* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
attributes have been consolidated to a single call to reduce code path length. -New parameter ABREAD/NOABREAD lets you control ASMRMFV's action when a VSAM read I/O error does occur: -ABREAD is the default and duplicates the current ASMRMFV behavior: a USER 0998 abend occurs on the first instance of a VSAM read I/O error and no further data is written to the RMFBSAM output file by ASMRMFV. Alias is ABR. -NOABREAD suppresses the above abend and the action taken depends on the data being read. Alias is NOABR. On the first read of an RMF III VSAM data set the action is to close this VSAM dataset and then open the next RMF VSAM file (if any). Otherwise, if the error occurs on a subsequent read for sample set data, the problem sample set is skipped and ASMRMFV attempts to read the next sample set from the same data set (if any). -ASMRMFV ends with Return Code=12 if NOABREAD is in effect and a read I/O error has occurred. -Program prologue documentation is updated to discuss ABREAD and NOABREAD and new Return Code 12. -With either ABREAD or NOABREAD, the output RMFBSAM file is likely to be missing data after a read I/O error. -NOTE: When ASMRMFV is being executed under the CLRMFV CLIST, the abend (if ABREAD in effect) is trapped, and with either ABREAD or NOABREAD, CLRMFV will eventually end with Return Code=12 after a VSAM read I/O error. You could use the JCL COND parm to test for RC=12 in the subsequent PDB build, if you want to also terminate the creation of the RMF III PDB datasets after a VSAM error. Thanks to Rodger Foreman, TransUnion, USA Thanks to Betty Wong, Bank of America, USA
ASMIMSL7 containing invalid Packed Decimal data, by skipping the ASMIMSL8 0C7'ing-records so the rest of the log can be read. If Aug 18, 2012 invalid ARRVDATE is found, a new error "NOPD" message Aug 30, 2012 will be written to the job log with the first 25 bytes of the record in hex. Since we have added this diagnostic, no reports of NOPD messages have been reported. -ASMIMSL8: ASMIMSL7 ABEND 0C4 occurred when an INQUEUE record had zero record length in the first four bytes, in a record that was to be discarded due to other errors (dupe key). This iteration deletes the defective record without the 0C4 ABEND. -We have been unable to replicate either the 0C4 or 0C7 condition with files from three runs, so it appears the source is a corrupted INQUEUE file, and since a VERY small number of records are being recycled, the safest solution now would be to eliminate the use of INQUEUE for a few cycles to confirm that was the culprit, and then it can be reintroduced while watching the job logs for MXG error messages, in case the corruption reoccurs. Change 30.159 Support for Microsoft Exchange 2010 incompatible changes VMACNTSM to MSEXCHANGEIS and MSEXCHANGEIS MAILBOX objects. Creates Aug 18, 2012 variables MSIS0001-MSIS0051 MSISCQTH in MSEXCHIS dataset, and in dataset MSEXCHPU variables MSPU0001-MSPU0074 and new per-second rate variables MSGDLVPS MSGRECPS MSGSNTPS MSGSUBPS (whose new per-sec values are also converted to per-minute and stored in the existing per-minute variables MSGDLVPM MSGRECPM MSGSNTPM MSGSUBPM). Thanks to Stan Dylnicki, Royal Bank of Canada, CANADA.
VMXGSUM KEEPERS VAR1 VARS or VAR2 in the //WORK library, which Aug 15, 2012 had no impact on your MXG programs, but as they were also Sep 4, 2012 left behind in the MXG QA job, they were seen as "real" MXG datasets and created DOCVER entries. -New handling of error conditions when interactive: rather than terminating the SAS Session, global macro MXGOPTKILL that is set in VMXGOPTK (the macro that drives the keep logic and is part of VMXGSUM) gracefully shuts down the VMXGSUM execution but preserves the interactive session. -Messages about too-long-text were revised.
UTILEXCL tailoring members now identifies the number of fields and Aug 13, 2012 the lengths that must be updated in those three members. It may be necessary to use the MCTSSDRL field in the DO groups for these three optional segments. Thanks to Scott Barry, SBBWorks Inc, USA. Change 30.156 -Variables TOTLSESS ACTVSESS INACSESS in TERMSERV dataset EXNTICAS for the Terminal Services object were input in that order IMACNTSM in Windows Server 2003 (NTVERSN='5.2'). In Windows Server VMACNTSM 2008-R2 (NTVERSN='6.1') the order was changed to ACTVSESS VMXGINIT INACSESS TOTLSESS, so now, one of two INPUT statements is Aug 10, 2012 executed, based on NTVERSN. -Support for ICA Session object creates ICASESSN dataset. dddddd dataset description NTICAS ICASESSN ICA Session Thanks to ???, ???, ??? Thanks to Mark Friedman, Demand Technology, USA.
EXT1209A nine records creates new TYP1209A dataset IMAC120 dddddd dataset description VMAC120 T1209A TYP1209A WebSphere Subtype 9 ASYNC VMXGINIT These variables in TYP1209A and TYP1209E correspond and Aug 9, 2012 can be used to compare observations in those datasets: Oct 16, 2012 TYP1209A TYP1209E SM1209GP SM1209CG SM1209GQ SM1209CR SM1209GR SM1209CS SM1209GT SM1209CU SM1209GV SM1209CV -Unrelated, variable SM1209CX was found to have been incorrectly divided by 4096, in TYP1209E dataset. -Oct 16: The August change also corrected variables SM1209DL, SM1209DM, and SM1209DN service units. Thanks to Joseph L. Babcock, JPM Chase, USA. Thanks to Tom Bubnash, SSA, USA. Thanks to Babatunde Ashiru, SSA, USA. Thanks to Jim Hare, Shared Services, CANADA. Change 30.154 Velocity Software ZVPS (a/k/a XAM) new User Record with EXXAMUSV with USERTYPE='VCPU' creates new dataset: FORMATS dddddd dataset description IMACXAM XAMUSV XMUSVCPU Virtual CPU per User Statistics VMACXAM Dataset XMUSVCPU contains the USEACT and USEINT variables VMXGINIT plus Virtual CPU Address (VMDCPUAD) and Engine Type in Aug 9, 2012 VMDPUTYP, decoded by new MGXAMTY format (CP, ZIIP, etc), so there is one observation for each USERID and VMDCPUAD for each MONWRITE interval. Thanks to Patricia Hansen, ADP, USA. Thanks to Mike Chaves, ADP, USA.
VMXGUOW -PROGRAM names PROG1-PROGnn are created (TRAN1-TRANnn), Aug 8, 2012 with the HOWDEEP= argument (default 10) setting the "nn" to be created, but you must also use KUOWIDC to list the variables you want to be kept. You can put the statement MACRO _KUOWIDC PROG1=PROG9 % in IMACKEEP tailoring, or use %LET MACKEEP= MACRO _KUOWIDC PROG1-PROG9 %; in SYSIN. -Documentation of CICS transaction counts: An observation in CICSTRAN cay be only one of many CICS "segments" that make up a "unit of work", UOW, a "real" transaction. Counting the number of observations in the CICSTRAN dataset does NOT tell you how many actual CICS Thanks to Charles Savikas, DCF, State of Florida, USA. "transactions", or "units-of-work" were executed. MRO trans typically have at least two observations in the CICSTRAN dataset for a unit of work: one is from the TOR region/applid and one is from the AOR region. In ASUMCICS dataset (summarized CICSTRAN), the variable NUMTRANS is the number of CICSTRAN observations, i.e., the inflated count of MRO CICS segments. In ASUMCICX dataset (summarized ASUMUOW), the variable NUMTRANS is the number of ASUMUOW observations, i.e., the "correct" count of transactions/units-of-work). In ASUMUOW dataset, (created from CICSTRAN/DB2ACCT/MQ): MROTRAN - Count of CICSTRAN observations in this UOW, inflated segment count. DB2TRAN - Count of DB2ACCT observations in this UOW. MQTRAN - Count of MQMACCT/MQMACCTQ obs in this UOW. Thanks to Scott Barry, SBBWorks Inc, USA.
VMACDCOL "KEEP= exit" for dataset DCOLDSET, intending to add some Aug 6, 2012 new variables and to DROP= some existing variables: MACRO _KDCODSN NEWVAR1 NEWVAR2 DROP= OLDVAR1 OLDVAR2 % This compiler error is unique to the VMACDCOL structure, and is circumvented by relocating the &MXGFILNM macro variable to a different place in the source code. -The SAS compiler erroneously expanded that _KDCODSN macro token TWICE, causing the first instance of DROP= to apply to all of the NEW variables in the second instance of the expansion, so NONE of the new variables were kept. This error was precipitated by Change 29.296, which added a new-style macro variable token, &MXGFILNM, at the end of each KEEP= list (so you could add variable INFILENM to all DCOLLECT datasets). Apparently, even when &MXGFILNM is blank, when it is the last token before the _KDCODSN token, then if _KDCODSN is not blank, the SAS old-style macro compiler expanded that macro token TWICE. But, by relocating that &MXGFILNM text to a separate line, before the last variable line in the MACRO _VDCODSN KEEP= list, so it is no longer immediately prior to _KDCODSN token, that second unwanted expansion is avoided and your normal _Kdddddd tailoring works as expected. The error was in both SAS 9.2 and 9.3 on Windows and z/OS SAS 9.1.3 SP4. Thanks to Paul Maradin, HP, USA.
VMACRMFV -ZRBASM file was not output to the PDB when TYPSRMFV Aug 4, 2012 member was used to build an RMF Monitor III PDB. -Macro call to _SZRBASM was missing from _SRMFV macro. -_BZRBASM macro for ZRBASM file misspelled ASMJESID variable as ASIJESID causing SAS error during PDB build. -Change 27.100 altered the EXZRBLCP macro to conserve disk space for inactive logical processors by suppressing output to the ZRBLCP file when LCPUPDTM was zero (which includes LPARs that are categorized as PHYSICAL). However, all processor entries for PHYSICAL LPARs are required to determine the number of physical engines on a CEC for an engine type such as a ZAAP or ZIIP. One or more of these can legitimately be inactive during an RMF III MINTIME interval. The number of physical engines for each engine type is required to compute Physical % UTIL as in the RMF III CPC panel. Entries for PHYSICAL LPARs are now always output regardless of activity. -Variable SSHRMFVN (RMF version number) added to these MXG created RMF III "PDB" SAS datasets: ZRBSVPP, ZRBSVPW, ZRBSVPC, ZRBSVPZ, ZRBSVPR, ZRBSVPG. -Variable SSHSMPNR (number of valid MINTIME samples) added to these RMF III PDB datasets: ZRBGEI, ZRBCFC, ZRBCFI, ZRBCPD, ZRBCPU, ZRBLCP, ZRBCSR, ZRBDVT, ZRBENC, ZRBENT, ZRBSHD, ZRBPGP, ZRBRED, ZRBRCDB, ZRBRCDS, ZRBRCDR, ZRBOPD, ZRBSPG, ZRBSVPP, ZRBSVPW, ZRBSVPC, ZRBSVPZ, ZRBSVPR, ZRBSVPG, ZRBUWDEV, ZRBUWSTO, ZRBUWJES, ZRBUWHSM, ZRBUWENQ, ZRBUWMNT, ZRBUWMSG. -Variable SSHGOMNT (Gatherer MINTIME option) added to these RMF III PDB files: ZRBGEI, ZRBASI, ZRBCFC, ZRBCFI, ZRBCPD, ZRBCPU, ZRBLCP, ZRBCSR, ZRBDVT, ZRBENC, ZRBENT, ZRBSHD, ZRBPGP, ZRBRED, ZRBRCDB, ZRBRCDS, ZRBRCDR, ZRBRCDT, ZRBRCDX, ZRBRCDD, ZRBOPD, ZRBSPG, ZRBSVPP, ZRBSVPW, ZRBSVPC, ZRBSVPZ, ZRBSVPR, ZRBSVPG, ZRBUWDEV, ZRBUWSTO, ZRBUWJES, ZRBUWHSM, ZRBUWENQ, ZRBUWMNT, ZRBUWMSG. Thanks to Randy Hewitt, HP Canada, Canada Change 30.150 NDM-CDI record 'EX' caused "UNKNOWN SUBTYPE" log messages VMACNDM and no DSECT is yet received to decode the 40 bytes of Aug 3, 2012 hex zeros at the end of the record, but the front end has the same fields as the NDMDT record so the 'EX' record is output in NDMDT, pending receipt of documentation. Thanks to Paul Volpi, UHC, USA. Change 30.149 The KEEPALL=YES default is used in most MXG code because VMXGSUM it is usually safer, as it protects if there is an INCODE= Aug 3, 2012 argument that can reference a variable that is not in one Aug 7, 2012 of the parameter lists. However, in some invocations of VMXGSUM, it caused a VARIABLE NOT FOUND error that caused PROC MEANS to fail, when a variable did not exist, due to an error in the VMXGSUM logic that only rebuilt the KEEP list after the first data step when KEEPALL=NO. Now, that rebuild is always executed with either NO or YES. -VMXGSUM was redated Aug 7; the above logic was revised. Thanks to Robert Kirkland, MedMutual, USA.
VMXGINIT MXGSOURC exists ("MXGNAMES" dynamic allocation was used, Aug 1, 2012 which requires FIXORDER dataset to be created to print the //SOURCLIB DSNAMEs messages at MXG startup), so the FIXORDER creation can be bypassed for static allocation. And FIXORDER is not compressed when it is created, to suppress a cosmetic warning messages about size increase.
ANALDB2R in OMEGAMON XE for DB2 Performance Monitor Version 4.2. VFMT102 Since the detail and trace formats are essentially the Jul 31, 2012 same, PMAUD03 now simply invokes PMAUD02. AUDIT= parameter is expanded to support these options: DDFTRAN - DDF TRANSLATION . ESTTRST - Establish Trusted Context CRTTRST - Create Trusted Context KERBERO - KERBEROS/ENCRYPTED Connection
change removes the timestamp when VFMT102 is called. The LINESIZE option is now set to 200 at the beginning and then reset to the original value at the end of ANALDB2R, as some of the reports can get very wide. Thanks to Alyona Bertneski, JPMorgan, USA. Change 30.146 Cosmetic/maybe. The %DO loop variable &I was changed to VMXGDSNL the more "unique" &DSNL to possibly avoid conflict with Jul 31, 2012 user code that was wrapped around VMXGDSNL in TYPEQACS. Thanks to Stephen Hoar, Lloyds Banking, ENGLAND. Change 30.145 Variable CORRNAME and CORRNUM were incorrectly decoded VMACDB2 for QWACATYP=4 (CICS) connections. Jul 31, 2012
READDB2 e.g. IFCIDS=ACCOUNT 105 107 269, the T102Sxxx datasets Jul 31, 2012 were not copied to the PDBOUT= destination. -If you used the LDB2xxx parameters to limit the variables kept or dropped and used the syntax: LDB2ACC=PDB//KEEP/varlist the // did not work as documented; a space was required, but this change now properly supports // or / / as the null argument.
VMXGDBSS incorrectly summed: Jul 31, 2012 ASUMDBSS - QB1TVPL QB2TVPL QB3TVPL QB4TVPL QB1TCBA QB2TCBA QB3TCBA QB4TCBA ASUMDBSB - QBSTVPL QBSTCBA They now contain the MAX value of the end-of-interval values, across each summary interval. -In addition, because DB2 V10 writes fixed one-minute data and so those variables are end-of-interval values, these average value variables are now created: ASUMDBSS - AVG1VPL AVG2VPL AVGTVPL AVGTVPL AVG1CBA AVG2CBA AVGTCBA AVGTCBA ASUMDBSB - AVGTVPL AVGTCBA -Labels were created for these four variables: ASUMDBSS - THRDFTPT THRDUSE TOTTHRD ASUMDBSB - READS Change 30.142 %VMXGPRNT, which "PROC PRINTs" with both the LABEL and VMXGPRNT VARIABLE as heading, is enhanced to allow FMTLST= option Jul 31, 2012 to specify the printed format. See ANAL113 for example.
VMACXAM observations as that is what is in the record. Jul 31, 2012 -Values in AVERAGTM='TOTALTIME/TOTALACKED IN FAL' and the values in TOTALTIM='TOT ROUND TRIP TIME(MS)' in XMTCPAPP dataset are large negative values in many records. -Both problems will be forwarded to the vendor. -Variable NAMENODE is now kept in XMTCPAPP dataset. Thanks to Andrew Petersen, CSC, AUSTRALIA. Change 30.140 -Support for IFCID 269 and 270 populates existing T102S269 FORMATS and T102S270 datasets with QW0269xx/QW0270xx variables. VFMT102 -Formats created for some QW0269xx/QW0270xx variables. VMAC102 -VFMT102 revised to eliminate the time values. Jul 31, 2012 -SQL Text long-variables have multiple blanks removed with COMPBL function. Thanks to Alyona Bertneski, JPMorgan, USA.
VMAC110 now know is a z/OS Unix name for a directory, primarily Jul 30, 2012 used for Web service binding files. Thanks to Dale Slaughter, Transamerica, USA. Change 30.138 ASG TMON/DB2 creates invalid DB2 101 records that causes VMACDB2H MXG to print ERROR: INVALID DB2 101 RECORD messages, but Jul 27, 2012 since the ASG records have their header segment at the Aug 3, 2012 start of the record, whereas IBM's header segment is at the end of the record, I've updated these error messages to identify "ASG-CREATED" and to identify the ASG fix of TE03737 in the error messages. These records are DELETED. Change 30.023 detected these errors, but printed ALL of the defective records; this change limits printing to 3. Thanks to Marty Pruden, Purina Nestle, USA. Change 30.137 Updates to NMON/TOPAS Monitor for AIX and LINUX. EXNMONDF -All BBBP single-value-per-line configuration entries for EXNMONEN the lsconf, lparstat, lsattr EL, vmstat V, and vmstat S EXNMONLP subtypes are now output in NMONBBBP dataset. All labels EXNMONMP now contain the subtype name. Variables from each subtype EXNMONVM are grouped in the LABEL statement so they are adjacent IMACNMON when printed. Variables BBBP001-BBBP0146 now exist. VMACNMON -These "BBBP,ending" multi-value-per-line entries are VMXGINIT output in these new datasets: Jul 28, 2012 DDDDDD DATASET DESCRIPTION Aug 17, 2012 NMONDF NMONBBBPENDDF BBBP,ending df m Sep 13, 2012 NMONLP NMONBBBPENDLP BBBP,ending lparstat h NMONMP NMONBBBPENDMP BBBP,ending mpstat d NMONVM NMONBBBPENDVM BBBP,ending vmstat i -The "BBBP,ending" single-value-per-line entries from the "ending vmstat v" and "ending vmstat s" subtypes are output in this new dataset: DDDDDD DATASET DESCRIPTION NMONDF NMONBBBPENDING BBBP,ending vmstat v and s -Entries PCPUnn and SCPUnn create TEMPCPUNP and TEMPCPUNS that are merged into existing NMONCPUD, detail CPU metrics by CPUNR. -Entries PCPU_ALL and SCPU_ALL_ are input and variables retained and added to NMONINTV. -Entries DISKRIO, DISKWIO, DISKAVGRIO, DISKAVGWIO are now supported, and variables of the same name are added to the NMONDISK dataset. -The internal logic for the DISKxxxxN "suffixed" records, written when there are more disk devices than will fit on a single record, is restructured to use a 2-elment array instead of a separate explicit DO group for each suffix. These arrays support 50 suffixed records with 512 devices per record, versus the original fixed 21 suffixes. -ERROR: ARRAY Subscript Out Of Range - DISKSERV DISKSVCTM DISKWAIT DISKWAITTM objects corrected, and protection for "NANQ" text for "not a number" for BBBPENDLP04, Aug 17. -Sep 13: BBBPENDVM05 increased to 64 bytes.
VMAC50 TY50PCIR TY50PCIV TY50PCIT TY50PCIU TY50PDFR Jul 25, 2012 are missing if SUBCHPOL is not 'READ'; IBM had changed the value to "RD/n" without documentation. The MXG test is now revised to detect READ if SUBCHPOL starts with RD. Thanks to John McLaughlin, Bank of America, USA. Change 30.135 %VGETALOC is a new macro for use with AUTOALOC=YES (that VGETALOC VMXGALOC option creates directory names containing date) Jul 24, 2012 that lets you select (getdaterange=,typeofdata=) a group of "PDBs"/directories that are to be read, constructing a "SET" statement in a DATA-step using VMXGSET.
%vgetaloc(getdaterange=01jul12 24jul12, typeofdata=daily, basedir=c:\mxg, datefmt=date7. ); data jobs; %vmxgset(dataset=jobs); will search the c:\mxg directory for all directories with a name that contains a date in ddmmmyy format, selects those with date between 01Jul12 and 24Jul12 inclusive, and allocates them as PDB1- PDB24 as there are 24 dates in your selection criteria. These libnames are passed to VMXGSET which build the logical SET statement to read all 24 LIBNAMEs for the JOBS dataset you requested in VMXGSET set pdb1.jobs pdb2.jobs pdb3.jobs ... pdb24.jobs; (i.e., the number of libnames found by vgetaloc).
ANALDB2R that you were using a backlevel ANALDB2R, because that Jul 24, 2012 statement at the bottom of ANALDB2R still had 30.02 in both 30.03 and 30.04. While the MXG QA stream has a test for VMXGVERS value, the line was in lower case and the QA test expected upper case. Now, the QA test looks first for lower case and then for version number. Since there is a %PUT with the ANALDB2R last update date, the currency of this ANALDB2R member was confirmed on the SAS log, so this was only a cosmetic error. But it gets this change text because the user had to waste his time to report the potential problem due to my error! Thanks to R. van der Zande, KLM, THE NETHERLANDS.
IMACDBNZ IBM DB2 Analytics Accelerator IDAA product uses Netezza. VMACDB2 -Similar to MXG support for optional CICS data segments, Jul 23, 2012 support for optional Q8AC DB2 Accounting variables is in a comment block in the new IMACDBNZ member that contains the code, labels, and formats, so none of the Netezza variables will exist in DB2ACCT, until you copy IMACDBNZ into your "USERID.SOURCLIB tailoring library and then remove that comment block. -Because the volume of DB2 statistics records is small, the Netezza Q8ST DB2 Statistics variables are created in Yüklə 28,67 Mb. Dostları ilə paylaş:
|