* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
elapsed times. MXGACC01=YES creates a summary by QWHSSTCK QWHSSSID QWHCATYP and QWHCAID. The datetime value will reflect the value in the interval parameter. If you use an interval of DATE/WEEK/MONTH the format of QWHSSTCK is adjusted to DATETIME7. and any other interval value it is adjusted to DATETIME13. If no interval is specified it is adjusted to datetime16. Example 1: %ANALDB2R(MXGACC01=YES,INTERVAL=DATE, PMACC01=NO,PMACC02=NO,PMSTA02=NO); creates a summary of all DB2ACCT records in the input PDB.DB2ACCT dataset, at date interval. Example 2: %ANALDB2R(MXGACC01= IF QWHCATYP=4; ,INTERVAL=DATE, PMACC01=NO,PMACC02=NO,PMSTA02=NO); creates a summary of all CICS DB2 account records with an interval of date. -MXGACC02=YES creates an unsorted detail report for every record in DB2ACCT. Given the volume this is most likely NOT what you want to do, without some selection criteria. Example 1: %ANALDB2R(MXGACC02=IF ELAPSTM GT 60 AND QWHCATYP=4;, PMACC01=NO,PMACC02=NO,PMSTA02=NO); creates the detail report, but only for CICS DB2 that executed for over 60 seconds. Thanks to Ron Wells, Springleaf Financial Services, USA. Thanks to Richard Brooks, Springleaf Financial Services, USA. Change 30.185 -VFMT102 is invoked by READDB2/ANALDB2R to build formats ANALDB2R to decode the hex DBID and OBID values in the DB2 IFCID READDB2 105/107 records, but it could be executed many times if VFMT102 multiple IFCIDS or multiple reports were requested. This VMXGINIT change prevents multiple executions within a single pass Sep 15, 2012 of data by setting the global macro variable VFMT102 when it is run the first time, and testing it to prevent the re-execution. If ANALDB2R/READDB2 is restarted in the same SAS session (possibly pointing to different data), VFMT102 will be re-executed. -Variable QW0145LN is printed instead of QW0145LO and $4. format is added for QW0145SC in ANALDB2R audit reports. Change 30.184 New INCODE= parameter added to permit input selection. ANALINIT For example: Sep 12, 2012 %ANALINIT(INCODE=IF DATEPART(READTIME) GE TODAY()-1;); would report on only jobs submitted since yesterday. Change 30.183 MXG 30.05/30.06. PROC FORMAT overlapping range error can DOC occur with READDB2/ANALDB2R. Change 30.140 removed the Sep 15, 2012 datetime from the format, but duplicates can exist if the input spans a very large time frame. VFMT102 in 30.04 or earlier can be used, so it is stored in the new VFMT102X, created ONLY so that support@mxg.com can send that one (renamed) member if you encounter the error. The VFMT102 was altered in Change 30.185 so it cannot be sent without sending additional members to circumvent this error. But: VFMT102X was removed in MXG 30.30. Change 30.182 Support for zNEXT EC12 processors with 101 engines. VMAC7072 There are no new variables kept in TYPE70 since all of Sep 8, 2012 the CPU-specific variables for CPU 64 or higher have been output in TYPE70EN dataset since Change 28.175. Change 30.181 The "Split 70" logic did not provide exits for four data EXTY70EC sets TYPE70EC, TYPE70EL, TYPE70EN, and TYPE70SP because EXTY70EL they were thought to be internal and temporary, but they EXTY70EN now have exits implemented so that users can make changes EXTY70SP to those four datasets if needed. VMAC7072
Sep 7, 2012 Thanks to Ruth Leonhardt, Xerox, USA. Change 30.180 -Two problems both found in testing that does not impact VMXGSUM the output but can improve efficiency of execution. Sep 7, 2012 Case 1: %VMXGSUM(INDATA=PDB.DB2ACCT, OUTDATA=DB2SUM, SUMBY=QWHSSSID, SUM=DB2TCBTM, MAX=MAXCPU, INCODE=MAXCPU=DB2TCBTM; ); In this case, the MXGSUM1 dataset contained all 500+ variables from DB2ACCT when only 3 were needed, because the KEEP= list was not being applied to the output dataset. Now, all variables in the parameter lists is added as a KEEP statement to reduce DASD space required. Case 2: %VMXGSUM(INDATA=PDB.DB2ACCT, OUTDATA=DB2SUM, SUM=DB2TCBTM, MAX=MAXCPU ); In this case, the MXGSUM2 dataset was created by the SORT and since there was no INCODE, the MAXCPU variable did not exist so the PROC MEANS failed with a variable not found error. Now, the KEEP logic after the SORT is NOT suppressed when there is no first DATA step. -Cosmetic. Deletion of temporary datasets used during the course of VMXGSUM moved from the end to the points in the code where they are no longer needed avoiding confusion of the programmer on when/where they should be deleted. Change 30.179 Formula changes from John Burg's SHARE presentation. For ASUM113 the z196 (VN=2) 0.59 replaced 0.57, 1.67 replaced 1.6, & VMAC113 0.61 replaced 1.00; the z10 (VN=1) 0.31 replaced 1.00. Sep 3, 2012 Sep 17, 2012
ANALDB2R caused ERROR 48-59 citing PUT @1 &SORT $CHAR16 text. That Sep 2, 2012 $CHAR format is not only NOT needed here (prints SORTBY= Sep 4, 2012 variable's value, format not needed because all possible variables here are MXG-formatted), it is THE PRESENCE of the $CHAR format that causes this error, so its REMOVAL eliminates this unintended consequence and oversight, and now numeric variables like CONNTYPE and QWHCATYP that exposed the error can be used in your SORTEDBY list, and they will be added in a new QA test of ANALDB2R. -In testing this change it was discovered that if you ran PMACC03 without all of the input datasets (DB2ACCTP, DB2ACCTB, DB2ACCTG) it would fail calling ASUMDB2p/b/g members for those datasets. If the datasets do not exist now, there will be an MXGNOTE printed on the log telling you what was found, with processing to continue without the missing datasets. Thanks to Howard L. Curtis, Duke Energy, USA. ====== Changes thru 30.177 were in MXG 30.06 dated Sep 1, 2012========= Change 30.177 SAS 9.2 z/OS ONLY VGETOBS didn't recognize tape datasets. VGETOBS On z/OS, SAS 9.2 returns a blank in DICTIONARY.TABLES for Aug 30, 2012 the number of OBS in a tape dataset, which caused VGETOBS to return a blank value, but both SAS 9.1.3 and SAS 9.3 returned a bogus set of non-blank characters that proved there was a tape dataset, so VGETOBS returned a one value (it found a tape dataset), and VGETOBS could print a log message that it had detected non-blank non-numeric text, (but only if the NOEXIMSG internal option was enabled to print suppressed log notes). Now, a blank value invokes a new test (VGETDSN length GT 0) for the dataset name that is used to return a value of one for a tape dataset. Thanks to Robert Stoffregen, Great Lakes Educational Loan Svcs, USA.
ANALDB2R are being reported, and new options for the AUDIT= parm: Aug 28, 2012 AUDIT=DML < runs as it currently does uses 143 and 144 Aug 31, 2012 AUDIT=DMLREAD - processes only IFCID 144 for DML AUDIT=DMLWRIT - processes only IFCID 143 for DML and a missing semicolon was found. -Requesting PMACC02=YES by itself, after reading SMF plus executing the ASUMDB2x members, resulted in overlaying many old-style macros that caused error messages. Those old-style substitution macros are now cleared in PMACC02. Change 30.175 Support for IFCID=271 DB2 AUDIT PERMISSIONS trace record. VMAC102 Dataset existed but none of the IFCID-specific fields Aug 28, 2012 were input.
VGETALOC dates formats for mmddyy, ddmmyy, and yymmdd, but mmddyy VMXGALOC and ddmmyy generate slashes in the constructed directory Aug 28, 2012 names; this change allows 6 8 and 10 byte formats for the Sep 16, 2012 mmddyy and ddmmyy but will replace them with the formats mmddyyd and ddmmyyd that use a - rather than a /, like the yymmdd format that already uses a - as documented. -VGETALOC uses %LEFT %macro supplied by SAS in SASAUTOS directory and MXG options in AUTOEXEU.SAS or AUTOEXEC.SAS include SASAUTOS and MAUTOSOURCE to "autocall" it, but %LEFT was not actually required in VGETALOC, so it was removed so VGETALOC would not fail even if the site did not have their SASAUTOS correctly allocated in AUTOEXEU. -GETDATERANGE= parameter will now accept a number of days relative to today to search. For example: %VGETALOC(GETDATERANGE=-5 -9,BASEDIR=C:\MXG, DATEFMT=YYMMDD.); will look for directories in C:\MXG with dates of 12-09-06 thru 12-09-02 inclusive. There is now also an MXGNOTE if any directories are not found. Thanks to Patricia J. Jones, DST Systems, USA. Change 30.173 WARNING: "INVALID QMDA SEGMENT ACCOUNT values are blank" VMACDB2 for records with QMDALEN=258, but QMDAASLN=110 (so there Aug 27, 2012 really were only 110-56=54 possible data bytes remaining) were printed because MXG incorrectly used QMDASFLN=247 as the length of accounting data. But these records don't contain JOB accounting data: text after QMDASFLN is an exact repeat of the first 28 bytes of the QMDA segment, QMDAPTYP,QMDAPVER,QMDAPREL,QMDAPMOD, and QMDAPLAT. Now, MXG uses QMDAASLN instead of QMDASFLN to print any warnings (so these false positives won't print), and the text that follows QMDASFLN is INPUT into QMDAACCT based on QMDAASLN. But, since QMDAACCT has never been KEPT in PDB.DB2ACCT this change was really just cosmetic so the false warning won't be printed on the log. Thanks to Paul Billick, QVC, USA.
VMACRMFV -New variables are added to the ZRBCPU file: Aug 28, 2012 CPUCAPAD CAPACITY*ADJUSTMENT*INDICATION CPUCAPRS CAPACITY*CHANGE*REASON CPUDTIME TIME*DELTA*BETWEEN 2*DIAG CALLS CPUIMMSU IMAGE*CAPACITY*MSU*LIMIT NOTE: CPUDTIME will be missing for LPARs running as VM Guests.
-New variables are added to the ZRBLCP file: LPARCLUS LPAR*CLUSTER*NAME LPARGRPM GROUP*MAXIMUM*LICENSE*UNITS LPARGRPN CAPACITY*GROUP*NAME*FOR*LPAR LPARONCS CENTRAL*STORAGE*ONLINE*IN MB LPAROSNA OS*INSTANCE*NAME LPARUPID USER*PARTITION*ID NOTE: Variables LPARCLUS, LPARGRPN, LPARGRPM, LPARONCS require RMF Monitor III data at the z/OS 1.12 Level to be populated. -Change 29.013 for CPCGRPLM, text and the INPUT statement, were both reversed; the four byte variable is input first and then the reserved four bytes are skipped.
VMXGINIT IBM had never externalized that DSECT, so MXG had never Aug 27, 2012 created any observations, and because it kept showing up as an exception on QA reports. However, I also removed the dataset token macro variables &PCICMNR and &WCICMNR from VMXGINIT, which cause UNRESOLVED MACRO ERROR when the new VMXGINIT was used with MXG 30.02 or earlier. I should never have removed those macro variables and so they are restored to be defined in VMXGINIT. Thanks to Pat Hansen, ADP, USA. Thanks to Mike Chaves, ADP, USA.
EXCICLDR with only 70,000 CICSTRAN observations; there is one obs Aug 27, 2012 for each program that has been active, but almost all of the observations have zero Fetch Count (LDRFC) so this change now output CICLDR observations ONLY if LDRFC is no zero; that test is located in the EXCICLDR exit member in case you ever want to change and output them all. Thanks to Chuck Hopf, Independent Consultant, USA. Change 30.169 -Dataset T102S083, IFCID=83, always had zero observations FORMATS due to an LE test for QWHSRELN that should have been GE. VMAC102 -IFCID=145 record with unexpected QWT02R1L=0 ABENDed with Aug 25, 2012 invalid DO loop value. The records are continuations for Oct 12, 2012 an SQL command with a total of 98,567 characters, so that multiple SMF record are written, and the continuation SMF records do not contain the expected R2O segment, so MXG's logic was revised. However, a problem has been opened with DB2 Support because the QW0145LE length-field is not correct, and there is no explicit offset field to the SQL text in both the initial and continuation records, so the MXG input is heuristic based on the available records and the actual SQL text length in each observation is in the MXG-calculated QW0145LNTXT variable. -Oct 12: IBM APAR PM74186 corrects the SQL length error. -These new-in-DB2-V10 QW0145 fields are INPUT and kept: QW0145IS QW0145SI QW0145OBNUM QW0145ACNUM QW0145LL QW0145LNTXT and format $MGD145I decodes the Isolation Level QW0145IS. Thanks to Chuck Hopf, Independent Consultant, USA. Change 30.168 Enhancement to ASM program that recreates RECFM=VB or VBS ASMDBLKU data from downloaded ASCII "RECFM=U" that is now reloaded Aug 24, 2012 back to z/OS into 32760 byte chunks that ASMDBLKU reads Aug 29, 2012 to reconstruct the original VB/VBS file (perhaps so you can terse it and send it on to a vendor), for sites that don't have SAS on z/OS (when you would instead use the MXG UDEBLOCK program to recreate valid VB/VBS files). -JCL PARM field values of PARM=VB or PARM=VBS are now supported. The default is VBS if no PARM field is coded. PARM=VB creates RECFM=VB,LRECL=32756,BLKSIZE=0, while PARM=VBS creates RECFM=VBS,LRECL=32760,BLKSIZE=0 as in prior versions of ASMDBLKU. Any other PARM field values will result in error message DBLKU29S and termination with Return Code 16. -New message DBLKU29I shows PARM field value if present. -Message DBLKU25I now also shows the average LRECL of the input SMFIN file. -Message DBLKU26I now also shows the average LRECL of the output SMFOUT file. -New message DBLKU27I shows the number of input SMFIN file records that were read at the minimum and maximum LRECL values. This provides an idea if the record lengths on input were biased toward these values. -New message DBLKU28I shows the number of output SMFOUT file records that were written at the minimum and maximum LRECL values. This provides an idea if the record lengths on output were biased toward these values. -Maximum LRECL value for the output SMFOUT file was incorrect in message DBLKU26I.
VMACIMS 3826 is deleted, to eliminate the unwanted log messages. Aug 24, 2012 Thanks to David J. Schumann, Blue Cross of Minnesota, USA. Change 30.166 Change 30.153 supported keeping PROGRAM and PROG1-PROGnn ASUMCICX variables in PDB.ASUMUOW, but its tailoring instructions ASUMUOW were incomplete, and you may need this updated IMACUOW. IMACUOW To add PROGRAM (and optional PROG1-PROGnn variables): VMXGUOW - you must update ASUMUOW to add (only) variable PROGRAM Aug 24, 2012 to the MACRO _KUOWIDC definition, - you must update ASUMUOW to add PROGRAM, plus any of the PROG1-PROGnn variables that you want kept, into your MACRO _KUOWPTH definition, - if the PROGRAM name you want/need isn't the PROGRAM in that last CICSTRAN in this UOW, then you will need to use this new IMACUOW tailoring and start with Case 5 as your MACRO _TRANUOW selection, but be aware there still could be additional site-dependent mods required. This example adds PROGRAM and PROG1-PROG3: %LET MACKEEP=%QUOTE( MACRO _YESOBS % MACRO _NOOBS % MACRO _KUOWPTH EXECAPPL PROGRAM PROG1-PROG3 % MACRO _KUOWIDC APPLID USER LUNAME SYSTEM TERMINAL USERCHAR PROGRAM %
%INCLUDE SOURCLIB(VMXGUOW); _NOOBS
OPTIONS NODSNFERR NOVNFERR; _SUOWCIC /*SORT CICSTRAN DATA */ _SUOWDB2 /*SORT DB2 DATA */ _SUOWMQ /*SORT MQ SERIES DATA */ _SUOWSPN /* CREATE ASUMUOW DATASET */ %VMXGUOW; _YESOBS OPTIONS DSNFERR VNFERR; If you use ASUMCICX to summarize the PDB.ASUMUOW CICS in PDB.CICS output dataset, you may need to add PROGRAM to the _BSUxxxx macro, as also now documented in comments. The only code changed in this change was in IMACUOW, but that is only the addition of the new Case 5 example. Thanks to Charles Savikas, DCF, State of Florida, USA. Change 30.165 Cosmetic. Existing $MGTMDAC format assigned to QW0142AC, VMAC102 to decode A/C/D to ALTER/CREATE/DROP. Note that when an Aug 23, 2012 MXG $MGxxxxx character format is assigned to a variable, that variable must be in a LENGTH statement with the input length; in the absence of a LENGTH statement, the old $1 character variable is changed to the length of the longest text in the $VALUE statement for that format.
EXTAPCPU six datasets; EXTAPIPU EXTAPJOB DDDDDD MXG MXG CREATED EXTAPOBJ DATASET DATASET DATASET FROM EXTAPPHY SUFFIX NAME LABEL INFILE EXTAPVOL IMACTAPR TAPCPU TAPRCPU TANDEM PROGNOSIS CPU TAPRCPU TYPETAPR TAPIPU TAPRIPU TANDEM PROGNOSIS IPU TAPRIPU TYPSTAPR TAPJOB TAPRJOBS TANDEM PROGNOSIS JOBS TAPRJOB VMACTAPR TAPOBJ TAPROBJECT TANDEM PROGNOSIS OBJECT TAPROBJ VMXGINIT TAPPHY TAPRPHYSVOL TANDEM PROGNOSIS PHYSVOL TAPRPHY Aug 24, 2012 TAPVOL TAPRVOLUME TANDEM PROGNOSIS VOLUME TAPRVOL JCLTEST9
Aug 26, 2012 The TYPETAPR or TYPSTAPR members will create all six MXG Aug 31, 2012 datasets, reading each of the six input files, but the Sep 25, 2012 individual datasets can be created with the _Tdddddd to create and the _Sdddddd token to sort each one. See examples in comments in VMACTAPR. -Sep 25: DD MON YYYY HH:MM:SS format for CPU001 supported. -These problems need to be resolved by Prognosis vendor: JOBS fields: 014,034,040-044,082,083: a. Invalid values in 014 - IR_ELAPTIME_STRING 1:29
109-05:50:31 48 b. Invalid values in 034 - IR_STARTTIM_STRING Values of 19 AUG 2012 and 02 MAY 2012 but the IR_STARTTIM_TIMESTAMP is the same these records. c. What are the units, where's the decimal, one obs has: 040 IR_SENDRESP value 0.0185 042 IR_PERFORM value 0.0104 043 IR_RESPTIME value 0.0185 044 IR_CPUSECS value 133877 - if in microseconds, then 0.133877 is larger than response. d. Range of values are strange: 082 IR_SEQNUM - 1617862657, 8, and 1628929 083 IR_ANCSEQNO - 1617861633, 0, and 1895169
062 IR_FULLTIME_TIMESTAMP 0 063 IR_FULLTIME_GMT_OFFSET 0
0:00 (no seconds), but format 20 Aug 2012 0:00:00 is used in the other records.
have enough digits to decode correctly; other records have 212212195260000000 microseconds, which does have enough digits and after adding the Tandem constant to convert from proleptic time, is a valid GMT datetime.
but that is changed to "CRLF" if the file is copied to a Windows platform. Macro variable MXGCRLF transparently sets the TERMSTR option of the INFILE statement to use the correct delimiter. It is set to TERMSTR=CRLF on WIN platforms, to TERMSTR=LF on all other ASCII platforms, and it is blank on z/OS. It can be changed with %LET. -The clue you have the wrong record delimiter is an error "ARRAY EXCEEDED" that prints the first two records where you can see the text date field after the last IR header name is listed as another field in that first record. -The JOBS "comma-delimited" file has double quotes around all character fields, and some contain embedded commas, which required special handling due to MXG's use of SCAN function which has no DSD option. A vendor option to to change the delimiter to a tab would have been nice. Double quotes were removed, and the commas restored. -GMT Offset is not provided in the IPU record. -All fields with "PERC" are labeled (MXG-style) "PCT".
VMACWWW C63BCB3DB05CA94A1ECF2B1C13C048097E3456C966691FE6EEB4168 Aug 21, 2012 D55EB32905A6221CB8CA5C525 Sep 28, 2012 that did not contain the expected equal sign caused MXG to loop; that unexpected text value is now protected. -Data records with ^M (CR) were created when IIS file was moved to AIX could be stripped using DOS2UNIX command, but can also be read with MXG using %LET MXGCRLF = TERMSTR= CRLF ; before the %INCLUDE of TYPEWWW, because this change added &MXGCRLF in the INFILE statement in VMACWWW. Thanks to Xiaobo Zhang, FISERV, USA. Change 30.162 Documentation. The RACF field AUDITFID is MXG variable VMAC80A RACF264 in RACF event datasets TYPE8028/29/30/44/45/8055, VMACRACF created from SMF 80 records, a 16-byte $HEX32 formatted Aug 21, 2012 variable because it contains both hex and text: e.g.: 'C8C6E2F0F0F93D0B00000000004D385B'X. and the RACF field AUDITFID is MXG variable AUDITID in dataset RACF0900 from IRRDBU00 RACF UNLOAD UTILITY data, a $32 character variable with the above hex as text. Both variables have been in MXG for years, but I have added "AUDITFID" text to both variable's label. Thanks to Mark Jacobs, Time Customer Service, USA.
ASMRMFV -A new message RMFV033E is issued when a VSAM Read Error Aug 19, 2012 occurs to show the current and maximum Relative Record Number, RRN, of the RMF III record being read, and the return/reason codes from the failed read operation. This message will aid in the investigation of sporadic VSAM read I/O errors that go away when ASMRMFV is rerun. We will use these messages to circumvent the error. -NOTE: We have never had a VSAM Read error repeat with a rerun; if you error with ABREAD please add a //SYSUDUMP, and send the dump to support@mxg.com so we can resolve. -The SHOWCB macro calls that collect RMF III VSAM file Yüklə 28,67 Mb. Dostları ilə paylaş:
|