* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
that externally and correctly deaccumulates NPMINPMT. DIFF28 is included by member TYPE28, but if you have added type 28 processing to your BUILDPDB, then you must add the statement %INCLUDE SOURCLIB(DIFF28); in member EXPDBOUT to properly deaccumulate NPMINPMT. Thanks to Steve Donahue, Blue Cross and Blue Shield of Texas, USA.
ANALUOW -Member ASUMUOW combines all of the pieces of CICS MRO ASUMCICX transaction (i.e., TOR, AOR, and FOR records for the ASUMUOW same Unit-of-Work are combined, and the correct TRANNAME IMACCICS is picked up), and (if the transaction called DB2) also TRNDCICX merges in all of the DB2ACCT records to create dataset Aug 28, 1997 PDB.ASUMUOW that gives a complete picture of each CICS unit-of-work. This revision to ASUMUOW adds CICS wait durations/counts to PDB.ASUMUOW so you can determine the cause of poor CICS reponse time. -Member ANALUOW is a parameter driven report to examine why CICS transactions were delayed, using PDB.ASUMUOW. -Member ASUMCICX is an alternative summary member, like ASUMCICS to create dataset PDB.CICS, but member ASUMCICX will use dataset PDB.ASUMUOW as its input, and also keeps the wait durations/counts. -Member TRNDCICX is an alternative trending member, like TRNDCICS, but it keeps the wait durations/counts. If you have CICS MRO and/or CICS talking to DB2, you should change the %INCLUDE SOURCLIB(ASUMCICS) in your BUILDPDB job to %INCLUDE SOURCLIB(ASUMUOW,ASUMCICX) so as to build both PDB.ASUMUOW and the enhanced PDB.CICS. The JCLPDB6 example has been updated to build both. The example macro _KCICTRN in member IMACCICS that keeps only the variables needed by ASUMCICS has been revised to keep the original ASUMCICS set or the new ASUMCICX set of variables, to reduce the size of CICSTRAN. Thanks to Chuck Hopf, MBNA, USA. Change 15.204 Tape Capacity Planning graphs based on TREND database GRAFTAPE and MXG ASMTAPES tape allocation and trending records, Aug 27, 1997 and also the HSC and LSM data from an STK silo. If you saw Chuck's paper at CMG94, this is the code that generated those graphs. Thanks to Chuck Hopf, MBNA, USA. Change 15.203 Trending of NTSMF's NTINTRV dataset creates summarized TRNDNTIN TREND.TRNDNTIM for long term trending of NTSMF systems Aug 27, 1997 with minimal storage required. Thanks to Chuck Hopf, MBNA, USA. Change 15.202 Enhanced measurement of Tape Allocation uses the Interval TRNDTALO Allocation record created by ASMTAPES at ML-13, MXG 15.02 Aug 27, 1997 or later, so you must have installed ML-13 before you can use this enhancement. The interval record eliminates any undercounting of devices still allocated and improves the accuracy of tape allocation counting. TRNDTALO assumes you have created WEEK.ASUMTALO in your WEEKBLD/WEEKBLDT member from your daily PDB.ASUMTALO data sets; you will need to add the data step in your weekly. Thanks to Chuck Hopf, MBNA, USA. Change 15.201 Related members in a way. ASUMCACH ASUMCACH creates the union of the TYPE74_1 and TYPE74_5 ANALEMC data (device level and cache reporter). This is useful EMCRANK in figuring out problems in the IO subsystem but it will EMCMAP also be part of the "I/O Stimulator" that will generate Aug 27, 1997 control cards for Stress Test based on your RMF data. ANALEMC takes the ASUMCACH data and carries it a step closer to the hardware by using a series of formats to map the logical to the physical in an EMC world. It will also do this for Spectris and HDS 7700 boxes but no formats are required. It creates datasets RANK, RAIDVOL, EMCMAP, and VOLUMES. Each is a slightly different level of summarization except for EMCMAP which is used to apply labels to graphs with the annotate facility of SAS/GRAPH. EMCRANK and EMCMAP use the output of ANALEMC to build graphs of the performance of EMC boxes. This is a grid 8*12 where each box represents on physical SCSI disk on the inside of the box. Each box contains the VOLSER, RANK, and DEVNR of the logical MVS volumes and EMCRANK applies colors to represent IO rates. EMCMAP leaves all of the boxes empty of everything except text to make the text a little easier to read. Find the line that says if text=:'EM' then color='ORANGE'; This is used to alter the color of the text. In this code, volumes that contain no data start with EM in the VOLSER. Making those volumes ORANGE makes them stand out and is a quick visual indicator of available capacity. Thanks to Chuck Hopf, MBNA, USA. Change 15.200 Support for Peregrine's Service Center SMF record creates EXTYSVCC new dataset TYPESVCC with one set of variables and two IMACSVCC subtypes, which can be identified by variable SVCCTYPE: TYPESVCC SVCCTYPE Description VMACSVCC 1 Service Center Statistics Aug 27, 1997 2 Service Center User Termination Thanks to Chuck Hopf, MBNA, USA.
TYPETMS5 new flag variables to both DSNB and Volume records: VMACTMS5 Dataset TMSREC: Aug 25, 1997 Variable Description ADDLFILS='ADDITIONAL*FILES*EXIST IN*VOLSET?' COPYCAT ='CREATED*BY*COPYCAT?' DEFEXPOO='DEFAULT*EXPDT*USED AT OPEN*OUTPUT?' DEFEXPOO='NON*RESIDENT*TAPE?' DEGAUSSD='TAPE*HAS BEEN*DEGAUSSED?' EMVREL ='RELEASED*BY EXTERNAL VAULT*MANAGER?' ERASEREQ='DATASET*ERASE*REQUIRED?' FILEISCA='FILE*ON OS*CATALOG?' SMSEXPIR='EXPIRED*BY SMS*MAX RETENTION?' VAULTREQ='VAULT*SPECIFIC*REQUEST?' VOLACTV ='ACTUAL*VOLSER*IN USE?' Dataset DSNBREC: Variable Description DSNBABND='CLOSED* BY ABEND*PROCESS?' DSNBDFXU='DEFAULT*EXPDT*USED?' DSNBECAT='EXPIRED* BY CATLG*INTERFACE?' DSNBISCA='FILE IS*ON OS*CATALOG?' DSNBLBL ='B1*SECURITY*LABEL?' DSNBTMSI='EXPIRED* BY TMS*INTERFACE?' DSNBWSCA='FILE WAS*ON OS*CATALOG?' The Release 5.2 data records show that TMS julian dates in year 2000 are supported. The 311th day in year 2000 will have julian data of 100311. Thanks to Xiaobo Zhang, Insurance Services Office, Inc, USA. Change 15.198 CICS reports were updated to add new data step for CICLDR ANALCISH processing to calculate the "average fetch time" where Aug 25, 1997 previously the "total fetch time" was printed. Thanks to ???, ???, USA. Change 15.197 ACF2 dataset ACF2JR did not populate variable ACLFLDVL VMACACF2 if ACLARCFL=02x, because that field type had not been Aug 24, 1997 previously seen. The data value is a hex string rather than the EBCDIC characters normally stored in ACLFLDVL, but rather than creating yet another variable for the hex representation, I chose to use ACFLLDVL, but to convert it to the printable-hex value using: SUBSTR(ACLFLDVL,1,2*ACLFLDLG)=PUT(ACLFLDVL,$HEX100.); so the three-byte data value of '010203'x in the ACF2 record will cause ACLFLDVL to contain six bytes 010203 as printable characters. Thanks to Peter J. Lines, NatWest Bank, ENGLAND. Change 15.196 New utility old-style MACRO _SMFGOOD defined in VMACSMF VMACSMF was needed to read an SMF tape with truncated records at Aug 24, 1997 the end of each block followed by a trashed record at the start of the next block. As an example of its use: MACRO STOPOVER MISSOVER % %INCLUDE SOURCLIB(VMACSMF,VMAC80A); DATA _VAR80A; _SMFGOOD ; _SMF ; _CDE80A ; RUN; deletes the trashed records and prevents invalid-smftime error messages and hex record dumps from being printed, while the MISSOVER prevents input-statement-exceeded errors from causing stopover on the truncated records. At the same time, variable SMFRECNR is now created in the _SMF macro, so it can be added to a _Kxxxyyy macro if you need to know which SMF record created an observation. Very few users will need to use these enhancements!
VMACHURN changes themselves were compatibly made, you must change Aug 23, 1997 VMACHURN to read them. Change IF SMFHVR EQ 02X THEN DO; Aug 26, 1997 to read IF SMFHVR LT 02X THEN DO; and MXG will tolerate (i.e., process the records correctly) Version 3.0, but to exploit (i.e., pick up the new fields) you will need this change. New dataset HURN06 Lock Manager Statistics is created for both subtype 22 (Interval) and subtype 6 (Termination) as both records are identical in structure. New fields were added to existing datasets: Control Region Start Time variables HU50CRTM, HU51CRTM HU52CRTM HU60CRTM HU61CRTM HU62CRTM and HU72CRTM are added to HURN501 HURN511 HURN521 HURN601 HURN611 HURN621 and HURN721 datasets. Control Region Session Identifier variables HU60UNQI HU61UNQI HU62UNQI HU72UNQI are added to HURN602 HURN612 HURN622 and HURN722 datasets. Thanks to Trevor Ede, Bank of New Zealand Ltd, NEW ZEALAND Thanks to Glenn Bowman, Wakefern Food Corporation, USA. Change 15.194 The MXG default for MEMSIZE was raised to 64M (from 48M) CONFIG because if you add lots of additional record processing Aug 22, 1997 (DMON,ENDV,FOCU,HSM,IDMS,28,NSPY,NTCP,PROS,80A,SASU, STC,SYNC, and X37, for example!), you will need slightly more than 48M of virtual addressability. Since there is no cost to large virtual storage limit for programs that do not need it, changing the MXG Maximum Default may save someone diagnostic time to resolve OUT OF MEMORY errors. The symptoms were that the IEF374I message on SYSLOG had: VIRT = 1572K SYS = 308K EXT = 47316K SYS=10768K the EXT value suggested a 48M limit was set somewhere! Why 64M? Why not 80M? No good reason. -z/OS 1.12 message IEF032I/33I replaced IEF374I/276I. Thanks to Susan Walters, Michelin Tires, USA. Change 15.193 Another invalid '04' Catalog Cell requires protection. VMAC6156 Add the IF SKIP GE 2 THEN DO; and END; statements VMACCTLG around the two statements to end up with: Aug 22, 1997 IF SKIP GE 2 THEN DO; INPUT VOLHKYLN &PIB.2. @; SKIP=SKIP-2; END; The '04' cell had nulls for VOLSER, and only had a low key range; the high key range was non-existent. Thanks to Dave Millard, Southern California Edison, USA. Change 15.192 Support for HMF SMF record Subtype 11 (DS3 Statistics) DIFFHMF adds new dataset TYPEHMFB, with 44 counter and endpoint EXTYHMFB values for each link (HMFBUSNO) during each polled IMACHMF interval. Since the HMF counters are accumulated, the VMACHMF dataset TYPEHMFB must be de-accumulated by logic added Aug 21, 1997 to DIFFHMF. Sep 2 update: all non-endpoint counters Sep 2, 1997 are protected for wrapping. Thanks to Anne Robbins, Bank of New York, USA. Change 15.191 ANALDB2R fails with ERROR 31-185 if the SORTBY list did ANALDB2R not contain PLAN. The error was because variable PACKID Aug 19, 1997 was not initialized as character; as long as PLAN was in the SORTBY list, PACKID's definition was picked up by the assignment IF ... THEN PACKID=PLAN; but that statement was not generated when SORTBY did not list PLAN. To correct: Find %MACRO PMACC02; and then find RETAIN PAGENUM; (about 957 lines later) and then insert LENGTH PACKID $18; after that RETAIN statement. Thanks to John Shuck, SunTrust Companies, USA. Change 15.190 Support for 5 new NTSMF Objects, validation of untested ADOCNTSM objects and some new counters, and utility for Discovery EXNTCNME Records. EXNTFULC -Five new NTSMF Objects create six new datasets: EXNTPEP5 Dataset Object Name ENNTSNAA CONECTME Connect-ME EXNTSNAC FULCRUM Fulcrum Server EXNTWBSS SNAADAPT SNA ADAPTER SNAD1c1 IMACNTSM SNACONN SNA Connections VMACNTSM WEBPRSRV WEB Proxy Server Service Aug 19, 1997 PENTP5 PENTIUM (see next note) Aug 25, 1997 -Two forms of the PENTIUM object exist. For PENTIUM PRO P6 processors, the record contains the expected 68 fields to output in the PENTIUM dataset, but for PENTIUM P5, there there are only 58 fields, with completely different field names, so the new PENTP5 dataset is now created from PENTIUM objects with only 58 fields. Instead of sensible variable names, variables PENTP501-PENTP548 are created (its a lot quicker, and I expect little need for these counters); there are only 48 variables because ten of the fields are place-holders. Mark has discovered that for either Pentium object, only two pairs of Pentium counters can be enabled at a time, and the counter values are accumulated so that you will need to subtract adjacent observations to calculate the delta using SAS's DIF() function until MXG provides a de-accumulation member, but this support is probably all that is needed for this hardware-level data. -Five untested NTSMF records have been validated; three have instance names (cannot tell without test data, as the Discovery records do not identify instances): Dataset Object Name Instance MSEXCHIP MSEXCHANGE INTERNET PROTOCOLS PROTNAME MSEXCHIS MSEXCHANGEIS 3 new vars MSEXCHMC MSEACHANGEMTA CONNECTIONS CONNNAME MSEXCHPU MSEXCHANGEIS PUBLIC 35 new vars NETWSEGM NETWORK SEGMENT SEGMNAME -The new utility macro, _UNTDISC, that can be used to print the Discovery records, is defined in VMACNTSM, and comments before the MACRO definition show how to use the utility to find new counters and new objects. -In Microsoft's NT Resource Kit, there is a utility \PerfTool\CntrTool\Cntrlist.exe that will provide detail explain text for all counters; unfortunately it does not always work in decoding new objects and has been known to abend, and Microsoft does not support their ResKit, but now you know it is there. Thanks to Jim Quigley, Consolidated Edison, USA. Change 15.189 Support for new ADSM VM Account Records from VM/ESA. IMACVM New dataset VMADSM matches the TYPE42AD dataset created TYPEVM from type 42 subtype 14 SMF records on OS/390 and MVS. VMAC42 While MVS puts the data in one SMF record, VM Accounting, Aug 15, 1997 forever limited to an 80 byte physical card image record, has broken new ground by defining subtype and sequence numbers so that more than 80 bytes can be recorded for an account event. The subtype 1 and subtype 2 of the 'C0' must be found in sequence in the VM Accounting input, or MXG will Delete with Error Message. Since Multiple User products can create 'C0' VM Accounting Records, the User ID Name of the ADSM Server must be hardcoded in TYPEVM; MXG expects the name to start with DSMSERV or records will not be recognized by MXG as ADSM-created. The new variable CLSESTYP is now created also in TYPE42AD; that location was a reserved field in original ADSM SMF doc. VM also creates a subtype 3 ADSM record, but I need test data to decode and validate that record. I took this opportunity to update the VM Accounting Support in MXG; the _KVMxxxx macros defined in IMACACCT are now actually referenced in the TYPEVM Data Statement and labels for the data sets were clarified: VMID Dataset Description 01/C1 VMSESSN Virtual Machine Resource Usage 02/C2 none Dedicated Devices 03/C3 none Temporary Disk Space 04 VMLOGON Invalid LOGON Password 05 VMDISK LINK to Not-Owned Protected Disk 06 VMLINK Invalid LINK Password 07 VMVCNA SNA/CCS Logoff/Disconnect 08 VMLOGOFF User Logoff/Disconnect/Shutdown/Force 09 none ACNT ALL - ISFC Accounting ISFC Initialization ISFI Accounting ISFC Conversation Accounting ISFS - Conversation Start ISFA - Conversation Active ISFE - Conversation End ISFC Link Statistics ISFL ISFC Termination Accounting ISFT 0B VMDEVICE Virtual Disk in Storage C0 VMADSM USER=:DSMSERV ADSM Accounting VMRSCS USER=:RSCS RSCS Accounting VMSQLxxx USER=:SQLDBA SQL Accounting VMSQLINI SQL INIT VMSQLSYS SQL SYSTEM VMSQLTRM SQL TERM VMSQLUSR SQL LOCAL USER VMSQLRMT SQL REMOTE USER VMUSRDAT USER=other USER Data
data is provided. The VM/ESA Account records still do not contain four-digit years (except in new ADSM record) but MXG's Change 15.167 does protect VM Account Records for the year 2000. Note: CHANGE 28.157 Added Support for the 09 record. Thanks to Ned Hammond, American Management Systems, USA.
VMACOPC because 19 reserved bytes after ETCNAME were not skipped. Aug 15, 1997 -Insert +19 between ETCNAME and ETCDESC to correct. -Dataset OPC23 only had obs for TRLEVT23='C' (Completed) or 'E' (Error). Now, events 'S' (Start) and 'X' (Reset) are also output, but date variable TRLDUR23 is nonmissing only for the 'C' and 'E' events. -Invalid OPC29 records have been observed, but are thought to be an IBM problem, which is under investigation. Thanks to Randy Shumate, LEXIS-NEXIS, USA. Change 15.187 CMF. Variable C279SSI in dataset CMF27C93 was INPUT as VMACCMF &PIB.2., while variable C279CSSI in dataset CMF27CSC was Aug 7, 1997 INPUT as $CHAR2., but both variables are Subsystem ID, Aug 28, 1997 and should be the same so they can be MERGED. Therefore add variable C279SSI to the $HEX4. format list, and change C279SSI &PIB.2. to C279SSI $CHAR2. -Aug 28: New variable C279DDIC is created by adding the Base address to the Device ID and converting to character so that new variable C279DDIC matches CMF27UA1 in the CMF27CSC dataset, because merging datasets CMF27CSC and CSC27C93 must be done BY both SSI and DDI. C279DDIC=SUBSTR(PUT(INPUT(CMF27CCU,S370FPIB2.)+ INPUT(C279DDI,S370FPIB1.),HEX8.),5,4); Thanks to Neil Ervin, Banc One Services Corporation, USA. Change 15.186 STK 4400 SMF records. STCLMU dataset character variables VMACSTC LSBECON1 and LSBECON2 were $CHAR6 fields, but are now Aug 6, 1997 replaced with four variables: LSBELSM1='PASSTHRU*PORT 1*CONNECTED TO*LSM' LSBEPAN1='PASSTHRU*PORT 1*CONNECTED TO*PANEL; LSBELSM2='PASSTHRU*PORT 2*CONNECTED TO*LSM' LSBEPAN2='PASSTHRU*PORT 2*CONNECTED TO*PANEL' Thanks to Bob Gauthier, American Stores Company, USA. Thanks to Bruce Sloss, PNC Financial Corporation, USA.
VMAC50 records are now greater than 78 bytes (although I still Aug 6, 1997 can find no documentation of the IBM change). The test LENGTH EQ 78+OFFSMF must be LENGTH GE 78+OFFSMF. Also, variable NCPSLOWS may be defective, as it has values of zero or 7FFFFFFx (which I am also pursuing with IBM). Note added Sep 30: It is not possible to distinguish between an XCF MPC Group Record and a non-XCF MPC Group record using ATTCHTYP or VERSNO, but for an XCF MPC Group record, both DEV and BSIZE will be zero, while for a non-XCF MPC Group record, at least one will be nonzero. Thanks to Bill Feeney, Hennepin County, USA. Change 15.184 TMON/DB2. Change 15.114 did not properly handle the "DE" VMACTMDB record, which only has part of the standard header. Also Aug 6, 1997 the dates input with MMDDYY informat were changed to use &NUM.2 so that VMACTMDB will correctly execute under ASCII systems as well as under MVS. Thanks to Luc Gariepy, Regie des rentes du Quebec, CANADA Change 15.183 Dataset TYPE72GO could have observations OUTPUT that had VMAC7072 no resource consumption for that period, because the MXG Aug 4, 1997 variable NOACTVTY was not reset to 0 for the second and subsequent periods within a Service Class. The new test for _I_ GE 2 was added to zero NOACTVTY (which is then used in EXTY72GO to control execution of the OUTPUT): DO _I_=1 TO NRSCS; IF _I_ GE 2 THEN NOACTVTY=0; INPUT @COLSCS Thanks to Ove Dall-Hansen, CODAN Insurance, DENMARK. Change 15.182 Variable VELOCITY in TYPE72GO was wrong for service class VMAC7072 periods with Discretionary or System goals; the value was Aug 4, 1997 carried forward from prior periods. Since neither System nor Discretionary goals have velocity, the variable is now set missing with: ELSE IF R723CRGF='S' OR R723CRGF='D' THEN DO; PERFINDX=.; END;
Thanks to Brenda Rabinowitz, Change 15.181 INVALID ARGUMENT TO FUNCTION INPUT in BETA93 SMF record VMACBETA is caused by a value of '*RELOAD*' in the JCTJOBID field Aug 4, 1997 from which MXG extracts the job's JES Number JESNR, for the BETA93 RELOAD step. Adding two question marks: JESNR=INPUT(SUBSTR(JCTJOBID,4,5), ?? 5.); to the INPUT function supresses the NOTE and the hex dump of the input record. With or without the question-mark modifiers, JESNR will be a missing value. Thanks to Winfried Waldeyer, BHF-Bank Frankfurt, GERMANY.
VMACVMON old-style macro has text in column 1 is circumvented in VMACVMXA MXG by ensuring a blank is in column one in the MXG code VMACXAM members that worked under MVS but failed under Windows. ANALDBTR This error was seen long ago, when initially testing MXG Aug 3, 1997 on ASCII platforms; it showed up now as I began testing of all of MXG under Windows. The symptom was a SAS error message "Variable ACTIVEDSVMAXUS longer than 8 bytes", (even though ACTIVE and DVSMAXUS were on separate source lines), but the variable names are in an old-style macro and the variable name started in column one! Inserting a blank in column eliminates the compiler's confusion in these members previously untested on ASCII platforms.
IMACICBB support the year 2000 in MV for CICS Release 5.2. Their YEAR2000 date yymmdd0F will be changed INCOMPATIBLY to yyyymmdd, Aug 1, 1997 so you will need MXG 15.04 or later before you install that release (late 1997?), if you have enabled IMACICBB to process their statistics segments in type 110 SMF records. MXG tests the 0F byte to determine the format. -Boole & Babbage MainView for IMS (was IMF) will support the year 2000 in Release 3.2, using yyyydddF format in place of 00yydddF, but MXG already supported that format so there was no MXG change to TYPECIMS code. Yüklə 28,67 Mb. Dostları ilə paylaş:
|