* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
// SPACE=(CYL,(100,100)) DATA PDB.TYPETALO; SET REALPDB.TYPETALO; IF TYPETMNT=5 THEN DELETE; %INCLUDE SOURCLIB(ASUMTALO); PROC COPY INDD=REALPDB OUTDD=TAPECOPY; EXCLUDE ASUMTALO; PROC DELETE DATA=REALPDB._ALL_; PROC COPY INDD=TAPECOPY OUTDD=REALPDB; PROC COPY INDD=PDB OUTDD=REALPDB; SELECT ASUMTALO; Thanks to Jonathan A. McVey, Lowe's Companie, Inc, USA. Thanks to David Zeems, Bank of America, USA. Change 17.241 The Tandem Utility to read "Unstructured" records was UTANDSTR re-commented and set up to convert the five Measure files VMACTAND that MXG supports, and the table of LRECLs for both the Sep 30, 1999 "Normal" and "Unstructured" tandem records in VMACTAND was updated.
ANALDSET nulled out in ANALDSET, causing ERROR 455-185. New lines Sep 29, 1999 XY ADD NAME=EXTY30TD /* override and output no type30td records */ were added to STEPONE. Thanks to Freddie Arie, Lone Star Gas, USA. Change 17.239 Variable ENDDTE was not converted back into a SAS date, VMACXCOM because ENDTIME was created so ENDDTE was redundant, but Sep 28, 1999 since ENDDTE was also kept, it is now a valid SAS date. Thanks to Andrew Nyiri, EDS Australia, AUSTRALIA. Change 17.238 RMFINTRV didn't support non-contiguous shift definitions, RMFINTRV like a three hour peak-shift that includes 10:00-11:59 Sep 28, 1999 and 14:00-14:59, because some TYPE7xD datasets were not resorted after the value of STARTIME was changed by the DURSET macro in member IMACRMFI. Now, all of the TYPE7xD data sets are sorted BY SYSPLEX SYSTEM STARTIME, so as long as your shift definition for this peak shift resets STARTIME to 10:00 as the start of this three hour shift, all three hours will be summarized as expected. Thanks to Jim Quigley, Con Edison, USA. Change 17.237 NETSPY X.25 line stats are in NSPYNPSI dataset instead of VMACNSPY dataset NSPYLINE (only non-X.25 lines are there), and now Sep 28, 1999 the line utilization variables SLNU/PLNU are calculated and kept in NSPYNSPI, for NSPNRECI='04'x (X.25 lines). Variable NSPX25R is no longer kept in NSPYLINE dataset, because NSPYLINE contains no X.25 resources and the existence of that variable was misleading. Thanks to Sharon Hindle, EDS Australia, AUSTRALIA. Change 17.236 Invalid user SMF records written with '40'x for record ID VMAC64 caused TYPE64 code to fail, so protection was added to Sep 27, 1999 prevent the INPUT EXCEEDED RECORD error. The INPUT is now conditionally executed: IF 137+LENEXT+OFFSMF+114 LE LENGTH THEN INPUT @137+LENEXT+OFFSMF Thanks to Fred Kaelber, Pacific Bell, USA.
VMACQAPM changes in LRECL of the new records), but if you change Sep 27, 1999 the LRECLs in your JCL, then MXG 16.02-17.06 "tolerate" the new records: MXG does not fail, reads all of the existing fields, but the new fields will not be input until you install this change, in MXG 17.07 or later). 4.2 4.4 File LRECL LRECL QAPMBUS 57 111 QAPMDISK 346 352 QAPMECL 276 277 QAPMJOBS 812 819 QAPMLIOP 144 150 QAPMMIOP 232 270 QAPMRESP 58 68 QAPMSYS 3250 3268 -Dataset QAPMCONF now has only one observation, with all of the Configuration options in that one observation, instead of a separate observation for each GKEY value. -Dataset QAPMSYS has three new variables, SYSHEAO, SYHFTH and SYHFTS. -Dataset QAPMJOBS has two new variables: JBSVIF JBTFLT. -Dataset QAPMDISK has one new variable: DSCERC
VMACIPAC for the Caller Type, new variable IPCALLER in dataset Sep 26, 1999 IPAC03, compatibly. Thanks to Steve Clark, California Federal Bank, USA. Change 17.233 TRND70 was enhanced to contain OMVSAVG, OMVSMIN and TRND70 OMVSMAX to track number of Open Edition users. Sep 26, 1999 Thanks to Diane Eppestine, Southwestern Bell, USA. Change 17.232 The new PDB.ASUMCEC dataset was created BY SYSPLEX CECSER ASUM70PR but SYSPLEX should not have been used in the summary (a Sep 26, 1999 CEC can have LPARs in different SYSPLEXes) and has been removed. Also, the "PDB" was hardcoded, but it should have been the "&PDBMXG" macro reference (which defaults to "PDB" but permits global destination changes by ITSV). Thanks to Randy Shumate, LEXIS-NEXIS, USA. Thanks to Toby Olberding, Hudson Williams, USA. Change 17.231 The date variables XPMLKDT and XPMXPDT were formatted as VMACSFTA DATE9, but they contained Julian dates and had not been Sep 26, 1999 converted to SAS dates. These lines were inserted after their input so they will now contain valid SAS dates: XPMLKDT=DATEJUL(XPMLKDT); XPMZPDT=DATEJUL(XPMZPDT); The actual code change was more extensive, as it includes the Y2K JULDATE algorithm from member YEAR2000. Thanks to Ove Dall-Hansen, CODAN Insurance A/S, DENMARK.
VMACSTC match the format printed on the STK Silo LSM Info report Sep 26, 1999 but once STK Technical Support told me how they drop the Oct 5, 1999 first nybble and print three character positions with the last nybble of ACS in one hex and both nybbles of LSM in two hex positions, both fields now match the STK utility report format from the STK user SMF record. However, truncated subtype 1, 4, and 7 records that cause INPUT RECORD EXCEEDED RECORD LENGTH, STOPOVER abends were found and are under now protected in VMACSTC. One circumvention for these short records (prior to this change) that eliminates the ABEND and lets MXG read thru all of the SMF data is to use MACRO STOPOVER MISSOVER % after //SYSIN to change MXG's default of STOPOVER to MISSOVER. However, MISSOVER will still OUTPUT each of the bad records, so the MXG datasets will contain unexpected missing values, possibly causing your report programs to die with divide by zero exceptions!
in the IMACFILE exit member, using: IF ID=222 AND SUBTYPE=7 AND LENGTH LT 140 THEN DELETE; IF ID=222 AND SUBTYPE=4 AND LENGTH LT 552 THEN DELETE; IF ID=222 AND SUBTYPE=1 AND LENGTH LT 198 THEN DELETE; Thanks to Jorge Fong, Depository Trust, USA Thanks to Joseph J. Faska, Depository Trust, USA.
IMACEXCL in macro _CICXCOM did not test for SMFPSRVR, so if you Sep 26, 1999 had both _CICXCOM and _CICXCU4 enabled, the _CICXCOM code was executed for the 4.1 region, causing INVALID RECORD DELETED error message and STOPOVER abend. Inserting IF SMFPSRVR LE 3 THEN DO; ... END; inside the MACRO _CICXCOM definition in IMACEXCL allows the old and new version's excludes to coexist. Thanks to Charles McAfee, Cover-all, CANADA.
ASMIMSLG are now implemented as the standard MXG support members. ASMIMSL5 (Those changes were added in new "X" members for testing ASMIMSL6 and with no reported errors, they have now become the MXG EXIMSBMP support for IMS log processing.) In addition, the exit EXIMSTRN members for the IMSTRAN, IMSBMPS, and IMSFASTP datasets JCLIMSLG are now used (i.e., the _EIMSxxx macro did not call the JCLIMSL5 %INCLUDE SOURCLIB(EXIMSxxx); ) so that the MXG exit JCLIMSL6 facility can be exploited by ITSV. TYPEIMSA
TYPEIMSB I strongly recommend that you install the new IMS support VMACIMSA and use it to reassemble the ASMIMSLx member to re-create Sep 25, 1999 the MXGIMSLn program, which is then used in the JCLIMSLx JCL example. Using these new members will correct the SubQ 6 times as well as provide the many enhancements to IMS log processing described in member ADOCIMXA. Change 17.227 The SAP IMS log record 'AE'x was NOT Y2K Ready until SAP VMACIMSA Release 5.0i, which (INCOMPATIBLY) changed the date from VMACIMS 0DDMMYYF to YYYYMMDD format. This change supports the YEAR2000 new format, adds Y2K protection for their old YY format, Sep 25, 1999 and also corrects an MXG error in the HH/MM/SS fields that could cause the SAPTIME date to be a day later than the actual date (those fields are actually PK format, but they were input with PIB format, so an HH of '18'x or larger became 25 hours, and bumped the date when HH was added to create SAPTIME. Thanks to Alain Kerninon, Generali AG, AUSTRIA. Thanks to Ottfried Dorok, ICG Informationssys, GERMANY. Change 17.226 An expiration date of 2099366 caused INVALID VALUE FOR VMACDCOL DATEJUL FUNCTION and UMEXPDT or DCDEXPDT were missing, Sep 18, 1999 although ORGEXPDT still contained the 2099366 value. Since 2099366 itself is an invalid value (2099 is not a leap year), MXG's algorithm for invalid dates (which is to set the calculated expiration date to 31Dec2099 so that any calculation on days-to-expire will be large) is now extended to protect the invalid date of 2099366, by inserting (YREXPDT=199 AND DAYEXPDT GE 365) OR after the (YREXPDT=99 AND DAYEXPDT GE 365) OR statement in two places. Thanks to Harald Seifert, HUK-Coburg, GERMANY. Thanks to J. Hohring, HUK-Coburg, GERMANY. Change 17.225 SAS/GRAPH examples are now updated to report GRAFWORK BY SYSPLEX SYSTEM; instead of just Sep 18, 1999 BY SYSTEM; Thanks to Tom Kelman, Sun Trust, USA. Change 17.224 Member IMACKEEP was not included in TYPETAND, but it now TYPETAND has been added after the include of member VMACTAND. Sep 17, 1999 Thanks to Steve Hurley, National Westminster Bank, ENGLAND. Change 17.223 Report example fails if there were more than one SYSTEMs ANAL42 data, because the PROC SORT DATA=JIM2 OUT=JIM; and DSD Sep 15, 1999 BY DESCENDING IOCOUNT; should have been BY SYSTE DESCENDING IOCOUNT; Thanks to Freddie Arie, Lone Star Gas, USA.
FORMATS 3000 Enterprise Server's new EIO (Emulated I/O) and DSD Sep 15, 1999 (Direct System Device) type of CHPIDs. MXG Format named $MG073CD decodes the CHPID type and was updated for these two new entries, plus additional entries that were in the complete list (which should have been in the APAR text, but was not printed there).
VMACNTSM STARTIME missing because PREVWEBS=SYSTEM should have been Aug 27, 1999 PREVWEBS=WEBSERV. Additional variables are deaccumulated and minor other corrections were made. Thanks to Jim Quigley, CONED, USA.
VMAC110 JCRLL=0 MXG error message, because MXG expected the new Aug 27, 1999 Journal Format SMF records in CICS 5.2 instead of 5.1! To correct, find the two IF statement with comments that contain "Journal Format" and change them to now read: IF SMFPSRVR LT 51 THEN DO /* PRE 5.1 JOURNAL FORMAT */ ELSE IF SMFPSRVR GE 51 THEN DO; /*POST 5.1 JOURNAL FORMAT*/ Thanks to Andre van de Riet, Hudson Williams Europe, THE NETHERLANDS. Change 17.219 SAS ITSV CPDUPUPD fails with APPARENT MACRO error because DIFFTPX an old statement: OPTIONS MAUTOSOURCE SASAUTOS=SOURCLIB; DIFFROSC was still in these members. The SASAUTOS was changed to TYPEDOS OPTIONS MAUTOSOURCE SASAUTOS=(SOURCLIB SASAUTOS); back in TYPEIMS Change 16.066, and since they are set in member CONFIG, TYPEIMS1 those old statements in these members were deleted so as Aug 23, 1999 to eliminate the error. Thanks to Jens Mohring, HUK Coburg, GERMANY. Thanks to Harald Seifert, HUK Coburg, GERMANY.
VMAC80 the need for a specific test to recognize Top Secret type VMAC80A 80 records from the standard IBM type 80 record). To fix Aug 23, 1999 MXG to support this release, you need only to add OR RACFVRSN=051X to the existing test for Top Secret Versions. Thanks to Peter Vines, THE University of Virginia, USA.
VMAC94 of invoking the _ETY94 substitution macro, so overriding Aug 16, 1999 the _ETY94 did not work. Now it does. Thanks to Tom Parker, CSC/Hogan, USA. ======Changes thru 17.216 were in MXG 17.06 dated Aug 12, 1999====== Change 17.216 To be Y2K Ready, ASMTAPES must be assembled with "ES6" VMACTMNT when the MXGTMNT Tape Mount Monitor program is assembled Aug 12, 1999 for under OS/390 R1.3 or later. Your old MXGTMNT program that had been assembled with "ES5" or "ESA" still ran on OS/390 R1.3, so there was no error to alert you that the "ES6" was now required, but if you are still running the old MXGTMNT after January, 2000, the date in the INITTIME field will still be 1900! IBM moved that date from a 3 to a 4 byte field that is only seen by "ES6". ASMTAPES documents that "ES6" is for OS/390 R1.3, but I had not previously highlighted that requirement. This change protects VMACTMNT; if 1900 records are still being written at your site in Jan, 2000, the variable INITTIME in TYPETMNT and TYPETALO will be correct. Thanks to Steve Donahue, Blue Cross and Blue Shield of Texas, USA.
ANALSTOR variable R791ESCT in Change 8.167, but the report was not Aug 10, 1999 updated until now. Thanks to Tony P. Steward, Post Office, ENGLAND. Change 17.214 OS/390 R2.8 is already supported in MXG, as there were several only minor, compatible changes, most of which have been Aug 10, 1999 released as APARS to the type 89, 92, and 94 SMF records. MXG Versions 16.09 or later can process the R2.8 records. Change 17.213 Support for the Import/Export Statistics section in SMF VMAC94 type 94 adds new variables to dataset TYPE94. Aug 10, 1999
VMAC110 of A17DT was hex '00' in the record, but MXG's statement Aug 8, 1999 A17DT=TRANSLATE(A17DT,' ','00'x) was incorrectly changing the '00'x to a blank, causing the two 40-byte sections to be skipped. Delete that line and these records will be properly decoded in to the CICFCR statistics dataset. Thanks to Bruce Widlund, Merrill Consultants, USA.
VMAC74 in only one record per SYSPLEX, and you can't pick which Aug 8, 1999 SYSTEM's record has the info, accoring to this IBM reply: To avoid collection of redundant data within a sysplex, RMF selects one system as a 'master system'. This selection is RMF internal, there is no option to select the master system. For this 'master system' all CF data, SMF 74-4 data, are collected. For the other systems within a sysplex, the redundant but expensive CF-data are omitted (expensive in terms of CPU utilization). That explains why we see the 'Connectivity Data Section' and 'Structure Data Section' only for one system within a sysplex. For more details, please see apar OW31699, as this was introduced with this APAR. Thanks to Dennis Wong, New York Life Insurance, USA. Change 17.210 -PDB.ASUMTAPE can have the wrong JOB/READTIME/INITTIME for ASUMTAPE some jobs from TYPETMNT, but these events are detected Aug 8, 1999 and the TYPETALO information is used to identify the job that caused the mount. These mismatched events occur only for jobs that went thru allocation recovery and are usually few in number (5 of 2500). -PDB.ASUMTAPE can have small negative values for TAPMTDTM because the TENDTIME is the sampled MXGTMNT end of mount, but TY21TIME is the exact dismount timestamp. The value will be less than two seconds (the default MXGTMNT sample rate) so to avoid negative numbers for the duration a a tape is mounted, these negatives are now set to zero. Thanks to Gene Rahe, First American, USA. Thanks to Judy Arroyo, Summit Bank, USA. Change 17.209 Support for NTSMF: Lotus Notes, SMTPDS & SMTPRS objects: EXNTLNAG Object Dataset Variables EXNTLNCA Lotus.Notes.Agent LONOAGNT 8 EXNTLNDB Lotus.Notes.Calendar LONOCALN 6 EXNTLNDI Lotus.Notes.Database LONODB 21 EXNTLNDV Lotus.Notes.Disk LONODISK 2 EXNTLNDO Lotus.Notes.Disk.Inventory LONODINV 1 EXNTLNMA Lotus.Notes.Domino LONODOMI 66 EXNTLNME Lotus.Notes.Mail LONOMAIL 16
EXNTLNNL Lotus.Notes.Network LONONETW 11 EXNTLNRE Lotus.Notes.Network.Log LONONETL 2 EXNTLNSE Lotus.Notes.Replica LONOREPL 5 EXNTLNWE Lotus.Notes.Server LONOSERV 16 VMACNTSM Lotus.Notes.Web LONOWEB 39 VMXGINIT SMTPDS SMTPDS 4 Aug 7, 1999 SMTPRS SMTPRS 4 -NTSMF 2.1.1 could write an interval begin (0,0) with the DURATM=0, which caused STARTIME to be missing. While now fixed, this change also protects that case and uses the previous interval to construct STARTIME. Change 17.208 The RECFM=F and LRECL=256/512 were removed from the VMACSPMG INFILE statements so that the DCB attributes are read Aug 6, 1999 from the dataset. SpaceManager files can be F or FB. Thanks to Anke Mineur, dvg Hannover, GERMANY. ======Changes thru 17.207 were in MXG 17.05 dated Aug 5, 1999====== Change 17.207 Microsoft SP4 and HotFixes have changed the version text VMACNTSM from "Service Pack 3" to "Service Pack 4, RC 1.2", which Aug 5, 1999 was passed by NTSMF into its output comma delimited file, so an extra new field was inadvertently inserted in this configuration record. MXG now looks for that syntax and appends the "RC 1.2" to the NTVERSN. NTSMF will protect the field in a future version to blank the comma. Thanks to Norbert Ravarani, European Commission, LUXEMBURG. Change 17.206 Member IMACJBCK (the "job check" exit to select SMF IMACJBCK job-related records by JOB/READTIME/etc) was not changed VMXGINIT to support dynamic changes, but by defining MACJBCK in Aug 5, 1999 VMXGINIT and locating &MACJBCK; at the end of IMACJBCK you can now use %LET MACJBCK= ... ; syntax instream instead of editing your tailoring into IMACJBCK member. Thanks to Dr. Alexander Raeder, Karstadt AG, GERMANY.
VMXGHSM and bytes physically written, are now created in dataset Aug 5, 1999 MCT from the HSM catalogs. Many sites now use DCOLLECT rather than reading the catalogs directly with MXG. Thanks to Dave Jackson, Capital One Financial Services, USA.
VMACCIMS truncating timestamps and record number. Change the Aug 5, 1999 both DEFAULT=4 to DEFAULT=8. Thanks to Alan Green, Zurich Co., ENGLAND. Thanks to Brian Sznger, Zurich Co, ENGLAND.
ASUM70PR Complete=Yes (which makes sense: to reduce overhead when Aug 4, 1999 there's nothing to share), but MXG's PDB.ASUM70PR reports 100% CPU busy in that LPAR (LCPUPDTM, partition dispatch time is 100% in a Wait Complete=YES LPAR) because MXG did not see this before! The immediate fix to your ASUM70PR member is to insert this DO group in the INCODE= section, immediately after the "LPARDUR=DURATM;" statement: IF LCPUDED='Y' AND LCPUWAIT='Y' THEN DO; IF PCTCPUBY GT . THEN DO; LCPUPDTM=LPARCPUS*PCTCPUBY*DURATM/100; LCPUEDTM=LCPUPDTM; END; ELSE DO;
LCPUPDTM=.; LCPUEDTM=.; END; END;
The PCTCPUBY that is tested is MVS CPU busy for Dedicated CPUs, but it exists only in the SMF record from the MVS System running in that LPAR. In the ASUM70PR observation from other LPARs in this CEC (Central Electronic Complex) the dedicated LPAR CPU busy measures will be missing, as they simply are unknown to the other LPARs. If you have only one LPAR in a CEC, then there is only one SYSTEM and the ASUM70PR data will be valid with the above fix. However, this error is an opportunity to design a better solution to measuring CEC/LPAR CPU busy. The ASUM70PR dataset is created per SYSTEM, but what you need is now the new PDB.ASUMCEC dataset that summarizes TYPE70PR data by the CECSER (CPU Serial Number of the CEC/CPC), so the hardware platform is measured by PDB.ASUMCEC, with true CPU busy for each of the LPARs running on that CEC. The structure of PDB.ASUMCEC has the same variable names (except for SYSTEM, which doesn't have any meaning here), so your existing reporting should require minimal change. But MXG has to read all of the type 70 records from every MVS system that executes in an LPAR with Dedicated CPU to create true CPU measurements; you cannot measure CPU busy in a Dedicated-CPU-LPAR from any other LPAR's SMF record; if MXG does not see the true CPU busy record, the CPU for those dedicated engines will be missing values and the total CPU busy will be wrong. Check your EXTY70PR exit to see if there was user code added to output records from only one system; delete that logic if you find it, so you output all records. The bottom line is that PDB.ASUMCEC should have been the dataset created long ago, but now it should correct the existing error and give you a single record per box to measure total and per-LPAR CPU utilizations. Thanks to Brandan Hock Kwan Wong, DBS, Singapore. Thanks to Chee Hwee Chua, DBS, Singapore. Change 17.202 This example report program processes VTS SMF type 94 ADOC94 records and replicates IBM PGM=VTSSTATS reports. The ANAL94 report is documented in member ADOC94. Remember to use Aug 2, 1999 only one system's data to report on a library box. Thanks to Bruce Widlund, Merrill Consultants, USA.
VMAC92 in your type 92 OMVS/USS file records; that field is now Aug 2, 1999 input if present. The APAR also adds two subtypes to SMF Aug 10, 1999 type 94 that create two new MXG datasets for OMVS/USS: Subtype Dataset Description Yüklə 28,67 Mb. Dostları ilə paylaş:
|