* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
Sep 8, 1986 be consistent with example in the book, and report ANALTURN title identifies units. Change 04.100 Cleanup of DATABASE data set built from TYPE1415 and Sep 8, 1986 step records. Unreferenced variables from step record ANALDSET are now deleted, saving space and avoiding confusion. Thanks to Eustace Fernandez, Bow Valley Industries, CANADA. Change 04.99 Support for IDMS log data and exit code to create SMF Sep 21, 1986 records from IDMS log data. This is in addition to the IDMSEXIT IDMS Performance Monitor SMF record supported by MXG IDMSLOG Change 4.61. All documentation is in member IDMSEXIT, IMACIDMS which contains the ASM source for an IDMS exit which will create four SMF records with IDMS log-type data VMAC200-203 without reading the IDMS DCLOG. Members TYPE200-3 will TYPE200-203 create labeled MXG data sets from these MXG IDMS SMF EXTY200-203 records. Member IDMSLOG will read the DCLOG itself, but variables are not labelled (yet). Member IMACIDMS defines the actual MXG IDMS SMF record IDs. There are additional reports in IDMS.... members. Thanks to Peter Bailey, Software Product Services, Woking, ENGLAND.
Sep 12, 1986 impacting changes in data format (DOSJOBS especially). TYPEDOS Five new data sets are now created by MXG from V2: FORMATS DOSBEGIN (startup statistics), DOSNET (network stats) EXDOSBEG DOSPOOL (spool accounting), DOSXPCC (connection), and EXDOSNET DOSXRC (transmit/receive spool). New member IMACDOSA EXDOSPOL now defines DOSJOBS variable APLICATN; code had been EXDOSXPC inline. Read comments at beginning of member TYPEDOS IMACDOSA for all V2 documentation and DOS changes Thanks to Paul Ehresmann, Compusource, USA. Change 04.97 Several cleanup changes by 4.1 ESP user. TYPE75 data Sep 6, 1986 not created if RMFINTRV run standalone. UTILCICS had RMFINTRV missing semicolon in comments. Two CICSTRAN variables, UTILCICS IWAITCN & IWAITTM were not in KEEP list (they exist VMAC110 only if DLI counting is enabled in member IMACICDL). Thanks to Norbert Korsche, OMV-AG, AUSTRIA. Change 04.96 Several 4.1 ESP changes. Inconsistency between JES2 Sep 6, 1986 and JES3 _NODUP macro corrected. OUTDEVCE & SYSTEM BUILDPDB appended to TYPE6 bylist in NODUP SORT to avoid dupes BUILDPD3 which had been encountered. Division by DURATM in type VMAC74 74 protected for DURATM=0. CPUTM added to PDB.JOBS. Thanks to Bill Cohen, Drexel Burnham, USA. Thanks to M. Morris, Northern VA Highway Department, USA.
Sep 6, 1986 manual were coded. I have never had type 59 records VMAC59 to test, and no site has ever validated this code.
Sep 6, 1986 and will now process compressed data, if the INFILE EXITMONI exit TMON (JCL + SOURCE are in EXITMONI) is installed. IMACMONI IMACMONI defaults to uncompressed format only. Once TYPEMONI TMON is installed, either format is supported. Thanks to Bob Rutledge, Sherwin Williams Paint, USA. (I can see why - they know he's so valuable you'll try to steal him).
Sep 5, 1986 without PP43887. Test should be at 51 with that PTF. VMAC110 MXG was updated to test for PTF and then test at 50 or 51 as appropriate. Thanks to Glen Wall, Databank, NEW ZEALAND.
Sep 5, 1986 a 3084 was split into two 3081's. Only if the CPU was VMAC7072 online during the entire interval will the CPU and its contribution to wait be counted. If a processor is offline at any time during the interval, it will not be counted in NRCPUS and its wait will not be added to CPUWAITM. Now, the CPUWAITn variable for that CPU will contain the actual wait for that CPU (before this change, it's CPUWAITM was set equal to DURATM). Thanks to Andy Yu, B.C. Systems-Hi Tech Systems, CANADA. Change 04.91 Substantial changes in DB2 data. MXG did not correctly Sep 4, 1986 capture all of the DB2 data segments. See description VMACDB2 in comments in VMACDB2. ANALDB2
Thanks to Martha Hall, Metropolitan Life, USA. Change 04.90 Type 40 (dynamic allocation) record did not include Sep 4, 1986 +OFFSMF in line 67, causing error if VSAM SMF file VMAC40 was read by MXG. Minimal impact, as 40 data is in 30 Thanks to Chuck Hopf, Computer Language Research (FASTAX), USA. =====Changes thru 4.89 were ALPHA Version 4.2========================= Change 04.89 Support for the ROSCOE Response Time Monitor records Aug 4, 1986 (new in ROSCOE Release 5.4). Reports are provided, TYPEROSC and Chuck Hopf's paper (in DOCRRTM) discusses the VMACROSC good and the not so good of this response data. This JCLROSC code is preliminary; variables are 8 rather than the DOCROSC 4 bytes they will be, etc. The reports match okay.
Aug 4, 1986 detect when operators have varied tape drives online ANALVARY simultaneously to multiple systems. Thanks to Chuck Hopf, Computer Language Research (FASTAX), USA. Change 04.87 SAS/GRAPH reports using RMFINTRV data were revised to Aug 4, 1986 use SAS Version 5 Graphics catalogs. CLSTRPLY allows GRAFRMFI TSO users to REPLAY these graphs. CLSTRPLY
Aug 4, 1986 in the example. IMACPUXA
Thanks to Huddie Dean, Chilton Corp Dallas, USA. Change 04.85 VM Monitor Support is now credible. The PERFORM, USER, Aug 4, 1986 and DASTAP records are almost complete, and the VM MAP TYPEVMON reports have been mostly validated. Read all of the notes and comments in VMACVMON. More reports still remain, and some important variables are not yet deduced by MXG (help IS solicited), but tests under VM HPO 3.4 look good. No HPO 4.2 data has been tested, but I think the code will work for VM Release 3 and 4, with or without HPO. Thanks to Allan Russell, SAS Institute Europe, GERMANY. Thanks to Daniel Delorge, SAS Institute, FRANCE.
Aug 3, 1986 product) data uncovered a few errors which are fixed. TYPEMONI Most of the variables are already documented in MXG Book Chapter 40 (CICSTRAN and CICSYSTM sections), but also note that THE MONITOR only provides the elapsed (attach to detach) response time; the IRESP (internal response, ELAPSED-WTTCIOTM, which excludes the user think and typing time) is not captured by THE MONITOR. This change affected many lines to the MXG 4.1 code. Thanks to Neil Ervin, Borg Warner Chemicals, USA. Thanks to Bob Rutledge, Sherwin Williams Paint, USA.
Jul 17, 1986 segment, which was required for OUTPUT. This fix adds VMAC80 %%INCLUDE SOURCLIB(EXTY8) to handle these cases. Thanks to Tom Wiebe, NERCO, USA. Change 04.82 INPUT statement did not include +OFFSMF offset to read Jul 11, 1986 VSAM SMF data correctly. Normal SMF data was handled VMAC8911 okay. Thanks to Chuck Hopf, Computer Language Research (FASTAX), USA. Change 04.81 VMONCU data was incorrectly documented by IBM. An obs Jul 10, 1986 will be created only if the device actually had busy VMACVMON measured from CPU or from APU. Thanks to Terry Magill, NAS, USA. Change 04.80 CICSTRAN IRESPTM (Internal response) was non-zero even Jun 30, 1986 if ERRFLAG was on, because it was carried from prior VMAC110 transaction, because it was not initialized to missing before each calculation. Thanks to Evangeline Jacobs, Stearns Catalytic, USA. Thanks to James Tummins, Stearns Catalytic, USA. Change 04.79 GDGATTR changed to PIB1., HEX2. from character, as Jun 22, 1986 it conflicted with VMAC6367. (Not caught by UTILXREF VMAC6156 because neither variable is actually KEPT, and only caused problem if 6156 and 6367 simultaneously built.) Thanks to Malcolm Morgan, Wachovia Bank, USA. Thanks to Norbert Riehout, Great Western, USA. -----Changes thru 4.78 were in Pre-Release of Version 4.1. Change 04.78 Correction to Change 4.55 for RACF 1.7. The OFFSET Jun 9, 1986 equations added by 4.55 was restored to the original: VMAC80 OFFSET=OFFSET+1+OFFSMF; Type 80s from TOPSECRET need OFFSET=OFFSET-3+OFFSMF, still under investigation with that vendor. Thanks to Tom Weibe, NERCO, USA. Change 04.77 CICS Transaction variable TRMCHRCN, total characters Jun 9, 1986 in and out, was not created for CICS 1.7. (It was an VMAC110 INPUTed variable in 1.6). Thanks to Barbara Watters, EXXON Houston, USA. -----Changes thru 4.76 completed May 31, 1986 Release of Version 4.1. Change 04.76 Variable CPUTM in the ROSCO... data sets was renamed May 31, 1986 to CPUTCBTM to be consistent with its meaning. Note VMACROSC that CPUTM was removed, which may cause your ROSCO... report programs (other than ANALROSC, which we fixed to use CPUTCBTM) to fail. (We had to correct this one, because the ROSCO label for CPUTM could override the correct TYPE30 label.) Sorry I didn't catch this one when I validated the original ROSCOE code. Change 04.75 Queueing percentages are now correct. EVENTS should May 31, 1986 have been the divident instead of NRSAMPLES. VMAC77
May 31, 1986 in different MXG data sets were resolved. Changes in DOC LABEL were to clarify and be consistent, but should not really be noticed; PHYSICAL*BLOCK SIZE*(BYTES) replaced BLOCK*SIZE, for example. Change 04.73 Change 4.65 was redesigned, and its description below May 31, 1986 was changed after printing of MXG Newsletter EIGHT. VMAC110
May 29, 1986 it didn't raise an error) due to IBM format changes we TYPEVM overlooked. Thanks to Jonathan Aliber, Fidelity Systems, USA.
May 28, 1986 records was expanded to also test for nulls (hex 0) as BUILDPDB NJE purge records are null, whereas JES are blank. Thanks to Al Loyd, DOD, Ft. Meade, Md, USA. Change 04.70 AVGRSPTM (average response time) variable added to the May 28, 1986 NLDM Type 39 data sets. VMAC39
May 27, 1986 you were supposed to (first). MXG EXIT Facility is now INSTALL documented in member INSTALL (essentially as it will be in Chapter 33 of the MXG Supplement). Additional sections in INSTALL cross reference member to product to MXG data set created to EXIT member names. Read it. Change 04.68 Member TYPEDOS was restructured to extend the MXG EXIT May 27, 1986 facility to DOS. See comments in TYPEDOS, and INSTALL. TYPEDOS It should not impact present DOS data sets in any way, EXDOS... unless you choose to enable it. Thanks to Barry Lewis, CIA, USA.
May 27, 1986 comment block around the SMF processing) did not keep RMFINTRV all of the RMF data sets which are kept when RMFINTRV is invoked by BUILDPDB. Standalone RMFINTRV now agrees Thanks to Barry Lewis, CIA, USA.
May 27, 1986 processors can cause BUILDPD3 to not recognize that BUILDPD3 the job is complete. _TIMEDIF macro is now defined IMACTIME in IMACTIME to allow installations to specify a time "fuzziness" for BUILDPD3 logic. Read IMACTIME. Thanks to Barry Lewis, CIA, USA. Change 04.65 Incorrect PTF information in IMACPTF can cause CICS May 31, 1986 data sets to be built with no apparent error, but VMAC110 containing invalid data (notably, TASKNR missing.) This was because MXG used the ?? marker to supress the dump of the record, the error message and _ERROR_ set by SAS for TASKNR. The ?? has been retained, but an MXG ERROR MESSAGE now notifies you that the CMF data format has been changes, and suggests you run UTILCICS which will identify your PTF level by APPLID. THE BAD OBSERVATIONS (TASKNR=.) ARE NOW DELETED FROM CICSTRAN. Thanks to Joseph J. Faska, Chemical Bank, USA. for lots of good input here.
May 19, 1986 Insert RT1DCDTW=DATEJUL(RT1DCDTW); Format 12.2 of VMACRTE RT1DCTIT,TTT changed to 8., TST,TUT,TWT to 14.4; IO is PAGES READ + PAGES WRITTEN; BST reports show BATCH for Program Type when Type Task is Batch.
May 13, 1986 Systems. Variable names were taken from existing MXG TYPEMONI data sets CICSTRAN and CICSYSTM (book, page 398) when FORMATS possible. The code has only been syntax checked; the MONITOR data tape did not arrive. Two test sites will validate before you read this; call before believing. Change 04.62 Split into five steps and better documented. Version 4 May 13, 1986 broke the SAS limit of 32768 SYSIN cards (it is fixed JCLTEST in SAS Version 5.15). JCLTEST MUST BE EXECUTED WITH A TESTIBM NEW VERSION OF MXG. Self documenting. New naming code: TESTUSER members used to test MXG will begin with TEST. TESTOTHR Old MXG testing members TYPETEST,OTHRTEST,TYPEPRNT, & OTHRPRNT are deleted in Version 4 to force the issue.
May 12, 1986 record. Called RTE because it was born as the Run Time TYPERTE Evaluator product of Business Software Technology. The FORMATS SMF record type (their default 230) must be enabled in EXTYRTE IMACRTE. The monitor provides task level resource data IMACRTE
Thanks to Myles McCarthy, Fidelity Systems, USA. Change 04.60 Support for RMF System Availability Management data May 12, 1986 records. Three data sets SAMINFO, SAMTERM & SAMSKED TYPESAM can be built from data captured by the new RMF/SAM component introduced in RMF 3.4. SAMTERM & SAMSKED are from installation data, SAMINFO is created from RMF data sent to the SAM Info/MGT data base, extracted with a SAM supplied utility. SAM detects system level availability events (IPLs, WAITs, LOOPs), and step terminations, but nothing at the subsystem level.
May 12, 1986 had omitted multiply by 16). In the process, all VMAC110 duration (TM) variables which had been created by (PIB4., then /62500) are now instead created with the equivalent (PIB4.6, then *16) for consistency. Thanks to Barbara Watters, EXXON Houston, USA. Change 04.58 Support for type 24 JES2 SPOOL OFFLOAD Program May 12, 1986 Product. Separate information for jobs offloaded VMAC24 and SYSOUT files offloaded from SPOOL. FORMATS
May 12, 1986 VTAM Tuning Statistics record. The existing SNA VTAM VMAC50 data is supplemented with new data for VTAM CTC use. Change 04.56 Support for BDT (Bulk Data Transfer) type 59 SMF May 10, 1986 record has been coded. It has not been validated by VMAC59 actual data records. Call if you have verified it.
May 9, 1986 decoded with several new formats. VMAC80 FORMATS
May 8, 1986 statements to avoid redundant SAS statements. This DOC has the added advantage that these MXG data sets will now have all variables in alphabetic order. VMAC7072 VMAC71 VMAC73 VMAC74 VMAC75 VMAC76 VMAC77 VMAC78 VMAC30
May 7, 1986 match RMF reports of same values. VMAC75
May 6, 1986 values which are now corrected. EXTFRMON is PIB4.0, VMAC71 not PIB4.1; MIGAGEAV is PIB4.1, not PIB4.0; and PVTMVTOT is now a rate per second with correct label. Change 04.51 New documentation shows the new measure of all path May 6, 1986 busy should be PCTPTHBY=100*PCTPTHBY/NRCMPTSM; VMAC78 Comparison of MXG PCTCUBSY (using RMF manual equation) does not exactly match % CU BUSY field on RMF report; usually close, sometimes MXG shows 26% while RMF only has 17%. Checking with IBM, think we are right. Thanks to Jan Van Lent, Fokker BV, Schiphol, THE NETHERLANDS.
May 6, 1986 prior day's date if the interval ended after midnight. VMAC38 Subtract NPATM from SMFTIME inside DATEPART function. Thanks to Roger Konydyk, Steelcase, USA. Change 04.49 IMACPTF added. A new maintenance facility for MXG in May 6, 1986 which MXG code to support certain IBM PTFs (Program IMACPTF Temporary Fixes) is enabled. Currently used only by VMAC110 to support CICS PTFs. Check here first if you find a PTF which changes data; it may already be supported in MXG, awaiting only your enablement here. Change 04.48 CICS PTF PP43887 adds new field in middle of record. May 6, 1986 For sites using CICS IRC (Inter Region Communication) VMAC110 the new variables WTIRIOTM and WTIRIOCN give the total time (and count of times) that this transaction waited for IO from the other CICS IRC region. You must update IMACPTF to enable _PP43887 when this PTF is installed. With the PTF installed and not enabled in MXG, the data in CICSTRAN variables will be wrong. Thanks to Sheldon Auerbach, AmeriTrust Cleveland, USA.
May 6, 1986 MXG _VAR.... macro which contains KEEP= list for all BUILDPDB/3 MXG data sets) must be INCLUDEd after EXPDBINC. Thanks to Malcolm Morgan, Wachovia Bank, USA. Change 04.46 Line 73 assumed OFFSET was from beginning of QSAM May 6, 1986 data, but it is from beginning of the BSAM record. VMAC80 Change +1 to -3 (to account 4-byte RDW difference). Thanks to Craig Feitzner, Citicorp, USA. Change 04.45 Support for Database 2 Release 2 changes to type 100 Apr 22, 1986 and 101 SMF records. VMACDB2 DB2STAT0: Q9STCTRC, QWSDCKPT, QWSDINV1-4 FORMATS DB2STAT1: QBSTIMW,"SEQ,"SPP,"SPD,"REE,"WEE,"DWT."DMC QISECT,"CTL,"CTG,"DBD,"DBDG,"DBDL,"FREE, "SKCT,"FAIL,"PAGE, QXALABON QTXALES,"LEN,"NPL DB2ACCT : QWACAJST,"ARNA,"ARNE,"ASC,"ASRB,"AWTI,"AWTL QXALABON
QTXALES,"LEN,"NPL QBACSEQ
decoded by MXG; when pushed by a user with data in hand, we will investigate its complexity and eventually support it. Change 04.44 Support for the NPDA type 37 SMF record. Apr 21, 1986 TYPE37 VMAC37
FORMATS Change 04.43 CMS Support, for the execution of MXG under CMS SAS. Apr 8, 1986 Installation instructions are contained in this new INSTALL member.
Apr 8, 1986 can be created from VM monitor records are documented TYPEVMON in INSTALL. All other documentation is in VMACVMON VMACVMON comments. Note that you must have SAS Version 5.12 or later to read Monitor data from the virtual reader (which needs MONITOR exit, new with 5.12). Change 04.41 All PROC DELETE's were changed to PROC DATASETS NOLIST Apr 8, 1986 because SAS no longer supports PROC DELETE. DOC BUILDPDB/3 ANALDB2,DOS,ESV,ROSC
Apr 7, 1986 was incorrect; when overflow occurred (almost never, VMACCIMS except for occasional big BMPs), overflow count was added to wrong bucket. Thanks to Ron Root, Sun Company Dallas, USA.
Apr 7, 1986 to a SAS error in the NODUP option of PROC SORT when BUILDPDB NODUP and KEEP= or DROP= were used. NODUP compared the BUILDPD3 records (wrongly) using the output buffer length. With KEEP or DROP, the output record is shorter than input, causing SAS to falsely detect and delete non-duplicate records. SAS HAS FIXED THE PROBLEM IN 5.15 AND LATER. (See SAS usage note 2079 for 5.08 and earlier). To enable duplicate data removal for all SORTS after you install SAS 5.15, EDIT BUILDPDB/3 and change the MACRO _NODUP % to MACRO _NODUP NODUP % as described in comments. This disabling of the NODUP option only affects the job records (6, 26, and 30); the majority of the PROC SORTs in BUILDPDB do not use KEEP= and thus NODUP is enabled. When duplicate records are encountered by the NODUP options, SAS tells you with a log NOTE. Thanks to Dale Ingold, SAS Institute Cary, USA. for quick response and repair.
Mar 26, 1986 be automatically deleted from the PDB files, and a BUILDPDB SAS NOTE on the LOG will tell you that duplicate data BUILDPD3 was found and deleted for each SAS data set. Note that this will still only remove identical SMF data in today's input SMF file. Re-processing data today (which was previously processed) would not be detectable by the NODUP options. This enhancement provides most of the function which the non-existent ANALDUPE member was intended to perform. Removal of duplicate data is done by the addition of the SAS NODUP option to the first SORT of the input data sets, and by adding variables to the BY list to force uniqueness for the NODUP option. Since the high level BY variables in those SORTs were not changed, your present reporting should not be impacted. You may want to take advantage of the new sort order, since it ususally will eliminate a sort in your reporting. Because CICSTRAN data set is created direct from the SMF data, there is no SORT in BUILDPDB/3 and there is no automatic duplicate removal in BUILDPDB/3. In your CICSTRAN report programs you normally must sort the data anyhow. To remove dupes, simply add NODUP to the PROC SORT and expand the BY list to include all the Yüklə 28,67 Mb. Dostları ilə paylaş:
|