* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
DCAKLN ='KEY*LENGTH* DCANLR ='LOGICAL*RECORDS' DCARKP ='RELATIVE*KEY*POSITION' DCARTR ='RETRIEVED*RECORDS' DCAUPR ='UPDATED*RECORDS' DCAVRRDS='VARIABLE LENGTH*REL RECORD*DATA SET?' DCOLDSET (Active Data Set Information) new variables: DCDALLFG='ALLOCATED*SPACE*RETURNED?'/ DCDNMBFG='UNUSABLESPACE*RETURNED?' DCDPDSEX='POSIX*FILE SYSTEM*FILE' DCDSECFG='SECONDARY*SPACE*RETURNED?' DCDSTRP ='STRIPED*DATA*SET' DCDUSEFG='USED*SPACE*RETURNED?' Additionally, if the DCOLLECT option SMSDATA is specified, these nine new MXG datasets will have observations that describe your SMS environment: DCOLAG - Aggregate Group Information DCOLBC - Base Configuration Information DCOLDC - Data Class Construct Information DCOLDR - OAM Optical Drive Record Information DCOLLB - OAM Optical Library Record Information DCOLMC - Management Class Construct Information DCOLSC - Storage Class Construct Information DCOLSG - Storage Group Construct Information DCOLVL - Storage Group Volume Information
no test data has yet been made available; with 277 new variables, I'll be surprised if it's perfect!
ASMIMSLG program to ABEND with 0C7. After the BZ READIN02 that is Feb 23, 1993 after the label READINPQ, insert this instruction: OI ARRVDATE+3,X'0F' to ensure the field is in packed decimal format. Thanks to Jeffrey S. Crum, Ashland Oil, USA. Change 10.303 TYPE1415 variable DSORG can be "PS" when really is "PO" VMAC1415 and MEMBER can be blank because the IBM record is wrong. Feb 23, 1993 If PDSs with member name specified are concatenated, the first type 14 has DSORG='PO' and correct MEMBER name, but the 2nd and subsequent type 14 records had DSORG="PS", and MEMBER was blank. The DSORG was wrong because both the JFCBORG and DCBDSORG are '40'X (PS) in the concatenation records, instead of '02'X (PO) as is in the first record. MEMBER is blank because MXG INPUTed it only for DSORG=PO (the location of MEMBER contains RELGDG if it is not a PDS and is a GDG!). Why are JFCBORG and DCBDSORG both "PS" for a PDS? Because when the member name exists in the JCL, OPEN treats the data set as sequential, even though the true DSORG (in the VTOC) is still "PO"! Since IBM will take some time to address this situation, MXG has circumvented the problem (hopefully): If MEMBER starts with + or -, it must be a RELGDG and MEMBER is blanked. Otherwise, if MEMBER is non-blank, it must be a PDS, and so MEMBER is valid and DSORG if forced to "PO". Thanks to Herbert G. Strozinsky, Burlington Northern Railroad, USA. Change 10.302 OPC/A data records are created as RECFM=U but test data I VMACOPC received had been copied to RECFM=VBS, misleading me to Feb 23, 1993 assume that RECFM=VBS had to be specified in macro _SMFOPC defined in VMACOPC. The RECFM=VBS parameter is now removed from VMACOPC, so MXG will read either VBS or U data, using the RECFM in the data set label. Change 10.301 This new report macro replicates IBM's RMF reports from ANALRMFR MXG's TYPE70xx thru TYPE78xx datasets. In addition to Feb 22, 1993 producing reports, the source code can be read to find out which MXG variables are used for RMF report values. This support includes all RMF reports except Channel and Trace, (which will be added when there is a groundswell of user requests, or when we feel like it, whichever comes first!) The report syntax is self-defined in this member. Change 10.300 Support for Empact's HiperCache product's SMF record adds EXHIPRSA two new dataset: EXHIPRVS HIPRSAM - Hipercache SAM data set activity IMACHIPR HIPRVSAM- Hipercache VSAM data set activity TYPEHIPR thanks for this significant user contribution. VMACHIPR
Feb 22, 1993 Thanks to David Childress, Lowe's Companies, Inc, USA. Change 10.299 RMFINTRV enhancements have added SIO75CNT (Paging SSCHs), RMFINTRV SIO74TAP (SIOs to tape devices), and TSO2-TSO4 variables Feb 22, 1993 SWAP,TRAN,RESP for 2nd thru 4th period TSO counts and response time. In addition, these swap counts: EXTAUX LOGAUX LOGEXT PHYAUX PHYEXT and page movement rates: HIPMIGRS HIPREADS HIPWRITS PGMIEXAU PGMVTOEX PWSSMIIN were added to RMFINTRV from the TYPE71 dataset. Thanks to Tom Parker, Hogan Systems, USA. Thanks to Chuck Hopf, Primerica, USA. Change 10.298 PDB.PRINT dataset now contains new variable TOTLINES, the BUILDPDB sum of PRINTLNE, PUNCHCRD, EXTWTRLN, for sites that still BUILDPD3 have impact printers. Note that TOTLINES is the number of Feb 22, 1993 print line images if JES2 or PSF prints line-mode data, but if PSF prints PSF-mode data, it counts print records, and a print record can be a single print line or it could be a multi-page graph. Because of this, variable SHEETPRN the actual number of sheets of paper printed, should be used, with TOTLINES only used for impact printers under JES. See member ACHAP31 for discussion of print charging. Change 10.297 Support for optional HOGAN application variables was IMACICUS updated to add new variables FPSSCRN, TCTPLD, TCBFUNC, VMAC110 PLDVERS, and correct variables FPSOPTN and FPSTYPCD, and Feb 22, 1993 the new variables were added to CICSTRAN KEEP= list (but, like all the optional variables in CICSTRAN, they will not exist unless the optional code is enabled in IMACICUS). Thanks to Tom Parker, Hogan Systems, USA.
ANALDB2R and PDBOUT= was not specified, ERROR "LIBREF SMF IS NOT IN Feb 21, 1993 A VALID FORMAT" resulted. ANALDBTR should have replaced the earlier "paring" macros, which were the source of the real error. Replacing the "paring" with %ANALDBTR invoked for those pairs eliminated the error. Specifying PDBOUT= circumvented the error. The PMACC02=YES accounting detail in the accounting detail report and the values reported as averages in the Buffer Pool Summary were reported incorrectly. In addition, a 'Total Average' column was added to be consistent with DB2PM.
ASUMDB2A were created but not kept because they were not in the Feb 21, 1993 SUM= list.
ADOCHPCS read MPE data (HP's older, proprietary operating system). EXHPCSSP Some minor logic bugs were also corrected. Macro HPCSDLM IMACCSSP was added to member IMACHPCS to identify the character(s) TYPEHPCS that are to be used as field delimiters when the PCS data VMACHPCS was created, since the delimiter is a part of the data Feb 21, 1993 collection command. Member ADOCHPCS now contains the specific SCOPE commands that was used to create the data that MXG TYPEHPCS support expects. In addition, that same ADOCHPCS member contains Chuck Hopf's fine paper on UNIX resource measurement, "An RMF Bigots View of Measuring UNIX Systems, or how i learned to type in lower case". Thanks to Doug McBride, Hewlett-Packard, USA. for patience with the untutored in this particular world of DP, and for the data and manuals that made this enhancement to MXG possible.
EXCMFCAR expanded to support essentially all subtypes; the complete EXCMFCCS status for each subtype is listed in comments at the start EXCMFCDS of member VMACCMF. Most of this enhancement is based on EXCMFCJS the sample SAS code distributed with CMF, so that names of EXCMFCOS variables and datasets are the same, but their sample code EXCMFCPQ was architecturally revised and edited into the MXG style. EXCMFCSC With this enhancement, these datasets are now created, EXCMFCSS and all have been tested with real data records, too! EXCMFC93
EXCMFDDS Subtype Dataset Description (Dataset Label) EXCMFGDA 00 CMFDEVIC CMF 00 DEV EXCMFJDS 00 CMFDOM CMF 00 DOM EXCMFPDS 00 CMFIPS CMF 00 IPS EXCMFPGS 00 CMFOBJ CMF 00 OBJ EXCMFPDS 00 CMFPG CMF 00 PG EXCMFUSS 00 CMFSRMC CMF 00 SRM EXCMFWOR 00 CMFEXTCC CMF 00 TCC VMACCMF 00 CMFEXTPG CMF 00 TPG Feb 21, 1993 00 CMFEXTRT CMF 00 TRT 01 CMFCPUQ CPU QUEUE COUNTS 01 CMFCPUS CPU AND CHANNEL SAMPLE DATA 02 CMF02PSD ASM DATA (PAGE/SWAP DATASETS) 03 CMF03PGS PAGING DATA (SRM SAMPING) 04 CMF04WOR WORKLOAD DATA (DOMAIN SAMPLING) 05 CMF05DDS DASD DEVICE DATA 05 CMF05TDS TAPE DEVICE DATA 06 CMF06GDA JES SPOOL ACTIVITY 06 CMF06JDS JES CLASS ACTIVITY 09 CMFASMQ ASM DATA 19 CMFTRACE I/O WORKLOAD TRACE 20 CMF20CCS TSO COMMAND SUMMARY 20 CMF20CSS TSO COMMAND DETAILS 21 CMF21USS TSO USER RESOURCES 23 CMFPG0V CMF 23 PG0V 23 CMFPRIOS CMF 23 PRIOS 27 CMF27C93 CACHE SAMPLING MODEL 3990-3 27 CMF27CAR CACHE SAMPLING MODEL 3880 27 CMF27CSD CACHE SAMPLING CSD 29 CMF29COS COMMON STORAGE COS 29 CMF29CJS COMMON STORAGE CJS 29 CMF29CDS COMMON STORAGE CDS Change 10.292 Support for IMF Release 2.8 uncovered MXG corrections: VMACCIMS Dataset CIMSTRAN variable INPQUETM was wrong for fastpath, Feb 20, 1993 and COREALOC is actually "Storage Available", so its label was changed. The IMF Release itself added these changes: Dataset CIMSTRAN: New variable UOWTIME is now created from existing variable ALPCBTRN. For DBCTL, ALPCBTRN is the IMS Transaction's Unit of Recovery, the same as the CICS Unit of Work ID, UOWID. MXG converts the first six bytes of ALPCBTRN into UOWTIME, the 6-byte part of UOWID that is the constant token across all records for the same transaction, and the token that is used to match IMS events to their CICS and/or DB2 counterparts. UOWTIME is always created, but is only valid when ALPCBTRS contains a UOR. The second field CICS/DB2 field used to match transaction, NETSNAME, the name of the originating network on which UOWTIME was created, is not yet in the IMF transaction record. Variable FLGSPECL now contains Q for Quick Reschedule. New variables TRNSKW and TRNSNKW count the SYNC buffer flush writes (Key-Writes and NonKey-Writes). New variable APPCIMS flags APPC/IMS transactions. Dataset CIMSDBDS: Two new counts: DBTDDNNO='DDN*DBTNO' DBTSECNO='SEC*DBTNO*(OVERFLOW)' and seven new flag (status) variables: DBTMSCNT='DBT CNTRS*OVERFLOWED?' DBTOSSB ='OSAM*SB IN*EFFECT?' DBTOVFLW='DBT*OVERFLOW*OCCURRED?' DBTTYDDN='I AM A*DDNAME*DBT?' DBTTYOTH='I AM A*CATCH-ALL*DBT?' DBTTYSEC='I AM A*SECONDARY*DBT?' DBTVSAM ='VSAM*ACCESS*METHOD?' Fortunately for everyone, the new fields used existing reserved fields, so the changes were compatible. Change 10.291 Support for Oracle 6.0.33.1.51 added two READTIME and JOB VMACORAC and ACCOUNTn fields, and two additional CPU time measures. Feb 20, 1993 CPUCICTM is the CICS subtask CPU time (now added to CPUTM as it was not previously captured). CPUXMETM is the CPU time spent in cross memory mode in the ORACLE address space, but as this is already captured in CPUTCBTM or CPUSRBTM, it is not added to CPUTM; CPUXMETM lets you see how much of CPUTM was spent in the ORACLE ASID. There is a new section for SQLNET data, but the contributor site did not have documentation nor data for that option. The USER field section is added; see comments in VMACORAC on how to keep the USER field if it exists in your data. Thanks to Martyn A. Jones, Data Sciences UK
TYPESLRI processing - it keeps only a subset of the statistics that Feb 19, 1993 MXG's new ASMIMSLG produces and it uses the output of IBMs DBFULTA0 as input for Fast Past processing, which greatly reduced the elapsed processing time. This contribution is documented in the author's letter at the beginning of the member, which is actually the JCL to build a PDS with the several members of this IMS log processing alternative. Thanks to George Denissof, Savings Bank Services Ltd, Espoo, FINLAND. Change 10.289 To bill tape datasets using TMS dataset DSNBRECD is built TYPETMS5 for the first dataset on the volume by OUTPUTing the info VMACTMS5 from the volume record, but if the first dataset on the Feb 19, 1993 volume was SCRatched or DELeted in the catalog, MXG still output to DSNBRECD, causing a billing charge for a "dead" tape. Now, the observation is OUTPUT only if the dataset is neither scratch nor deleted. Thanks to Chuck Hopf, Primerica, USA.
TYPEIMSA had not been moved into TYPEIMSA from TYPEIMS. Lines in Feb 19, 1993 TYPEIMSA from STRTTIME=GUTIME; thru OUTQUETM=OGUTIME... were deleted and replaced by TYPEIMS lines from /* TEST FOR MISSING thru RESPNSTM=ENDTIME-ARRVTIME;. Thanks to Lonnie T. Rimmer, Philip Morris, USA. Change 10.287 Format $MGAMDDT is no longer used (it was for an earlier FORMATS version of Amdahl's SPMS), and has been removed. Feb 19, 1993 Thanks to Adrian Reynolds, SAS Institute, ENGLAND. Change 10.286 MXG 10.5 only. Variable AVGMTPTM, the average mount time VMAC74 per TAPEMNTS (new in ESA 4.3) was incorrect. DEVMNTTM, Feb 19, 1993 the duration when a mount was pending is added to TYPE74.
EXTY94 SMF type 94 record is added, providing hourly statistics IMAC94 (min/max/avg counts/durations) of drives mounted, mount TYPE94 requests pending, demounts, ejects, audit requests, and VMAC94 the number of insert stores. Feb 18, 1993 Change 10.284 Amdahl MDF creates TYPE70PR observations with LPARNUM=0, ASUM70PR while the first LPAR in IBM's PRSM or Hitachi MLPF is=1. Feb 18, 1993 ASUM70PR will produce an error message "MORE THAN 16 LPAR" when it encountered the unexpected LPARNUM=0. Now, this summary member supports 17 LPARs, numbered 0 thru 16! Thanks to Jeff McFadyen, Ministry of Correctional Services, CANADA. Change 10.283 Some fields (notably R791TCPU) are accumulated in type 79 VMAC79 records created by RMF Monitor II. However, it appears it Feb 17, 1993 is not possible to deaccumulate the data. Using the sort order BY SYSTEM R79SES R791ASID R791JBN STARTIME does correctly sequence the TYPE791 dataset, but MVS/ESA 4.2.2 data contains adjacent intervals in which R791TCPU drops rather than increases. I assume these are either step transitions or reuse of the same ASID by the same Job in which the CPU clock is reset, but there is no indicator of STEPNR or JESNR in the record by which the reset can be recognized. Furthermore, even when two intervals happen to show an increase, it is possible that the CPU clock was reset, but the CPU time in the second interval just happened to be larger; in this case, if the values were deaccumulated, the wrong CPU time would have been created. In addition, the TYPE791 data contains accumulated counts for page activity when the task is swapped in, but those counts are zeros in the next record if the task is swapped out, requiring additional investigations in the absence of logic description from IBM as to when this occurs. Thus a safe algorithm for deaccumulation is not possible until IBM enhances the record and describes when fields are zeroed and when they are not! Thanks to Khalid Al-Harthi, Saudi Arabian Airlines, SAUDI ARABIA. Change 10.282 This suite of RACF reports was made more user friendly by ANALRACF removing references to &DAYNUM, and providing specific Feb 17, 1993 example of how to set &START and &END in JCLRACFR. All occurrences of RACFTERM $6. were changed to $8. MXG Format MG080TY was updated with new types. PROC DATASETS were added to delete unused work data sets, freeing DASD. Two new members now exist: JCLRACFR executes ANALRACF programs, and JCLPRINT prints reports created by ANALRACF. The internal members in ANALRACF that were altered are: JCLRACFR JCLPRINT MXGDAY1 REPRACF REPRACFB REPRACF2 REPRACF3 REPRACF4 WPDBRACF While ANALRACF provides a lot of reports, the new TYPE80A dataset may prove to be a better source of reporting. Much of the logic in ANALRACF was required because MXG split the type 80 record into multiple observations. The TYPE80A member creates one observation per type 80 into several datasets for improved access and reporting. Try it first! Thanks to George Waselus, State of Arizona, Department of Admin, USA. Change 10.281 DB2 Trace IFCID 0044 Lock fields were input incorrectly. VMAC102 Change INPUT QW0044KL PIB1. to read Feb 17, 1993 to read INPUT @I+8 QW0044KL PIB1. Thanks to Jason Lau, AMP Society, AUSTRALIA. Change 10.280 The 4th TCB (Secondary LU) CPU time was not summed into VMAC110 CPUTCBTM in dataset CICDS, and thus it was also not in the Feb 15, 1993 CICEODRV, CICINTRV, CICREQRV or CICUSSRV datasets.
VMAC535 created, and its type 5 record caused MXG to fail because Feb 4, 1993 I assumed the MVS/370 service unit fields were present in the record. Now their input is conditional! Thanks to Don Mosley, Farmland Industries, USA.
IMACICOC to their BSC segment in the type 110 record for both CICS IMACPTF Version 2 and CICS/ESA. This will corrupt the data in the Feb 4, 1993 other OMEGAMON segments, as well as any user data segments your installation might have added. And, of course, there is no way to detect whether or not you have installed this PTF by examining the type 110 record! As a result, you will need to update member IMACPTF to enable _QOC0109 to tell MXG when you have installed this OMEGAMON PTF. (Page 9 of Newsletter TWENTY-TWO discussed how to enable the MXG Support for OMEGAMON V550.) Thanks to Carol Harper, EG&G Idaho, USA. Change 10.277 DCOLLECT does not capture VSAM space used. IBM Appendix E VMACDCOL of the MVS/DFP Access Method Services For ICF documents Feb 4, 1993 that DCDUSESP is not 'valid' for VSAM datasets. MXG's ASMVVDS/TYPEVVDS processing, which reads the VVDSs, does capture space allocate and space used. However, under user pressure, IBM has announced, in ETR Item E3058,379 that DFSMS Release 1.1 (March, 1993) will add a field to the "A" record with space used value. Note: Change 10.305 added the new fields to "A" record, which is output to DCOLCLUS dataset; see ADOCDCOL. Thanks to Frank Vessell, ITT Consumer Financial Corporation, USA.
VMACX37 variables to the STOPX37 dataset. This support has not Feb 4, 1993 yet been tested with actual 3.4 data records.
VMAC73 should be SKIP=SKIP-8; instead of SKIP=LENCHDS-8;. (Note Feb 4, 1993 that line 030800 still must be SKIP=LENCHDS-8;). Thanks to Jim Nissen, Principal Financial Group, USA. Change 10.274 Support for Empact's POOL-DASD user SMF record, written EXTYPOOL whenever POOL-DASD manages an allocation. The text of the IMACPOOL WTO messages associated with the allocation override by TYPEPOOL POOL-DASD describes what action was taken. VMACPOOL Feb 3, 1993 Change 10.273 Support for Candle's IMS Transaction Reporting Facility, EXITRDAT ITRF, is added. MXG reads the output file created by the EXITRDB2 ITRF Batch Summary program to create these MXG datasets: EXITRMSG ITRFMSG - Message records EXITRMSO ITRFMSGO - Message Out records EXITRSUM ITRFDATA - Database Records (DL1 and Fastpath) IMACITRF ITRFSUM - Summary Records (DL1 and Fastpath) TYPEITRF ITRFDB2 - Summary Records (DB2) VMACITRF ITRF is a new part of Omegamon II for IMS. Its online Feb 3, 1993 component creates an IMS log record (default ID=160) that captures response time, CPU time, virtual storage used, and elapsed time of DL1 and DB2 calls. That log record, along with IBM log records 01,03,07,08,31,32,33,35,5607, 5901,5903,5936,and 5937 are the input to the ITRF Summary program, which constructs transaction events and creates the output file that is the input for MXG's TYPEITRF program. Although Candle provides an option to send all of those required IMS log records to SMF for processing by the Summary program, this is one instance in which I think it ill advised to use SMF because of the significant volume of data. (If Candle ever builds a transaction data in the online component and write a single record for an IMS transaction, I would strongly endorsed SMF as the correct destination for such a record!). Change 10.272 DB2 Accounting report 2nd occurrence "GET PAGE REQUEST" ANALDB2R is actually "BUFFERPOOL EXPANSIONS". The Average values Feb 3, 1993 of Buffer Pool statistics are incorrect; they were divided by the number of occurrences, then divided a second time. (We missed this careless error because in our test data we happened to have only one occurrence for each group!) Finally, the "Total" column will be relabeled to indicate it is the "Total of Averages", and a new column with the true total has been added to the report. Thanks to Tuomo Rahko, KOP / Kansallistieto, FINLAND. Change 10.271 Processing Landmark's Monitor For CICS data Release 9 may TYPEMON8 produce "INVALID DATA FOR TIESDATE" and "ERROR"INVALID DO Feb 3, 1993 LOOP CONTROL INFORMATION", if the Landmark Dump program does not specify CONVERT on the DUMP statement. (See their JCL samples TMON9DBU and TMON9FSU). Without the CONVERT operand, the dumped data is not converted to Release 8 format, which is the only data format that Landmark has documented externally. As long as DUMP CONVERT is used, MXG has no problem processing Release 9 data. MXG now detects this condition and writes an error message to the SAS log alerting you to the incorrect dumping. Landmark has also begun shipping of Release 1.0, which has the same data records as Release 9 (and which also requires the use of the CONVERT option). Thanks to Terry Baker, Royal Insurance, USA.
READDB2 %READDB2(INFILE=SMF,PDBOUT=XXX,IFCIDS=ACCOUNT STATISICS); Feb 3, 1993 did not write DB2ACCT,DB2STAT0,DB2STAT1 to the XXX ddname, because the defaults defined in IMACDB2 were not changed by READDB2. The macro definitions _DB2ACCT and _DB2STAT were replaced by new definitions of _LDB2ACC, _LDB2ST0, and _LDB2ST1 to correct this error.
READDB2 FOUND" if the USER=xx option is specified, because READDB2 Feb 3, 1993 expected the selected datasets to be in the WORK file. Thanks to Mike Marek, Kraft General Foods, USA. Change 10.268 10.4-10.5 only. INPUT STATEMENT EXCEEDED RECORD LENGTH VMAC28 resulted because variable LSAMINID was input twice! The Feb 3, 1993 second occurrence in the INPUT statement must be deleted. Thanks to Adrienne Wijlaars, Belastingdienst Automatiser, NETHERLANDS Change 10.267 Title statements in these graphic members still contained GRAFBNCH the options .F and .H of an ancient SAS version (82.4). GRAFCICS The period should have been removed from both options long GRAFVM ago! Feb 2, 1993 Thanks to Norbert Korsche, OMV-AG, AUSTRIA. Change 10.266 DB2 Trace subtype 191 (data set T102S191) is now decoded, VMAC102 although some of the internal diagnostic sections are not, Jan 28, 1993 nor is this strange subtype protected if the data length exceeds 5000 bytes. (DB2 self-limits a DB2 data block to 5000 bytes, even though the SMF record can be 32760 bytes. When an IFCIC=191 data block exceeds 5000 bytes, DB2 inserts two-byte length fields between the blocks to build the SMF record, but DB2 does not update the internal offsets inside the data blocks! This will require lots of nasty logic to either strip the length fields or to figure out how many length fields there are and then to correct the internal offsets to the external record locations! If someone really needs the diagnostic sections decoded and provides sample data, only then will I have to spend the time to figure out how to do it!) Thanks to Tom Parker, Hogan Systems, USA. =Changes thru 10.265 were included in MXG PreRelease 10.5, Jan 28, 1993= Change 10.265 Support for CA's IDMS Release 12 Performance Monitor SMF VMACIDMS records is confirmed. It was easy, as there were no Jan 28, 1993 changes in their SMF records! Thanks to Roger Clayton, Cryovac Division of W.R. Grace & Co, USA. Change 10.264 Support for NPM Release 1.6 added new values for several EX028CM7 formats and changed the FRP, NAM, NCD, and TRI segments. EX028LA6 New subtypes 84x,85x are output in the existing NPMINFRP EX028LA7 dataset. Six new data sets are created: EX028LA8 Subtype Dataset EX028LA9 80x NPMLANET Ethernet Lan EX028LAA A0x-A1x NPMLANOD ODLC Lan FORMATS C3x NPMLANSG Lan Segment Utilization VMAC028 C0x-C2x NPMCMLAN Start/Stop/Alter Lan Jan 28, 1993 C4x-C5x NPMLANES Lan Monitor Exception C6x-C7x NPMLANSA Lan Attach/Detach
VMAC71 second didn't so indicate in their label but now they do! Jan 27, 1993 (You can always determine if a variable is a rate simply by looking at all occurrences of the variable name in the MXG source VMAC.... member to see if it is divided by the interval duration. Thanks to Bill Fife, Computer Associates, USA.
EXNSPYLD NETSPY added STOPTIME to NSPYLU, NSPVTSWD to NSPYTR, EXNSPYLF NSPYLINE and NSPYNPSI datasets. Many new variables were IMACNSPY added to LANSPY dataset NSPYLANL, including frame and byte VMACNSPY counts for LLC, NetBios, IP, IPX, XNS, DEC, LAT, LAVC, ARP Jan 27, 1993 X.25, AppleTalk, NCP, SMB, VINES, NFS, FTP, ICMP, Telnet, Feb 20, 1993 Broadcast, Multicast, and SNA over Ethernet! Counts of frames and bytes were also added to NSPYLANS. New dataset NSPYLAND captures internetwork delays. Member EXNSPYLF controlled outputting of dataset NSPYLANF by testing IF SUM(a,b,c) THEN OUTPUT ..., which created observations only if the sum was non-zero, but two users were puzzled why it did not read IF SUM(a,b,c) GT 0 THEN OUTPUT..., so to avoid unnecessary puzzlement, it now uses GT 0! (Without the GT 0, the OUTPUT is executed if the sum is non-zero; since the a,b,c values were input as PIB4, they could never be negative, and there was no logic exposure, but it is more accurate and user friendly to use GT 0!) Feb 20. New Path Entry Layout for Extended Subtype D LAN SPY records was coded, and several incorrect lengths were revised. No data for testing yet available.
BUILDPDB Dataset Variable Description IMACPDB TYPE23 SYNCTIME - Datetime value when the SMF Global RMFINTRV TYPE30_V Recording Interval ended (the STATUS VMAC23 TYPE70xx function). This value is put into VMAC26J2 thru other SMF and RMF records so the data VMAC30 TYPE79xx can be matched exactly! However, the VMAC32 new SYNCTIME includes the GMT offset, VMAC33 so if you have a non-zero value for VMAC42 TIMEZONE= in SYS1.PARMLIB(CLOCKxx), VMAC60 SYNCTIME will be in GMT while all the VMAC6156 other timestamps will be in the LOCAL VMAC71 time zone! And, the GMT offset value VMAC73 is not provided in these records! VMAC76 You Synchronize SMF and RMF records VMAC77 with the SYNCVAL and INTVAL values Jan 28, 1993 in the SMFPRMxx member of parmlib. Feb 9, 1993 TYPE26J2 SUBMUSER - Submitting USERID NOTIFYND - Job end execution notify NODE NOTIFYUS - Job end execution notify USERID ACCOUNTn - Job card accounting fields finally!!! NRACCTFL - Number of accounting fields. LENACCTn - Length of nth accounting field DUPJBDLY - Flag that job was delayed due to Duplicate Job Name Hold event OFFLPURG - Flag that this purge record was due to Spool Offload event, and is not the actual purge event. (This flag requested only last year, is now used in BUILDPDB to solve the problem described in Change 9.243. TYPE30_V New INTBTIME and INTETIME values are added to the ID section, and INTETIME is now the end of the Interval (previously, INTETIME was taken from SMFTIME, which could be later than the actual end of the interval). If a task is swapped out at the end of the interval, a record is not written until the task is swapped back in, and SMFTIME is much later than INTETIME. The INTBTIME/INTETIME values are actually on the GMT clock, but because the old interval begin time (SMF30IST) remains on the local clock, the GMTOFFTM can be determined so that INTBTIME and INTETIME are converted to local time (like all the other type 30 times). Variable SYNCTIME is the GMT value of INTETIME that will match SYNCTIME in type 23 and type 70s, for matching synchronized SMF and RMF intervals. SMFTIME was also added to TYPE30_V data set. Data set TYPE30_V becomes the PDB.SMFINTRV data set. Also, DSSIZHWM was converted to bytes and format with MGBYTES format. TYPE33_1 Now contains JOB READTIME and STEPNAME. TYPE33_2 New APPC/MVS dataset contains resources at the conversation level, particularly needed if you uses an APPC/MVS Server address space to process multiple conversations concurrently, since now you can collect information for each conversation in the server address space. TYPE42SR New DFSMS data set provides response statistics for each Storage Class for each interval: AVGCONMS- Average connect time per SSCH AVGCUQMS- Average CU queue time per SSCH AVGDISMS- Average disconnect time per SSCH AVGPNDMS- Average pending time per SSCH CACHCAND- Count of cache candidates CACHHITS- Count of cache hits IOCOUNTS- Total i/o count RESPTIME- Average response time per SSCH STORCLAS- Average connect time per SSCH WRITCAND- Count of write candidates WRITHITS- Count of write hits TYPE42DS New DFSMS data set provides response statistics for each Data Set, with same statistics as in the TYPE42SR (above), with additional details. TYPE60 Existing field SMF60SUB/SMF61SUB which explained TYPE6156 if the VVR or ICF Catalog record was UPDATED or INSERTED or DELETED is now "Erroneous Field. Do Not Use", with no alternative field! TYPE70s Variable SYNCTIME was added to all RMF datasets for synchronization with SMF records. See above. Additionally, all RMF datasets have new variable INTRVSYN=Y if synchronization is in effect. TYPE71 CPUPAGTM, the CPU time spent in page movement in central store. When a particular type of frame (eg. below 16MB or nonreconfigurable) is mandated by a request but is not found in the available frames, the real storage manager uses a process called prefsteal to attempt to find a correct type of frame and move the contents of that frame elsewhere in central or expanded storage. The TCB/SRB timer is stopped during this process in consideration of the general desire that these times be as repeatable as possible. This new variable, CPUPAGTM, is the CPU time that was not previously captured during this process. TYPE72MN A significant MVS/ESA 4.3 RMF enhancement is the addition of many new RMF III variables to the TYPE72MN dataset, written for each performance group for each interval. New sampled measures for the percentage of time when each performance group was USING devices or CPU, or was DELAYed for devices, CPU, storage, ENQ, HSM, JES, MOUNT, message, XCF, will make TYPE72MN a very useful source of delay analysis. Additional measures of CSTORE and ESTORE usage and "long" logical swaps are included in these new variables: AVGELPTM=AVG ELAPSED*TIME PER*ENDED TRANS AVGQUETM=AVG JES/APPC*QUEUE TIME*PER ENDED CPUVECTM=VECTOR*CPU USED*DURATION LOGCSTOR=CSTORE FOR*LOGICALLY*SWAPPED USERS LOGESTOR=ESTORE FOR*LOGICALLY*SWAPPED USERS LONGESTO=LONG SWAPS*TO EXPANDED*STORAGE LONGLGSW=LONG*LOGICAL*SWAPS LONGPHYS=LONG SWAPS*TO PHYSICAL*AUXSTORE LSWSAMPS=TOTAL*LOGICAL SWAP*SAMPLES*/ PCTDLDEV=PCT WHEN*DEVICE*DELAY PCTDLENQ=PCT WHEN*ENQ*DELAY PCTDLHSM=PCT WHEN*HSM*DELAY PCTDLJES=PCT WHEN*JES*DELAY PCTDLMNT=PCT WHEN*MOUNT*DELAY PCTDLMSG=PCT WHEN*MESSAGE*DELAY PCTDLPRO=PCT WHEN*PROCESSOR*DELAY PCTDLSTO=PCT WHEN*STORAGE*DELAY PCTDLXCF=PCT WHEN*XCF*DELAY PCTUNKN =PCT WHEN*UNKNOWN*STATE PCTUSDEV=PCT WHEN*DEVICE*USING PCTUSPRO=PCT WHEN*PROCESSOR*USING PHYESTOR=ESTORE FOR*NON-LOGICAL*SWAPPED USERS PSWSAMPS=TOTAT*NON-LOGICAL*SWAP SAMPLES TRANS =ENDED*TRANSACTION*COUNT VALDSAMP=TOTAL*VALID*SAMPLES TYPE73 In MVS/ESA 4.2, APAR OY45991 PTF UY77343 now writes a CHPID segment for all 256 possible paths whether they exist or not; now MXG only OUTPUTs TYPE73 observations if the CHPID is ONLINE. TYPE74TD New "Tape Device" dataset contains the maximum number of tape devices allocated each interval, recorded if device class TAPE is being recorded. This dataset was also added to BUILDPDB/BUILDPD3 and WEEKBLD/MONTHBLD logic. TYPE74 New variable, TAPEMNTS, counts the number of tape mounts detected during the interval for each tape device. Since TAPEMNTS now exists, the duration of Mount Pending, DEVMTPTM is also now created. If a Mount is Pending (MTP) at the beginning or the ending of the interval, variables MTPATBET and MTPATEND are "Y". Only if MTP does not exist at either the begin or end will MXG calculate the average mount pending per tape mount per device, new variable AVGMTPTM. (Only when both flags are blank do we know for sure that the mount pending time duration and the number of mounts are exactly matched.) TYPE76 SAMPSKPD variable if samples were skipped TYPE77 SAMPSKPD variable if samples were skipped, and WAITS was PIB2., is now moved to a PIB4. field at the end of the record. TYPE792 R792EXCP was expanded to four bytes and moved to the end of the record. TYPE79C New variables R79FLAG, R79CPBY, R79CCPTS added. TYPE90 Variables MINBUFF and MAXBUFF are now reserved in IPL SMF, SET SMF, or SETSMF Operator Command event records. No real loss here, since the maximum number of buffers each interval is always in TYPE23 dataset. TYPE96 The new type 96 record was already supported in MXG 10.3, as TYPETIRS, for this record is created to account for "The Integrated Reasoning System" users resource usage. However, the name in the SMF manual for type 96 is "Cross Memory Service Provider Charge Back", is inaccurate; it's TIRS! BUILDPDB Logic was changed to use the type 26 OFFLPURG field to detect purge records created by the SPOOL Transfer/Offload program. New variables SUBMUSER DUPJBDLY OFFLPURG ACCOUNT1-ACCOUNT3 LENACCT1-LENACCT3 and NRACCTFL were added to the list of variables kept from TYPE26J2 (in member IMACPDB). The order of datasets in the MERGE for PDB.JOBS was changed, moving GOOD26J2 to be first so that the TYPE26 ACCOUNTn fields will be used in PDB.JOBS when they exist. (Leaving it where it was could have blanked out valid ACCOUNTn data since the SAS MERGE uses the values from the last dataset, for sites not yet on MVS/ESA 4.3!) RMFINTRV The new TYPE71 CPUPAGTM is added to RMFINTRV as a separate variable, and is also included in CPUTM, the sum of all captured CPU time; this will cause and increase in the capture ration of RMF. Change 10.260 Negative CPUACTTM and PCTCPUBY in TYPE70 under PR/SM can VMAC7072 occur if the LPARNUM for the PHYSICAL partition segment is Jan 24, 1993 non-zero and equal to a real LPARNUM value. The LPARNUM for the PHYSICAL partition should always be zero, but at a few sites it was erratically non-zero (which also caused RMF reports to fail!). IBM APAR OY55704 and PTF UY87723 acknowledge and fix the problem, but MXG now protects the data by inserting: IF LPARNAME='PHYSICAL' THEN LPARNUM=0; following the @; which follows the INPUT of LPARNUM. Thanks to Tom Hatcher, US Sprint Dallas, USA. Thanks to MP Welch, US Sprint Dallas, USA. Change 10.259 There are now always 255 CHPID segments in type 73 record VMAC73 even though there might not be 'FF'X CHPIDs defined in the Jan 24, 1993 system. Now, MXG only outputs TYPE73 observations if the CHPID is defined. APAR II06736 discusses this problem and suggested two possible algorithms; the above is the second algorithm in that APAR (beware - there are typo's in the 1st algorithm - Reference to SMF73FGS should be SMF73FG3, and the bit field names SMF73REP and SMF73VLD do not exist in the SMF manual - due to OCO!). Thanks to John Mattson, National Medical Enterprises, USA. Change 10.258 Variable QW0023DN was not kept in dataset T102S023, which VMAC102 caused a warning when using ANALDB2R report PMTRN01. It Jan 23, 1993 is reserved, and should not have been referenced in the report, but the better fix was to now keep it! Thanks to Andre Henry, AG 1824, BELGIUM.
VMAC7072 each section actually exists in the physical record. The VMAC71 previous validation only checked that the section offset VMAC73 and number of sections was non-zero, but did not protect VMAC74 for truncated records nor for unexpected invalid records. VMAC75 Now, if a bad record is encountered, a message will be VMAC76 printed on the SAS log, the bad record deleted, but the VMAC77 processing will continue, avoiding the STOPOVER abend! VMAC78 VMAC79
Jan 23, 1993 Thanks to Phil Hathaway, Spartan Stores, USA. Change 10.256 SMF type 70 INPUT STATEMENT EXCEEDED RECORD LENGTH occurs IMACTCP with IBM's TCP/IP SMF record because the IBM sample member Jan 22, 1993 used record ID of 70! IBM APAR PN34837 addresses this bad choice, and IBM has now assigned SMF record numbers 118 and 119 to the TCP/IP product SMF records. MXG has thus changed its default values in IMACTCP to agree with this IBM correction. To delete the unwanted TCP/IP records, you can add this code to member IMACFILE: IF ID=70 AND LENGTH LE 194 THEN DELETE; until you get the ID of TCP/IP records corrected. Change 10.255 ICF Catalog cells C4x and C9x were not printed because VMAC6156 the test prior to their PUT statements tested for 'C4' or Jan 14, 1993 'C9' (without the X indicator to test for hex value!). New Feb 17, 1993 variables DATANAME/DATAVOL1 and INDXNAME/INDXVOL1 for VSAM entries contain the VSAM Data Component name, First Data VOLSER, Index Component name, and First Index VOLSER. New variables DATACLAS, MGMTCLAS,and STORCLAS are non-blank if the Optional SMS Subcell for a Cluster record is found. Thanks to Steve Dzielak, First Interstate Bank of Arizona, USA.
VMXGHSM HSM BCD/MCDS records were wrong. Replace these lines: Jan 14, 1993 TTOCDLR PD3. +1 TTOCEFLG PIB2. TTOCXPDT PD3. @; and the subsequent lines that reference TTOCDLR/TTOCXPDT with these lines: TTOCDLR1 PIB1. TTOCDLR2 PIB2. +1 TTOCEFLG PIB1. TTOCXPD1 PIB1. TTOCXPD2 PIB2. @; TTOCDLR =TTOCDLR1*1000+TTOCDLR2; TTOCXPDT=TTOCXPD1*1000+TTOCXPD2; Unrelated, the line formatting TTOCTYP $HEX2. should be deleted, as this is an EBCDIC field. Thanks to Chris Fenn, Andersen Consulting, ENGLAND. Thanks to Chris Weston, SAS Institute Europe, GERMANY. Thanks to John Dalton, SAS Institute, ENGLAND. Change 10.253 QW0022OS, the DB2 optimizer's cost estimate, was supposed VMAC102 to have been changed by Change 10.174, but the code was Jan 14, 1993 not changed until now. Sorry for this carelessness! Thanks to Matthew Bildstone, Sun Life, ENGLAND. Thanks to Chris Weston, SAS Institute Europe, GERMANY.
GRAFLPAR should have been "by &incrd", as incrd is a macro variable Jan 14, 1993 and was missing the ampersand. Thanks to Chris Weston, SAS Institute Europe, GERMANY. =Changes thru 10.251 were included in MXG PreRelease 10.4, Jan 8, 1993== Change 10.251 Preliminary revised support for RACF & TOPSECRET type 80 EXTY8001 SMF record. The original TYPE80 data set built by MXG is EXTY8002 valid, but it is unwieldy for analysis, as it creates one EXTY8003 observation for each data segment in each type 80 record, EXTY8004 and left the decoding of the RACFDATA string to you! This EXTY8005 revision recognizes that each type 80 record represents an EXTY8006 event, and the data segments for each event should be kept EXTY8007 in a single observation. Instead of one TYPE80 dataset, EXTY80A the new logic creates nine datasets; seven for each RACF EXTY80CM event category, (TYPE8001-TYPE8007), TYPE80CM for RACF IMAC80A Commands, and TYPE80A in case records with no segments are TYPE80A encountered. The data segments in the TYPE80CM Command VMAC80A dataset are not decoded in this implementation, but the Jan 7, 1993 command and its issuer, etc., are decoded. Some variables may be formatted, and some revisions are likely, but the initial testing suggests that reporting and selection of RACF data will be much easier with this revision. The 7 event categories and their new dataset names are: TYPE8001 Job Initiation or TSO Logon TYPE8002 Resource Access TYPE8003 End of Volume TYPE8004 Rename Data Set TYPE8005 Scratch Dataset or Tape Volume TYPE8006 Delete Volume one of Multi-Volume TYPE8007 Define Data Set or Tape Volume TYPE80CM RACF Commands Issued TYPE80A Catchall for records with no data segments Thanks to Joseph J. Faska, Depository Trust, USA Change 10.250 Twenty-five more ADOC members (Documentation members, one ADOCx for each "Product" which described all data sets that MXG Jan 7, 1993 creates from that product data records) have been added. Most contain sample PROC PRINTs, but not all are fully completed. However, when an ADOC member exists, it is the first place to look for notes and contents of MXG datasets Each ADOC member corresponds to a Section of Chapter 40.
FORMATS were updated for new values for RACF 1.9. Jan 7, 1993
VMXGSUM the same as the SUM function, but stores the results in an Jan 7, 1993 8-byte field instead of a 4-byte field. This is rarely a problem with performance/capacity data, but in chargeback applications with large dollar amounts, pennies could be lost with only 4-byte stored variable lengths. Thanks to Chuck Hopf, Primerica, USA.
VMAC73 4.2.2 added 8 bytes to each CHPID section in type 73 SMF Jan 6, 1993 record, but MXG failed to skip over the new fields, which caused alternating good and invalid observations, and MXG had good records only for the first 128 CHPIDs! As best as I can discover, there was no TNL to the SMF Manual for this IBM change! The correction is to insert xx lines as shown in the following corrected code (note this is the last PCHANBY input, at the bottom of VMAC73);
exists @; insert SKIP=LENCHDS-8; " IF SKIP GT 0 THEN DO; " INPUT SMF73PBY PIB4.6 /*CHANPATH*BUSY TIME*SINCE LAST*/ " SMF73PTI PIB4.6 /*CHANPATH*MEASUREMENT*INTERVAL*/ " @; " SMF73PBY=1024*SMF73PBY; /* CVT FROM 1024 USEC UNITS */ " SMF73PTI=1024*SMF73PTI; /* CVT FROM 1024 USEC UNITS */ " SKIP=SKIP-8; /*this line is now correct in this change */ /*and in VMAC73, but in 10.4 and 10.5 it */ /*was wrong (SKIP=LENCHDS-8;) in both! */ " IF SKIP GT 0 THEN INPUT +SKIP @; insert END; exists IF CHANIND ='..1.....'B THEN CHANTYPE='BLOCK MUX';
UTILGETM the record contains a subtype in bytes 19-20, but the VMACSMF subtype is an EBCDIC 00 (hex F0F0) instead of hex 0000, Jan 9, 1993 and UTILGETM threw the record away. Now, when an invalid subtype value is found, a message is printed on the log, and INDEXST=1 is set so that the record will be written. VMACSMF was also changed to input type 36's subtype as an EBCDIC number. This change was revised after MXG 10.4. An "ARRAY SUBSCRIPT OUT OF RANGE" can occur with MXG 10.4; the statement SUBTYPE=1; should have been INDEXST=1. Thanks to Freddie Arie, Lone Star Gas, TEXAS.
EXTYASXT an SMF record, but is sent to a flat file that must be IMACASXT first uncompressed by a Legent-provided program and then TYPEASXT decoded by these MXG members. See the comments in member VMACASXT VMACASXT, which includes the JCL example for the two-step Jan 5, 1993 process. (Note that the SMF record created by ASTEX has been supported for years by MXG members VMACDMON, named DMON in MXG because it used to be DASDMON). Thanks to John Rosza, Depository Trust Company, USA.
EXSYTCUM VXSYTCUM - LPAR Management time by Physical CPU EXSYTCPM VXSYTCPM - Channel Path Busy by CHPID EXPRCCFN VXPRCCFN - ADD access to CRYPTO facility EXPRCCFF VXPRCCFF - REMOVE access to CRYPTO facility EXIODALS VXIODALS - 3495 Automated Tape Library Statistics VMACVMXA (only partially decoded, awaiting 3495 manual Jan 4, 1993 because that's where the record is documented) Perhaps of more importance, VM/ESA 2.0 added a number of new variables to several datasets: VXAPLSRV - 25 new counters now exist VXIODDEV - new measure SCDATIM "Device Active-Only time" VXMTRDDR - IBM no longer spans across physical blocks! VXMTRDEV - RDEVSHAR if device is shared VXMTRPAG - FBA if device is FBA VXMTRPRP - PFXIDVER identifies CPU model version number in this and other CPU-related datasets. VXSYTUSR - SYSDIALD/SYSLUCNT for dialed/SNA connected users, and flag in VXUSEATE/VXUSELOF/VXUSEACT and VXUSELOF data sets. VXUSELOF - Major enhancement adds resources consumed to the existing (but previously useless) LOGOFF event record so that it now describes the resources used by each CMS machine session. Not only resources (CPU, SSCH, Reserved Pages) but now a session logon time CALTODON, RACF group VMCGRPN, and user accounting number (only 8-bytes!) VMDACTNO make this a much more attractive source of session-level resources for accounting and capacity planning. This description only hits the highlights. The complete list of new variables is given in member DOCVER10 for the data sets beginning with VX - twenty five MXG data sets have new variables! This code has been tested with real VM/ESA 2.0 data that was provided by IBM (with my sincere thanks!), but it did not contain any of the new record types, so not all of the code has been tested! An existing error in MXG 10.1 was also corrected; the second occurrence of "MACRO _MSTOASS ..." should have been ASI instead of ASS in three places in that line. This error caused a 180 syntax error. Change 10.243 As noted on page 10 of MXG NEWSLETTER TWENTY-TWO, SAS ZAP VMXGVTOC V6-SYS-FILE-4673 (on June 1992 Usage Notes Tape) is needed Dec 29, 1992 for SAS 6.07 to avoid "CRITICAL ERROR IN VTOC" messages.
DIFFTPX to incorrect parsing, but the error can be circumvented by Dec 29, 1992 re-ordering the PROC SORT to read: PROC SORT NODUP %VMXGFOR DATA= _LTPXINT ; Thanks to Mike Marek, Kraft General Foods, USA.
JCLMNTH concatenation, and the %INCLUDE SOURCLIB(MONTHBLD); which Dec 17, 1992 should have been the last line of SAS code was missing. Thanks to Barry Lampkin, Polaroid, USA. Change 10.240 SMFSMALL step JCLTEST6 may produce "ARRAY OUT OF RANGE" UTILGETM error with SMF record types of 127, 128 or 255, depending Dec 16, 1992 on subtype value, because the index calculation was wrong. INDEX=(ID+1)*256+SUBTYPE; must be INDEX=ID*256+SUBTYPE+1; and ELSE IF ID LE 128 ... must be ELSE IF ID LE 127 .... Thanks to Tom Gillis, Southern National Bank, USA. Change 10.239 ASTEC variable RDMT must be input as PIB4. vice PIB4.6, VMACDMON and the statement RDMT=RDMT*128; must be deleted, as the Dec 16, 1992 units are seconds, not 128 microseconds. Variables RDTBK1 thru RDTBK4 must be input as PIB4.2 vice PIB4. Thanks to Chris Nielsen, Wells Fargo Bank, USA. Change 10.238 VM/ESA 2.0 MONWRITE data causes SFM-OR-CRR SAMPLE message VMACVMXA and then PROBABLE DATA LOSS message if you have enabled Dec 15, 1992 the Application Server data. (The MXG protection for new APL data was incorrectly coded.) Change the line reading SKIP=SKIP-CALDATLN-8; to these two lines: INPUT +SKIP @; SKIP=0; With this change, MXG tolerates VM/ESA 2.0 MONWRITE data records without error, skipping over the new fields and records. MXG will exploit VM/ESA (i.e., decode new data) shortly; work is in progress.
IMACACCT now only printed five times, and the text was made more Dec 15, 1992 clear. Thanks to Ann Wheeler, American President Lines, USA. Change 10.236 All occurrences of PIBR. should have been PIB4. I thought VMACM204 this one had been fixed, but it slipped through into 10.3! Dec 15, 1992 This causes FORMAT PIBR unknown during JCLTEST6 execution. Thanks to Freddie Arie, Lone Star Gas, TEXAS. ==Changes thru 10.235 are included in Dec 13, 1992 MXG PreRelease 10.3== Change 10.235 Existing graphic reports were enhanced in GRAFLPAR when GRAFLPAR is executed under SAS 6.07 or later using PROC GPLOT with Dec 13, 1992 filling colors in place of PROC GCHART with bar charts, as we exploit SAS 6.07 features. Thanks to Chuck Hopf, Primerica, USA.
VMAC110 CICS records is enhanced by printing a table of expected Dec 13, 1992 values of Data Length and Number of Fields (MCTSSDRL and MCTSSDCN) in the MXG error message to help you compare the actual and expected values. This has been the number one technical problem in MXG CICS record processing this year sites CICS person installed a CICS monitor which excludes fields, but the MXG person doesn't find out until MXG runs against the modified data! The new error text should help resolve the error without additional phone calls! Change 10.233 Variables USER, SQLUSER & SQLDBMAC were added to VMSQLxx TYPEVM datasets. SQLDBMAC is the full name of the SQL database Dec 13, 1992 machine (and is stored in from USER after line 028200). SQLUSER is TERM in VMSQLTRM, is SYSTEM in VMSQLSYS, and is INIT in VMSQLINI, but it has meaning in VMSQLUSR where it is taken from the connect process and is sometimes a different user. USER is always SQL/DS in the -INI, -SYS, and VMSQLTRM datasets, and at present useless, but it is kept for consistency and possible future changes by IBM! Thanks to Norbert Korsche, OMV-AG, AUSTRIA. Change 10.232 APPC type 33 account record header offsets were off by 6 VMAC33 bytes (the subtype and length of header were omitted). The Dec 13, 1992 offsets for OFFPROD thru NRTPUS in lines 68-76 must have 6 added to each value. Additionally, the TP name field is variable length, with a maximum of 64 bytes, so it is now input conditionally: SMF33TPL PIB2. @; IF SMF33TPL GT 0 THEN INPUT TPNAME $VARYING64. SMF33TPL @; Thanks to Robin Luff, Dun & Bradstreet Europe, ENGLAND. Change 10.231 Support for Velocity Software's XAMAP history data files. EXXAMCPB MXG creates several datasets from the four history files. EXXAMCPT This code has been tested with actual XAMAP data records EXXAMDEV and reviewed for reasonableness, but there are many fields EXXAMSYS and lots of code! EXXAMUSR See comments at the beginning of VMACXAM for instructions IMACXAM for processing the four XAMAP files. TYPEXAM Jul, 2003: EXXAMSYS, EXXAMTCP, EXXAMUSR members were VMACXAM removed in MXG 21.03 as multiple datasets Jan 4, 1993 are now created from those XAM data. Change 10.230 Boole & Babbage CMF type 78 record variable R783PT-number VMACCMF of times channel path was taken- is in error in CMF 4.3.3 Dec 11, 1992 until you install Boole's fix BAB1081. Thanks to Bill Stoneberg, National-Oilwell, USA. Change 10.229 STC 4400 SMF record variable LSBECON1/2 documentation was VMACSTC incorrect; these are not connect durations but rather are Dec 11, 1992 the LSM number and Panel number, and they are six bytes instead of four bytes! They are now input $CHAR6., format $HEX12., and were removed from the IF test for OUTPUT of STCLMU dataset. Also, variable LSMNUMBR is created. Thanks to Dean Ruhle, J.C. Penny, USA.
ASUM70PR time was in TYPE70PR, but sites without the APAR or sites Dec 1, 1992 with MDF (which does not yet record Effective Dispatch) ended up with LPAR management time equal to LPAR dispatch time. Delete the sixteen lines which set LPnUEDTM=0, and the overhead time will be zero/missing instead of wrong! Thanks to Diane Eppestine, Southwestern Bell, USA.
ANALVARY occurrences of XSWISS were changed to SWISS. This change DOCGRAF works with SAS 6.06, 6.07, 6.08 and later. However the GRAFBNCH SWISS font name is not supported in SAS 5.18, so if you GRAFCICS are still on 5.18, you must change SWISS back to XSWISS GRAFRMFI in these SAS/GRAPH examples. GRAFVM Dec 1, 1992 Thanks to Jim Border, Packaging Corporation of America, USA. Change 10.226 The MXG Tape Mount Monitor has been modified to recognize ASMTMNT the MIM Pseudo Mount event (see Change 10.200), to create a TMNTRTRN value of 3 when the Pseudo event is recognized. Dec 1, 1992 This change has been tested in a non-MIM shop, but the MIM site tests have not been completed to verify that in fact a record with TMNTRTRN=3 is created. Please verify! Change 10.225 The BY list for the WEEK.ASUM70PR dataset should not have WEEKBLD variables LPARNAME LPARNUM, as these variables don't exist Dec 1, 1992 in the summarized ASUM70PR dataset. Thanks to Wayne Bell, National General Insurance, USA. Change 10.224 JES3 TYPE84 INPUT STATEMENT EXCEEDED RECORD LENGTH error. VMAC84 a. Remove the line containing +4 immediately following Dec 1, 1992 JMFJSMXJ $CHAR8. .... b. Three lines later, change LOCJSOF=LOCJSOF+76; to read LOCJSOF=LOCJSOF+72; This MXG error uncovered two other IBM errors. JMFGSNUM is zero, but there is a GMS/MDS summary entry present, but the zero causes MXG to not input the entry. Also noted, R84JSNAM has two leading blanks for DSP names. These are being reported to IBM for repair. Thanks to Ellen Ulrich, Texas Instruments, USA.
VMACROSC ELSE IF MONITOR='JCL' OR MONITOR='JCK' THEN DO; but only Nov 17, 1992 in one line. The 2nd occurrence needed to be changed also! Thanks to Norbert Korsche, OMV-AG, AUSTRIA. Change 10.222 SIMWARE support missed two undocumented bytes, apparently VMACSIM added for alignment. After the INPUT of VUXBPGRM $CHAR20. Nov 17, 1992 insert a line with +2 to skip the extra bytes. Thanks to Mike Roybal, First National Bank in Albuquerque, USA. Change 10.221 DCOLLECT disk capacity dataset DCOLCAPD variable UCTOTAL VMACDCOL contains the volume capacity in Kilo-bytes in the DCOLLECT Nov 16, 1992 record, but was documented as tracks! MXG now multiplies UCTOTAL by 1024 (to convert Kilo-bytes to bytes) and then added UCTOTAL to the list of variables formatted MGBYTES so that the capacity will print 601M for a 3380 (that has 615,472 Kilo-bytes, or 630,243,328 bytes capacity. Thanks to Frank Vessell, ITT Consumer Financial Corporation, USA. Change 10.220 VMPRF for VM/ESA now counts a user as "ACTIVE" if either VMACVMXA the user consumed some virtual CPU time (VMDVTIME GT 0), Nov 13, 1992 or the user was not in the Dormant List at the end of the monitor interval (VMDSLIST NE '0B'X). MXG formerly used only the CPU test to count ACTIVE users, but now the test for ACTIVE has been expanded to include the VMDSLIST test. Note, however, that the whole issue of counting VM users is questionable, since the interval over which the count is taken completely controls the number of "ACTIVE" users. This site examined VXBYUSR with different intervals: 5-min interval active count in the 150 range 15-min interval active count in 240-440 range 30-min interval active count in 700-800 range Thanks to Anne Schroeder, Amway Corporation, USA. Change 10.219 IDMS DBKDBKEY was incorrectly documented by CA. Instead VMACIDMS of the DB Key in four bytes, the key is only the first 3 Nov 12, 1992 bytes, and the fourth byte is the DBKEY Line Index. MXG now inputs DBKDBKEY as PIB3, and creates the new variable DBKDBINX as the fourth byte. Thanks to Sal Fazzino, General Electric Capital Corporation, USA. Change 10.218 MXG 10.2 only. In adding new variables to support LPARS ASUM70PR 9 thru 16 (for Amdahl MDF), several variables were not Nov 11, 1992 properly RETAINed due to spelling errors, causing the ASUM70PR dataset to be corrupted - the effective dispatch times LPxUEDTM and its associated LPAR management time LPxMGTTM will be missing except in the last LPAR, and this caused CPUOVHTM and PCTOVHD to also be incorrect. The total Partition Dispatch Time, LPxUPDTM, fortunately, was not affected, so it is possible you may have not noticed this error. Once you make this change, you should rebuild ASUM70PR in each PDB that was created by MXG 10.2, by allocating the //PDB DD DSN=dsname,DISP=OLD, and then %INCLUDE SOURCLIB(ASUM70PR) to rebuild ASUM70PR. The correction is to look at the end of the RETAIN statement, and correct the spelling so that there are four sets of sixteen variables, each with names LPxUPDTM LPxUEDTM LPxNRPRC and LPxMGTTM, where x=1,2,...,8,9,A,B,C,D,E,F,G. Another minor correction, the 2nd occurrence of LPDMGTTM in the TIME12.2 statement should be LPGMGTTM, and four lines earlier, LPDUEDTM should have been LPGUEDTM. Thanks to Ron Kopfer, First Interstate Bank of Arizona, USA. Thanks to Norbert Korsche, OMV-AG, AUSTRIA. Change 10.217 MXG 10.2 only. MXG Tape Mount Monitor was changed in 10.2 ASMTMNT to default to MVS/ESA, but the SYSPARM parsing was not Nov 2, 1992 updated, which caused an ASM error if you specify TEST in SYSPARM (to write the records to a flat file). The parse logic values are now changed to read: AIF (K'&SYSPARM LT 7).NOSYS &ESA SETB ('&SYSPARM'(5,3) EQ 'ESA') AIF (K'&SYSPARM LT 12).NOTEST &TEST SETB ('&SYSPARM'(9,4) EQ 'TEST') AIF (K'&SYSPARM LT 16).NONUM &C SETC '&SYSPARM'(14,3)
VMAC110 for EXCLUDE logic, and a record with excluded fields could VMAC110M cause MXG to ABEND with INPUT STATEMENT EXCEEDED RECORD Nov 2, 1992 message. Additional testing logic should now prevent this ABEND and alert you that EXCLUDE was used in the CICS MCT. Thanks to Ron Kirk, Union Carbide, USA. Change 10.215 NPM Release 1.5.1 added six new subtypes, causing "INPUT EX028EV6 STATEMENT EXCEEDED RECORD LENGTH" error. That error can EX028EV7 be circumvented by replacing both occurrences in VMAC28 of EX028INE ELSE IF 21X LE NPMSUBTY LE 35X THEN .... VMAC28 with Oct 30, 1992 ELSE IF 21X LE NPMSUBTY LE 25X OR 31X LE NPMSUBTY LE 35X THEN .... Subtypes 26x,27x,28x,29x are session monitor exception and exception resolution events for applications, nodes, LUs& LU groups, and create new datasets NPMEVSAA and NPMEVSAL. (It was the unexpected fall-thru by a NPMSUBTY=26x record that caused the MXG ABEND.) Subtypes 82x and 83x report interval frame relay resources in new dataset NPMINFRP.
SH19-6835 supplement to SC31-6835 (the NPM "SMF" manual) described the new records, the supplement was not SLSS'd to anyone, so it took an MXG user to ABEND to alert me to call one of the two remaining NPM IBM'ers in the USA (NPM is now developed in Rome), who was able to fax me the six critical pages to add the support the next day, but IBM Publications is still researching why I didn't automatically get the change so that it could have been in Newsletter 22 last summer! Thanks to Adrienne Wijlaars, Belastingdienst Automatiser, NETHERLANDS
VMAC16 and Data Space usage, PROC Step name, input and output Oct 28, 1992 DSNs and their first VOLSER, performance group number, the average record length, and flags if HIPERSORT, DATASPACES, and/or Work Datasets were used. Additionally, counts of records and bytes in and out that were previously 4-bytes were relocated and expanded to 8 bytes.
VMAC30 incorrect INITTIME and REND (Reader End) timestamps. IBM Oct 28, 1992 is aware but has no fix at present. The INITTIME contains the initiator start time (not the step start time), and REND contains the initiator start completion time (not the end of Read-in), and hence in this defective record the INITTIME is less than REND! (Note that INITTIME can be less than READTIME, because READTIME is taken where the job originally read-in, and with NJE that could be in a different time zone!). These incorrect time stamps can be detected and corrected with this circumvention to set an INITTIME closer to the truth! IF . LT INITTIME LT REND THEN DO; INITTIME=DHMS(DATEPART(SMFTIME),0,0,LOADTIME); IF INITTIME GT SMFTIME THEN INITTIME=INITTIME-86400; END;
Yüklə 28,67 Mb. Dostları ilə paylaş:
|