* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
so that subtype 5 records are deleted, so you could use the new VMACTMNT with the (renamed) member ZSUMTALO in case you have problems. Check the counts closely.
ASUMTALO SIMULTANEOUSLY - IF YOU HAVE A TAILORED ASUMTALO MEMBER IN YOUR USERID.SOURCLIB, YOU MUST RETROFIT YOUR TAILORING INTO THE 15.01 ASUMTALO. (Or use the ZSUMTALO as described above).
determining which system's SMF had the earliest cutoff of SMF data (earliest last datetime). Data is "spun" to SPIN.SPINTALO if it is after the cutoff or if it is the last observation for an allocation event and it is also an interval record (we have to adjust the start time of the real allocation record during the next day's ASUMTALO so that we don't count the same drive-interval twice. Thanks to Chuck Hopf, MBNA, USA.
VMXGTAPE MISSING OR OUT OF RANGE when the length of the DSN= macro May 3, 1997 argument was three characters or less, but now will work with any length DDNAME. (VMXGTAPE is a %MACRO to find out if a DDNAME or SAS DATASET is on tape.)
ANALRMFR -Updates were made to logic for IOQU reports May 3, 1997 -Incorrect paging, added array IOPI(9) IOPINSTL, added RETAIN for IOP, ACTIVITY RATE and AVG Q LNGTH variables during MERGE. -MXGCHAN, changed PS=&PGS back to PS=66, because with PS=&PGS and N=PS (to arrange the contents of an entire printed page) does not work correctly. -MXGSMRY moved SWAPRATE=. and PRT=. to SUMRPT dataset to eliminate "MISSING VALUES" messages. Change 15.074 Support for Altai's ZARA Tape Management Release 1.2. TYPEZARA Well, almost. I have created all of the new variables VMACZARA in datasets ZARADSN and ZARAVOL, an enhanced TYPEZARA's May 3, 1997 back-end logic to also keep the new variables, but I do not have the DSECT for the header yet, nor do I have any test data yet, to the new code is in a never-execute DO group (IF RECTYP= -99 THEN DO;) Check for an update to this change when actual support has been tested. Note in this rare instance, the documentation is done before the code changes are complete! Changes to ZARADSN dataset: Variables added: FILABEND FILCONT FILCRSID FILENEXT FILEXPIR FILRCFM FILTPSTK
FILRECFM was replaced by FILRCFM, now is CHARACTER FILFLAG1 was replaced by FILABEND FILCONT FILEXPIR Changes to ZARAVOL dataset: Variables added: VOLAPERR VOLATRER VOLATWER VOLCBLVL VOLCHECK VOLCLEN VOLCREAT VOLFILEA VOLLABEL VOLLOCK VOLSTAT1 VOLSTAT2 Variables no longer created in Version 1.2: VOLSECUR VOLAPPID VOLGRPID VOLOBOX VOLFLG1 VOLFLG2 VOLPOOL VOLLABL was replaced by VOLLABEL, now is CHARACTER Thanks to ???, Nissan Motor Manufacturing (UK) Limited, ENGLAND. Change 15.073 Support for Virtual Tape Server additions to SMF type 94 VMAC94 provide an excellent set of measures for VTS physical and May 2, 1997 logical volume activity, including counts of mounts, by type, lots of min/max/avg for times to mount physical/ logical volumes, time virtual drives were mounted, the number of concurrent physical/virtual drives mounted, and bytes read/written to host channel/from tape devices!!
VMACILKA Interlink has always been wrong. The "FTPCRHS3" in the May 2, 1997 create for FTPCLHST should have been "FPTCLHS3". Thanks to Chris Weston, SAS Institute Cary, USA. Change 15.071 Support for MEMO subtype 8 record creates new MEMODIST EXTYMEMD dataset; subtype 8 records are written by MEMO during a VMACMEMO distribution by an internal process. May 2, 1997 Thanks to Thomas Peiper, Enator, SWEDEN.
DIFFDB2 datasets will have lots of negative values, and there are May 2, 1997 two observations with QWHSISEQ=1 created for each real startup. You can delete both the good and the bad start up record with IF QWHSISEQ=1 THEN DELETE; to clean up your reports quickly, but the logic error fixed by this change applied to each of the Stats datasets and involved creating an OUTFLAG=1 before the DO group for the first OUTPUT, then setting OUTFLAG=0 inside that DO group, and then adding AND OUTFLAG=1 to the subsequent IF QWHSISEQ test for the second OUTPUT. This error was introduced by Change 14.231. Thanks to Trevor Nightingale, SAS Institute Cary, USA.
VMACNSPY frequently missing when it should have had value. Apr 30, 1997 Variable LUNRSPSS was to be INPUT if NSPYENTL GE 112, but the NETSPY 4.4 record had NSPYENTL=108, so LUNRSPSS was not INPUT and hence missing, and LUNRSPSS is used to calculate ARSPHOST. The existing test IF NSPYENTL GE 112 must be change to IF NSPYENTL GE 108 and then between the INPUT lines for LUNRSPSS and LUATTCHS, insert: @; IF NSPYENTL GE 112 THEN INPUT to always INPUT thru LUNRSPSS. MXG was unaware of the shorter length segment. This error was noticed moving from MXG 13.02 to 14.14 with NetSpy 4.4; I do not think there is a problem with 4.5 and later, as I think its segments are always NSPYENTL=112 or greater. Thanks to Mark Zelden, Newark Electronics,USA. Change 15.068 Revised support for HP's MeasureWare for Sun systems has ADOCMWSU restructured the code to match the default REPORT macro, VMACMWSU and was validated with data. Examples of moving data via Apr 29, 1997 ftp are provided in ADOCMWSU's comments. Thanks to Steve Corson, Cincinatti Bell Information Service, USA. Thanks to Gary Alexander, BMC, USA. Change 15.067 Support for NETSPY Version 5.0 is included in MXG 14.14, VMACNSPY (there was no real change between NETSPY 4.7 and 5.0, and Apr 29, 1997 4.7 support was added in MXG 14.03). The DSECT shows fields added compatibly to the type 'X' (Alert) record, but this is not a primary subtype, so until I have test data (and a site that needs the new data in this fairly obscure subtype) I won't add the new fields.
VMAC7072 be TRANSEXC (transactions completing "execution phase" Apr 29, 1997 during the interval) instead of TRANS (transactions that completed during the interval). Thanks to Don Deese, (CPExpert), Computer Management Sciences, USA.
VMACEXC2 for devices with 4-digit UCB addresses of 8000x or higher Apr 29, 1997 were put in variables EXCPMSS/IOTMMSS (instead of the EXCP3390 or EXCP3480 device-type buckets), and these IOs were also not included in EXCPDASD or EXCPTAPE, although they were counted in EXCPTODD and EXCPTOTL variables. The old logic for the now defunct Mass Storage System, MSS has now been commented out, but it tested the first bit of DEVNR to identify MSS devices, causing this error.
VMAC71 the percentage of local page dataset slots that are in Apr 28, 1997 use. Find the maximum value of SLOTUTIL during the month to make sure you have enough page dataset slots defined. SLOTUTIL should always be less than 25% (because the ASM's contiguous slot allocation algorithm can move 30 pages in one SSCH only when there are 30 contiguous free slots, and at utilizations above 25%, ASM begins to not find 30 slots, so it tries to find 15, then 8... which causes lots of extra SSCHs to your page volumes at the very worst possible time - those few times when paging becomes a performance problem!). I have preached this concept, but had not provided the variable (and the value I used in class turns out to need to be changed): SLOTUTIL=(SLOTLOMN-SLOTUNMN)/SLOTLOMN; compares the minimum number of defined local slots with the minimum number of unused local slots to calculate the maximum utilization of slots during the interval. Thanks to Jean Quinkert, Inland Steel Company, USA.
EXTY30OM The variable OMVSEXNP (OMVS Program Name, "grep", "find", VMAC30 etc) is now 16 instead of 8 bytes. Apr 27, 1997 TYPETASK='A ', originally for APPC/ASCH type 30 records (other TYPETASK values are JOB/TSU/STC, taken from the JCTJOBID), is now also used for OMVS type 30 records, so there is no direct way to separate OMVS from APPC/ASCH in type 30 records. However, Goal Mode records contain WLMNAME and SRVCLASS names, which may be useful along with TYPETASK to group type 30 records.
segment, and there can be segments in interval subtypes 2 and 3, in step termination subtype 4, and in job term subtype 5, but there is only one slot per record for the OMVSEXNP Program Name, so the subtype 5 observations are useless for analysis. Here's what I observed: SUBTYPE INIT TERM STEPNAME SUBSTEP OMVSEGNR OMVSEXNP 3 59:02.16 02.29 STEP1 0 1 find 4 59:02.16 02.30 STEP1 0 1 find 3 59:02.30 02.51 *OMVSEX 1 1 grep 4 59:02.30 02.51 *OMVSEX 1 1 grep 5 59:02.16 02.52 0 1 5 59:02.16 02.52 0 2
step records. The single subtype 5 job record has two OMVS segments, representing the two OMVS invocations (one per each step, in this instance), but the OMVS segments in subtype 5 cannot be associated with their program name. As a result, I no longer output obs in TYPE30OM from subtype 5 records (although you can change my new default in member EXTY30OM). Change 15.062 Analysis of impact of DDCONS(NO) versus DDCONS(YES) with ANALDDCN regard to the number of duplicate DDs that are written to Apr 27, 1997 SMF when the (recommended) DDCONS(NO) option is specified. This analysis proves how little duplicates there are and refutes arguments that lots of additional SMF records will be created. See MVS Technical Note in Newsletter 32.
VMAC78 incorrect. The calculation of PCTDIRPT was changed: Apr 26, 1997 Variable R783DPB is now INPUT where PCTDIRPT was, and May 3, 1997 now the calculation of PCTDIRPT is changed to IF (CHPIDTKN+R783DPB+PCTCUBSY) GT 0 THEN PCTDIRPT=100*R783DPB/(CHPIDTKN+R783DPB+PCTCUBSY); ELSE PCTDIRPT=.; and the calculation of PCTCUBSY was changed to: IF SUM(CHPIDTKN,R783DPB,PCTCUBSY) GT 0 THEN PCTCUBSY=100*PCTCUBSY/SUM(CHPIDTKN,R783DPB,PCTCUBSY); ELSE PCTCUBSY=.; The SUM() statement is needed for PCTCUBSY calculation to protect for missing values (old versions of RMF do not have all three variables), but is not needed for PCTDIRPT because that calculation is inside loops which guarantee the existence of the three variables. Thanks to Thomas Neurauter, Fidelity Systems, USA.
VMAC110 DURATM and STARTIME were always missing, but now both Apr 26, 1997 variables are created in CICSEXCE for consistency. Thanks to Trevor Nightingale, SAS Institute Cary, USA. Change 15.059 If a MIM record segment contained MIMCNT=0, Version 3 VMACMIM segments after the zero were not read, because the line Apr 26, 1997 IF MIMCNT = 0 THEN DELETE; deleted the record. For the Version 4 record, that statement had been commented out. This change puts IF MIMCNT GT 0 THEN DO; ... END; around both of the OUTPUT statements for both versions, so only records with samples are output, but so that a zero segment doesn't stop the scan for other segments. Thanks to David L. Dittmar, IBM Global Services, USA.
FORMATS labeled in dataset TYPE74CA; it is the Cache interval VMAC74 and is used in all rate calculations (instead of the Apr 25, 1997 RMF variable DURATM). The IF CSCSS1='...1....'B line misspelled CSCSDFM0='N', should be CSCSDFM1='N'. The variables CSDEVPI and DEVPI have been incorrectly decoded by format MGCACPI, which was replaced by MGCACPD. New variables CSDPIDAT/CSDPIFAI/DEVPIDAT/DEVPIFAI are now replacements for CSDEVPI and DEVPI. Format values for MGCACRC format (for variable RCOL) were enhanced; this field/format describes record level caching mode. Thanks to Bruce Widlund, Merrill Consultants, USA. Change 15.057 New RACF events are now decoded by format MG80EV, with FORMATS new Open Edition events, process actions, etc. It is Apr 25, 1997 worth your while to read the list of possible events! Thanks to Jens Schlatter, Independent Consultant, GERMANY. Change 15.056 This utility is now a %MACRO which can be invoked with: UTILCONT %UTILCONT(PDB=WORK); Apr 25, 1997 to show the size (Megabytes, Observations, etc) of each SAS dataset in the PDB= Data Library, with one line per dataset. And it now works under Windows as well as MVS! It still requires an //INSTREAM or FILENAME INSTREAM to write to and read from (ok, it's crude, but it works, especially when I have filled my WORK file and need to see which datasets are big and which are small). This is a structural revision of the earlier UTILCONT. Thanks to Vickie Drake, Levi Strauss & Co.
ASUM70PR -ASUM70PR. In the calculation of TOTSHARE=SUM(LP0SHARE,.. CICINTRV variable LPESHARE was repeated; the first instance of DIFFTMDB LPESHARE should be LPDSHARE. VMAC42 -VMACQAPM. Variable OS2CT16 should have been divided by VMACQAPM 1000 and formatted TIME12.3, just like variable OS2CT14, VMACTMDB and now it is. Variables ASBRCV ASBTRN LSBRCV LSBTRN Apr 25, 1997 SDBRCV SDBREX SDBXMT SYBRCV SYBREX SYBXMT are now formatted with MGBYTES. Variables ending with PWT in QAPMSNA are formatted TIME12.3 now. -VMAC42. Variables SMF42JNC and SMF42JND are now length 8 with format DATETIME21.2, and variable SMF42JNE is format TIME12.3. -VMACTMDB. Numerous previously unlabelled variables are now labeled, and several DA9STCTx variables that were not kept in TMDBDA2 dataset, now are. -DIFFTMDB. Datasets TMDBDAB, TMDBDA2 and TMDBDAF are now SORTed and Deaccumulated to create DURATM and ENDTIME variables for these interval datasets. -CICINTRV. The first statement A06TEOT=DIFTEOE should be A06TEOE=DIFTEOE. The VMXGSUM for INDATA=SRTDLIR had INVOKEBY comment of INTDBU instead of INTDLIR. Variable A10IDQN appeared twice in KEEP= list for SORT of CICDQR. Variable A17DSTSW was misspelled as A17DTSW. Thanks to Chris Weston, SAS Cary, USA.
TYPETMON only one byte as a result of Change 14.342. Statement: Apr 25, 1997 IF SYSTEM=' ' AND SYSID=' ' AND TMMDREC='TA' .... inserted by that change must be changed to read: IF SYSTEM=' ' AND SYSID=' ' AND TMMDREC='TA' .... (i.e., four blanks instead of one blank for SYSTEM & SYSID). The inserted statement became the first assignment of the two variables, so SAS used it to set their length, in place of the later INPUT statements that had previously defined the variable's lengths. Thanks to Henry Schreiber, Amdahl, USA.
SMFPRM00 that are specified in member SMFPRM00 of SYS1.PARMLIB. Apr 24, 1997 Extensive discussion of why options are recommended. Will be expanded into discussion of to synchronize or not to synchronize to DDCONS(YES) or to DDCONS(NO) system delays during the SMF interval pop. Thanks to Henry Pomorski, Harris County, USA. Thanks to Robert Rudd, Harris County, USA. Thanks to Cheryl Watson Walker, Watson & Walker, USA. Change 15.052 Support for all AS/400 OS/400 Release 3.7.0 records. MXG VMACQAPM had earlier supported only the critical records (CONF, Apr 22, 1997 SYS,JOBS, and DISK), and had coded many new variables Apr 25, 1997 added to the end of the records, but the incompatible insertion of IOPRN (IOP RESOURCE NAME) was overlooked until hit with test records. These datasets were incompatibly changed between 3.6 and 3.7 by the insertion of IOPRN: QAPMHDLC,ASYN,BSC,X25,ECL,STNL,ETH,STNE,DDI, DDI,STND,FRLY,STNY,CIOP,DIOP,LIOP,MIOP,RESP,RWS,LAPD,IDLC and QAPMIOPD. In addition, IBM incompatibly changed the field lengths in QAPMPOOL, and fields were removed from QAPMRESP, QAPMX25, & QAPMRWS. Existing variables named CIDMAO/CIDMAI/DIDMAO/DIDMAI are now named CIKBY0/CIKBYI/ DIKBY0/DIKBYI, but MXG kept the original names and uses conversion from KBytes back to bytes so MGBYTES format will print same magnitudes before and after 3.7.0! Thanks to Len Marchant, Coca Cola Enterprises, USA. Thanks to Cheryl Howard, NationsBank, USA. Change 15.051 Change 14.321 (CF Structures) added trap if number of QO VMAC74 and SO sections were not equal (trap prints MXG DISCOVERY Apr 21, 1997 NO STRUCTURE MATCH on log), but trap springs if there was no QO section at all. While unexpected, I have added protect for SMF744QN=0 while we talk with IBM to find out why the record has Request Data sections but had no corresponding Structure Data Section. Thanks to Joe Schwartz, Cigna, USA. Change 15.050 Protection for products that do not support year 2000 has YEAR2000 been added in MXG Software. Windowing (00-59=20yy, 60-99 Apr 20, 1997 =19yy) has been added during INPUT of JULDATE fields if the century bit is not enabled (i.e., where the record vendor does not support the year 2000, MXG can still do the right thing and create the correct year 2000 date. A more extensive note and update to member YEAR2000 will be written and this text will be revised. As of May 5, JULDATEs have been windowed, but there are other forms to be protected, and member YEAR2000 has not yet been updated to identify the new category of products that are now "MXG-protected". Change 15.049 TIC_UTIL is now added to NPMLANOD and NPMINTRI datasets. VMAC28 (It was calculated but not kept in NPMINTRI, and the Apr 17, 1997 units of time-per-byte were corrected to nanoseconds in NPMLANOD). Thanks to Ian Hindshaw, Standard Bank of South Africa, SOUTH AFRICA.
IMACPDB are added to dataset PDB.PRINT, as they are often used to Apr 17, 1997 determine print characteristics in analysis. They were always captured in TYPE6 dataset, but were not in the list of variables (in IMACPDB) to be kept into PDB.PRINT. Thanks to Jerry Maier, First Chicago NBD, USA. Change 15.047 ML-13 of ASMTAPES contains enhancements to provide error ASMTAPES recovery for the mount monitor's AR code. ML-13 will not Apr 17, 1997 correct these problems, but it will recover from them so the monitor stays up. In addition, when the monitor is run in "DEBUG" mode, an ABEND in ML-13 will generate an SVC dump of both address spaces (monitor and monitored) which will hopefully provide enough diagnostics data that the next iteration will eliminate these errors. In addition, there are two other fixes: ML-13 prevents S0C4 ABENDS when the TCBTIO is zero, and ML-13 corrects the ALESERV return code in the TMNT008I message. Thanks to Ruth Whitney, CITICORP, USA. Thanks to Tom Parker, Hogan Systems, USA. Thanks to Will Evans, Consolidated Freight, USA. Change 15.046 Change 14.345 added code, but the 4DIGITUCB comment was FMXGUCBL coded into column 72, causing an ASSEMBLY errors such as Apr 13, 1997 BEGIN-TO-CONTINUE COLUMNS NOT BLANK. Move left one col. Thanks to Dr. Alexander Raeder, Karstadt AG, GERMANY. Thanks to Harmuth Beckmann, Karstadt AG, Germany
DOC DATETIME20., so as to display all four digits of years. YEAR2000 I also observed that truncated DATETIME formats will not Apr 13, 1997 display four-digit year for the date, hour, minute or second truncations - I hoped DATETIME9 could be enhanced to show the yyyy, but as it now prints seven positions, and increase could cause existing applications to fail (the wider field would shift right or overlay), so SAS Institute cannot make the change.
01FEB84 7-8-9 date,daily 01FEB84:22 10-11-12 hourly 01FEB84:22:33 13-14-15 minutely 01FEB84:22:33:59 16-17-18 secondly 01FEB1984:22:33:59 19-20-21-22 secondly,yyyy Thanks to Bruce Widlund, Merrill Consultants, USA.
FORMATS but now it does into 08X:SDSA, 09X:ESDSA, and 0AXRDSA. Apr 13, 1997 Thanks to Helmut Roese, COM Gmbh, GERMANY. Change 15.043 Change 15.037 changed TYPE116 variable QWHCTNO from VMAC116 &PIB.4. to &NUM.4., but data values that are not EBCDIC Apr 13, 1997 numbers (eg., 004DF3D4x), so the input is changed back back to &PIB.4., is now formatted as HEX8., and is now kept as LENGTH 5 (to display all four bytes numeric, an extra byte must be kept). I could have changed to a character variable of length 4 with $HEX8 format, but then prior TYPE116 datasets would be incompatible with newly build TYPE116, so I kept the variable as numeric. I am also asking IBM if these values are expected for the CICS Thread Number - their answer was that it really is a token and is not necessarily a number. Thanks to Helmut Roese, COM Gmbh, GERMANY.
FORMATS (It was decoded as 30:3GB before format MGTANDS was Apr 13, 1997 corrected by this change). Tandem Disk Drives type Apr 25, 1997 4245,4255,4265,4560, and 4570 are now decoded. Thanks to Hannu Koski, SJ Data, SWEDEN.
UTILCONT have BLKSIZE=, but now it does. Apr 1, 1997 Thanks to Vickie Drake, Levi Strauss & Co., USA. Change 15.040 IBM can write type 72 subtype 2 (TYPE72MN - RMF III MON) VMAC7072 records with zero for DURATM, which caused DIVISION BY Yüklə 28,67 Mb. Dostları ilə paylaş:
|