanks to Diane Eppestine, Southwestern Bell, USA. Change 15.329 The BUILDPDB logic for the SMF processing phase is now BUILDPDB more efficient, as the _CDExxxx macros are reordered to BUILDPD3 process the most frequent records first (the order is BUILD606 110, 30, 74, 100-101 and then the rest), and the _CDExxxx BUIL3606 macros are now inside ELSE IF ID= xxx then DO; blocks. BUILD001 This will result in significant reductions in CPU time! EXTYID Also, the ID dataset that was hardcoded in BUILPDx code IMACID has been externalized so it can be tailored to contain TYPEID additional variables (in IMACID's _KTYID macro) or can VMACID contain zero observations (by commenting out the OUTPUT Jan 7, 1998 statement in member EXTYID). See Change 15.354, which redesigned this change for more efficiency. Change 15.328 New macros _Sxxyyyy are defined in IMACs that will let IMACs you sort individual datasets exactly as that dataset is Jan 7, 1998 sorted into the PDB library in BUILDPDB/BUILDPD3. The _Sxxyyyy name matches the _L, _K macro names and the EXxxyyyy exit member naming conventions. The sort macro is needed by sites that have so much SMF data that they have to split their SMF data into parallel SMF files. An example of their use would be for a site that split their type 74 RMF records to a separate SMF file. To create the TYPE74 and TYPE74CA datasets into separate SAS data libraries they could now use:
//TYPE74 DD DSN=YOUR.TYPE74.PDB,DISP=NEW //TYPE74CA DD DSN=YOUR.TYPE74CA.PDB,DISP=NEW %INCLUDE SOURCLIB(TYPE74); %LET PDB74=TYPE74; _STY74
%LET PDB74CA=TYPE74CA; _STY74CA
MACRO _STY74 PROC SORT NODUP DATA=_LTY74 OUT=&PDB74..TYPE74 ; BY SYSTEM STARTIME DEVNR SMFTIME; %
unsorted TYPE74 dataset, and sorts it into the new &PDB74 macro for the DDname that was just added by Change 15.320. (The %VMXGFOR macro, needed so that PROC SORT's FORCE option is used so that you can SORT when OBS= is not OBS=MAX, has double percent signs because it is inside an old-style MACRO definition). A later MXG Technical note will discuss other examples. Only those IMACs for products that are in the default BUILDPDB/BUILDPD3 now have _Sxxxyyy macros, but over time I hope to provide an _S macro with the MXG "recommended" sort order for every MXG dataset. At the present time, none of the _S macros are used in the BUILxxxx members. Thanks to Billy Westland, Litton Computer Services, USA.
VMAC42 XRC (eXtended Remote Copy) SMF records, but the volume Jan 7, 1998 data is present only if the volume being copied is in a duplex state. I need sample data records before I can proceed to add support for that APAR, so if you can send sample data for a duplexed volume copied with XRC, do it! Change 15.326 Cosmetic. Labels were added for variables DCSNAME and NTINTRV PRODTYPE and NRCPUS, andNRCPUS is now kept in dataset VMACNTSM NTCONFIG. In NTINTRV, variables DCPQUED and DCPRATE are Jan 7, 1998 now correctly spelled as DPCQUED and DPCRATE. Change 15.325 The number of RACF DTP (44) segments kept is increased VMAC80A from 6 to 12 for RACFEVNT 10 and 13, and the MXG message Jan 7, 1998 was relocated so the total count in the record is now printed. See Change 15.289. Thanks to Tom Parker, Hogan Systems, USA.
VMACDCOL Variable DCDVOLSR is now RETAINed from the preceding Jan 7, 1998 dataset record (DCURCTYP='D', for DCOLDSET dataset) so that dataset DCOLCLUS now contains DCDVOLSR. Thanks to John W. Etter, StorageTek, USA. Change 15.323 All packed decimal inputs are now protected with ?? input VMACZARA modifier so that hex zeros won't create a SAS log note. Jan 5, 1998 The file record contains System ID at location 295, so new variable FILESYST is now created. Both statements IF FILEDATL=0 THEN DATECR=.; were typos and now read: IF FILEDATL=0 THEN DATELU=.; Thanks to Harry Olschewski, DeTeCSM/SES, GERMANY. Change 15.322 MXG 15.07 only. Remove SMTNTASK from the KEEP statement ANALCISH 91 lines after DATA &MXGDS1. Change SMTRANI to SMSTRNI Jan 3, 1998 in all 57 occurrences.
VMACRMFV -In dataset ZRBASI, variables ASITOTSV, ASISVINR, ASIGSPPI Jan 3, 1998 and ASIGASPD were wrong because ASICUSE, ASITOTD, and ASISRVO should have been input as &PIB.4. instead of as &PIB.2. New variables ASIOREPL, ASIOTOTU, and ASIIOU (new with OS/390 R1.1) are now created in ZRBASI. -In dataset ZRBGEI, variables GEIWLMTK and GEISPLXI should be input $EBCDIC8. instead of $CHAR8. (but under MVS the results are the same; only if you run MXG under ASCII SAS would you have noticed the error). Variable PCTCPUBY is now always missing, as its source field GEICPUOL is now reserved in OS/390. Thanks to Joseph J. Faska, Depository Trust, USA. Thanks to Peggy Dunne, Depository Trust Company, USA. Change 15.320 All hardcoded references to "PDB" as a DDNAME are now DOC replaced with a macro variable, &PDBxxxx, so that each VMXGINIT datasets can be re-directed to a different DD if needed. Jan 2, 1998 For example, if you want to write the TYPE74 dataset to Jan 12, 1998 a separate DDname of MY74, you would use: %LET PDB74=MY74; %INCLUDE SOURCLIB(BUILDPDB); to create dataset MY74.TYPE74 in the MY74 Libref. A separate &PDBxxxx macro exists for each dataset in the PDB architecture. The macro &PDB and "PDB." were changed to &PDBMXG macro name. All of the &PDBxxxx macros have a default value of "PDB". VMXGINIT defines these macros:
%LET PDBMXG =&DEFAULT; %LET PDB0 =&DEFAULT; /*TYPE0 >> PDB.IPLS */ %LET PDB0203=&DEFAULT; /*TYPE0203 */ %LET PDB6 =&DEFAULT; /*TYPE6 */ %LET PDB21 =&DEFAULT; /*TYPE21 >> PDB.TAPES */ %LET PDB25 =&DEFAULT; /*TYPE25 (JES3 ONLY) */ %LET PDB30UD=&DEFAULT; /*TYPE30_D >> PDB.DDSTATS*/ %LET PDB30UV=&DEFAULT; /*TYPE30_V >> PDB.SMFINTRV*/ %LET PDB30U1=&DEFAULT; /*TYPE30_1 */ %LET PDB30U4=&DEFAULT; /*TYPE30_4 */ %LET PDB30U5=&DEFAULT; /*TYPE30_5 */ %LET PDB30U6=&DEFAULT; /*TYPE30_6 */ %LET PDB30OM=&DEFAULT; /*TYPE30OM */ %LET PDB30MU=&DEFAULT; /*TYPE30MU */ %LET PDB70 =&DEFAULT; /*TYPE70 */ %LET PDB70PR=&DEFAULT; /*TYPE70PR */ %LET PDB71 =&DEFAULT; /*TYPE71 */ %LET PDB72 =&DEFAULT; /*TYPE72 */ %LET PDB72GO=&DEFAULT; /*TYPE72GO */ %LET PDB72DL=&DEFAULT; /*TYPE72DL */ %LET PDB72MN=&DEFAULT; /*TYPE72MN */ %LET PDB72SC=&DEFAULT; /*TYPE72SC */ %LET PDB7204=&DEFAULT; /*TYPE7204 */ %LET PDB73 =&DEFAULT; /*TYPE73 */ %LET PDB73L =&DEFAULT; /*TYPE73L */ %LET PDB73P =&DEFAULT; /*TYPE73P */ %LET PDB74 =&DEFAULT; /*TYPE74 */ %LET PDB74CA=&DEFAULT; /*TYPE74CA */ %LET PDB74CF=&DEFAULT; /*TYPE74CF */ %LET PDB74ST=&DEFAULT; /*TYPE74ST */ %LET PDB74TD=&DEFAULT; /*TYPE74TD */ %LET PDB75 =&DEFAULT; /*TYPE75 */ %LET PDB77 =&DEFAULT; /*TYPE77 */ %LET PDB78 =&DEFAULT; /*TYPE78 */ %LET PDB78CF=&DEFAULT; /*TYPE78CF */ %LET PDB78CU=&DEFAULT; /*TYPE78CU */ %LET PDB78IO=&DEFAULT; /*TYPE78IO */ %LET PDB78PA=&DEFAULT; /*TYPE78PA */ %LET PDB78SP=&DEFAULT; /*TYPE78SP */ %LET PDB78VS=&DEFAULT; /*TYPE78VS */ %LET PDBTMNT=&DEFAULT; /*TYPETMNT */ %LET PDBTSWP=&DEFAULT; /*TYPETSWP */ %LET PDBTALO=&DEFAULT; /*TYPETALO */ %LET PDBNJEP=&DEFAULT; /*PDB.NJEPURGE*/ %LET PDBJOBS=&DEFAULT; /*PDB.JOBS */ %LET PDBSTEP=&DEFAULT; /*PDB.STEPS*/ %LET PDBPRIN=&DEFAULT; /*PDB.PRINT*/ %LET PDBRMFI=&DEFAULT; /*PDB.RMFINTRV*/
SAS ITSV product can send all datasets to the //WORK file and then later copy them to their //DIRECT libref. This eliminated the needs discussed in Change 15.272. It is unlikely anyone else will need to use that option.
the change should be completely transparent to MXG users. However, users of SAS ITSV Version 2.0 and SAS/CPE must install this circumvention with MXG 15.08 or later: The following two lines of code must be inserted immediately before any %CMPROCES invocation: %INCLUDE SOURCLIB(VMXGINIT); %VMXGINIT(DEFAULT=WORK); Version 2.1 of ITSV does not need this circumvention.
BUIL3005 label datasets TYPE0203, TYPE7, TYPE21, TYPE75, and BUILD005 TYPE7204, and Labels for datasets PDB.JOBS, PDB.STEPS, BUILDPD3 PDB.PRINT, PDB.SMFINTRV and (JES2-only) PDB.NJEPURGE were BUILDPDB added in the BUILxxxx members. The labels in BUILxxxx DIFFDB2 members now contain BUILDPDB/BUILDPD3/BUILD005/BUIL3005 RMFINTRV so you can tell exactly which member created your PDB VMAC0203 library! Data set Label for RMFINTRV was added, and in VMAC21 DIFFDB2 labels were added for the de-accumulated VMAC7 statistics datasets (and they too indicate DEACCUM versus VMAC7072 the RAW tag now used in VMACDB2 for those same datasets). VMAC75 Data set labels are passed from input to output by PROC VMACDB2 SORTs, so only DATA steps were labeled. Dec 31, 1997 Thanks to Lindsay Oxenham, IBM Global Services, AUSTRALIA.
IMACACCT members that contain account fields and thus include Dec 31, 1997 member IMACACCT: VMAC20 VMAC26J2 VMAC26J3 VMAC30 VMAC33 VMAC434 VMAC535 VMACBETA VMACDB2 VMACNAF VMACORAC VMACSFTA VMACTSOM VMACXPSM XTYPEMVT XTYPEVS1 Thanks to Charlie Bloemker, Lockheed-Martin, USA.
ASMTAPES earlier release of MVS; in fact, the monitor program Dec 31, 1997 will loop under that archaic release. If you are still stuck with a back level MVS, use the back level ASMTAPES or ASMTMNT until you upgrade your MVS. Thanks to Scott Snyder, First Data Merchant Services, USA. Change 15.316 Support for OS/400 Release 4.1.0 (INCOMPATIBLE in three VMACQAPM datasets - QAPMETH, QAPMIDLC, and QAPMLAPD, because field Dec 30, 1997 lengths were changed and/or reserved fields were deleted) but the other "important" datasets were not changed. This support is based on examination of DSECTS, but has not been absolutely verified with AS400 4.1 data yet. Thanks to Len Marchant, Coca Cola, USA.
ASUMUOW merges/sums CICSTRAN and DB2ACCT into one observation for IMACUOW each unit of work. Member IMACUOW now defines new macro Dec 30, 1997 _TRANUOW that allows you to determine which CICS trans record is used to supply TRANNAME in the PDB.ASUMUOW data set that is created by ASUMUOW. For some environments, the TRANNAME wanted is the first transaction to end, but with CICS under OS/2, the first transaction name is CPMI and that mirror name is not the desired transaction name. Comments in IMACUOW and ASUMUOW discuss the three choices that now exist, and other tailoring options. Note that while ASUMUOW is automatically included in the JCLPDB6 example, so that dataset PDB.ASUMUOW will be created by default, the dataset PDB.ASUMUOW will contain zero observations until you tailor member IMACUOW; that way, there is no sort of CICSTRAN and DB2ACCT until you tell MXG you want to create observations in PDB.ASUMUOW. Thanks to Richard S. Ralston, Whirlpool Corporation, USA. Change 15.314 Cosmetic corrections (duplicate variable name in KEEP, DOC FORMAT, INFORMAT, or incorrect comment statement naming Dec 30, 1997 dataset OUTPUT, etc., were corrected in members: EXAIMRA EXAIMRD EXAIMRP EXAIMRR EXAIMRT EXAMIRS EXHPTEAP EXHPTECO EXHPTEDS EXHPTEGL EXHPTELA EXHPTEPR VMAC102 VMAC110 VMAC7072 VMAC99 VMACAIMR VMACCMFV VMACCTCP VMACDCOL VMACEREP VMACHBUF VMACHURN VMACNSPY VMACOMVT VMACSARX VMACTCP VMACTMO2 VMACTUX VMACVMXP VMACXPSM Thanks to Fredie Arie, Lone Star Gas, USA Change 15.313 Support for ICSS SMF type 103 (Internet Connection Secure DIFF103 Server, V2R2, but soon it will be renamed to Domino GO EXTY1031 Webserver V2R4, with no change in the record format). EXTY1032 Two new datasets are created for this webserver: IMAC103 Subtype Dataset Name Description TYPE103 1 TYPE1031 ICSS Configuration VMAC103 2 TYPE1032 ICSS Performance VMACSMF Many Configuration Values are decoded by formats. Dec 20, 1997 Member VMACSMF had to be updated, because the SMF 103 Jan 14, 1998 record does not correctly set the second bit in the Feb 20, 1998 first byte ("record has subtype field" bit). The TYPE1032 Performance record contains request counts like bytes received/sent, but there is no interval duration in the IBM record. As a result, member DIFF103 was required (it is automatically invoked by TYPE103) to create the interval duration variable DURATM. Although the record default interval is 15 minutes when the logging queue is full, if the server activity is minimal the logging queue fills more slowly and the DURATM will be much longer than the expected 15 minutes Note that if you add 103 processing to BUILDPDB, you will need to %INCLUDE the DIFF103 member in your EXPDBOUT member to create DURATM in dataset TYPE1032.
with an SMF header; I suspect we will see many future SMF records based on Simple Network Measurement Protocol Measurement Information Blocks! New information Feb 20, 1998: IBM is fixing the problem in the SMF 103 records and the fix should be in Lotus Domino Go Webserver 5.0 (LDGW) with GA in May, 1998. As I don't have the DSECT changes yet, you will need to get the then-current version of MXG to process the revised records (which should eliminate the need for DIFF103). Thanks to Peter Skov, Jyske Bank, DENMARK.
CHANGES after '15.07' in the %LET MXGVERS=15.07 statement, which VMXGINIT causes an ERROR message and variable MXGVERSN in TYPE70 Dec 19, 1997 to be missing. Also, the date in CHANGES was 17Dec97, but the real date (in VMXGINIT) was 18Dec97.
VMACAIM2 several fields to the existing AIM and AIM/NDB records, VMACAIM6 and new datasets from AIM/RDBII's type 98 SMF record VMACAIM7 with member TYPEAIMR. Datasets are now labeled and VMACAIMR alphabetized, and INPUT statements were collimated, but TYPEAIMR the real logic changes were discovered and contributed IMACAIMR by the cited codesharks! EXAIMRT Logic changes (i.e., incompatibilities between V12 and EXAIMRS V20) were made in VMACAIM2 and VMACAIM6 and the new EXAIMRP VMACAIMR. The other VMACs were changed only by adding Yüklə 28,67 Mb. Dostları ilə paylaş:
|