* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
the previous records, so you will require this change. Change 12.197 -APAF INPUT STATEMENT EXCEEDED RECORD LENGTH occurs when VMACAPAF a channel record does not contain channel type and mode. Nov 10, 1994 Insert after input of CHPIDBSY the statement: @; IF LENCHPID GE 12 THEN INPUT -APAF INVALID DATA FOR IOCDSLNO occurs when there is no IOCDS information in the record; a value of '00'x is not valid for a field input with &NUM.1. Insert a double questionmark in its input: IOCDSLNO ?? &NUM.1. to suppress the error message and associated record dump. Thanks to Willie Antman, Federal Deposit Insurance Corporation, USA. Change 12.196 Calculation of variables AOUTSZT and ARSPNET in NETSPY VMACNSPY dataset NSPYLU were changed by NETSPY, but I did not know Nov 9, 1994 about the changes until this user diligently validated my calculations with NETSPY's reports. -The AOUTSZT= calculation should have divided by TRANSNO (the number of transactions) instead of NETRSPNO (the number of transactions with measured response times) -The ARSPNET value calculated by MXG was missing if the NETRSPNO was zero, but the NETSPY logic now looks to see: if more than half of the transactions had a measured response; if not is the new bit in LUFDRSEQ flagging that the value in LRSPNET is the calculated ARSPNET value, or if not, then use the only data available to calculate the average network response. The new section of code in the MXG logic for NSPYRECs S, T, and U now reads: IF TRANSNO GT 0 THEN DO; AINPUTSZ=CINPUTSZ/TRANSNO; AOUTSZT=COTPUTSZ/TRANSNO; END; ELSE DO;
AINPUTSZ=.; AOUTSZT=.; END; IF NETRSPNO GE .5*TRANSNO AND TRANSNO GT 0 THEN ARSPNET=CRSPNET/TRANSNO; ELSE IF LUFDRSEQ='...1....'B THEN ARSPNET=LRSPNET; ELSE IF TRANSNO GT 0 THEN ARSPNET=CRSPNET/TRANSNO; ELSE ARSPNET=.; IF LUNRSPSS GT 0 THEN ARSPHOST=CRSPHOST/LUNRSPSS; ELSE ARSPHOST=.; IF OUTPUTNO GT 0 THEN AOUTSZW=COTPUTSZ/OUTPUTNO; ELSE AOUTSZW=.; IF NETRSPNO GT 0 THEN DO; /* calculation of ARSPNET, AOUTSZT was originally here */ IF SESSMGR='N' THEN DO; /* DONT CALC TARGET FOR ....*/ ... original code ... and then the two subsequent occurrences of ARSPNET=.; were deleted. Thanks to Wendy Gan, Malayan Banking Berhad, MALAYSIA. Thanks to Ms Foo Hooi Ping, Maybank, MALAYSIA.
WEEKBLDT was not applied to WEEKBLDT, causing TAPEMNTS NOT SORTED Nov 7, 1994 error. The correct list for dataset TAPEMNTS is: SYSTEM SHIFT DEVICE TMNTMTYP TMNTTIME Additionally, %INCLUDE SOURCLIB(EXPDBWEK) was also added to make WEEKBLDT consistent with WEEKBLD. Thanks to Gavin Ring, ALCATEL Australia, AUSTRALIA
VMACACHE CSAVAIL,CSOFFL,CNPINNED and CSPINNED are in K-bytes in Nov 7, 1994 the SMF record (while for all other devices, the values were in bytes!). As a result, those six variables are now multiplied by 1024 inside the nested DO groups for IF REPORLVL GE 1.6 ... and IF CACMODEL EQ 06X ... (so they will be printed correctly with MGBYTES format). Thanks to Russ Weid, CUNA Mutual Insurance Group, USA. Change 12.193 Type 6 variable SUPGROUP is now added to PDB.PRINT, as IMACPDB the destination group name has been found necessary for Nov 7, 1994 print accounting; macro _PDB6 now includes SUPGROUP. Thanks to Russ Weid, CUNA Mutual Insurance Group, USA. ===Changes thru 12.192 were included in MXG 12.04A dated Oct 23, 1994== Change 12.192 SMS Data Class, Storage Class and Management Class fields VMAC42 can only be 8 bytes, even though many records contained VMAC60 30-byte areas for these fields. This change changes all VMAC6156 inputs in all datasets to be length 8, so that the stored VMACDCOL size of the datasets containing these SMS constructs will VMACOMSM be minimized. (While some of these variables are named VMACVVDS consistently as DATACLAS, STORCLAS, MGMTCLAS, there are VMXGHSM many variant spellings when I instead used the field name Oct 23, 1994 in the DSECT as the variable's name.)
EXTMVCC The new version is incompatible because new subtypes now EXTMVCD replace old ones, so some existing datasets will have no EXTMVCF observations; however, MXG will not fail with new records EXTMVCH and some records (notably the system record) are still EXTMVCI processed successfully (albeit without new data added at EXTMVCP the end of the old record!). EXTMVCS Many new datasets are now created, some from new subtypes EXTMVDV and some from existing subtypes the MXG did not support. EXTMVIT The comments in member VMACTMVS identify which datasets EXTMVIX are created from which subtype. Some subtypes are not EXTMVJD yet decoded because the DSECT was not provided, and some EXTMVJDA were not decoded because no test data was provided, but EXTMVJDD Landmark provided both data and DSECTs for the important EXTMVJDI subtypes, so the new support has been tested with data, EXTMVJDO and its been out on the street for months with no report EXTMVJDR of any problems (but I doubt any power user has attacked EXTMVJDS each and every record type). Some testing notes: EXTMVJDT JDCPRES - large negative value for CSTORE residency EXTMVLC JDJTCBS - unreasonable (327,680) count of TCBs EXTMVMD and will update this note when I know more. EXTMVML
EXTMVSG Note: this change was in-progress in MXG 12.04A, and EXTMVXI existing datasets had new variables added, but the new IMACTMVS dataset decoding and validation was not added until MXG VMACTMVS 12.05, which is the minimum level for TMVS 1.3. Nov 15, 1994
VMACTCP with hex format (TCP/FTP/API-LCL/RMT-PO), but I now know Oct 21, 1994 they really are decimal numbers, so I have created new numeric variables (TCP/FTP/API-LCL/RMT-PN) and the label for the sets of variables indicated whether the Port Address is Hex (old PO variables) or Decimal (new PN's). Also, variables APILCLAD/APIRMTAD are now format $HEX8. Thanks to David Ehresman, University of Louisville, USA. Change 12.189 CICS 3.2.1 only. Statistics dataset CICDS duration VMAC110 variables with suffix TWT,TDT,TCT,ACT,SWT,SCT and prefix Oct 21, 1994 DSG,DS2, or DS3 were incorrectly converted. In the 18 lines preceding SKIP=SKIP-216;, make the variable name on the right hand side of the equal sign match the name on the left. Thanks to Tom Parker, Hogan Systems, USA.
VMAC28 preceded by INVALID DATA FOR IF message occurs because a Oct 21, 1994 line was dropped. Insert a line with @; immediately preceding IF LENAOF GE 128 THEN and this error will go away. In addition, this error discovered the new date time stamp SYNCTIME at the end of the NAD section, which is now captured by MXG. Thanks to Tom Parker, Hogan Systems, USA. Change 12.187 NPM type 28 NVV section caused CONVERSION note on the log VMAC28 and incorrect value for flag variables LNVVCACH,LNVVREMV, Oct 21, 1994 and LNVVMOUN, because the three variables tested should have been suffixed with "X". That is, change: IF LNVVCACH GT 0 THEN LNVVCACH='Y';ELSE LNVVCACH=' '; IF LNVVREMV GT 0 THEN LNVVREMV='Y';ELSE LNVVREMV=' '; IF LNVVMOUN GT 0 THEN LNVVMOUN='Y';ELSE LNVVMOUN=' '; to read:
IF LNVVCACX GT 0 THEN LNVVCACH='Y';ELSE LNVVCACH=' '; IF LNVVREMX GT 0 THEN LNVVREMV='Y';ELSE LNVVREMV=' '; IF LNVVMOUX GT 0 THEN LNVVMOUN='Y';ELSE LNVVMOUN=' '; Thanks to Tom Parker, Hogan Systems, USA. Change 12.186 Candle Omegamon for VTAM V160 incompatibly changed their VMACOMVT user SMF record, causing MXG error messages and possible Oct 20, 1994 looping. A four byte field was expanded to eight bytes in the middle of the exception record (to store VTAM NODE name), requiring extensive recoding in MXG; however, you can partially circumvent the error by inserting: IF NASMFPVN GE 'V160' THEN INPUT @LOC+60 @; immediately before the statement: %%INCLUDE SOURCLIB(EXOMVEXC); With that circumvention, the looping will not occur, and the rest of the OMVTxxxx datasets will be valid, but the OMVTEXCE dataset will still be trashed until you install the new MXG version. What's really ludicrous about their incompatible change (and its unnecessary abuse of your time to install a new version of MXG) is that there was no need for the incompatibility - there are two other areas, one of 12 bytes, one of 16 bytes, which could and should have been used to store the Node Name compatibly! Thanks to John Astle, National Australia Bank, AUSTRALIA.
ANALBLSR error, if the last observation in DSETOPEN did not meet Oct 20, 1994 the subsetting IF criteria. The subsetting IF statement IF TYPETASK=:'JOB' AND ACCESS=:'VSAM' ... etc must be replaced with this logic: IF NOT (TYPETASK=:'JOB' AND ACCESS=:'VSAM' AND (LSRSTAT='Y' OR LSRSTAT='?')) THEN DO; IF EOF THEN CALL SYMPUT('CANDIDAT',CANDIDAT); DELETE; END;
Thanks to Paul Hill, Midland Bank, UK. Change 12.184 NETSPY Token-Ring Dataset NSPYTR variable TIC_UTIL was 10 VMACNSPY times too large (the convert nanosec to microsec was Oct 20, 1994 off by 1000, but conversion to percent was off by 100!). In addition, the old equation used IFRAMESS/IFRAMESR to calculate TIC_UTIL (because that's what LEGENT used in its on-line display), but now LEGENT PTF 71565 corrects their on-line calculation using TRTFRAMS/TRTFRAMR instead of just I-Frames (so their on-line reports match their batch reports), so now the MXG calculation agrees with the LEGENT calculation with that PTF. The correct MXG equation (in 2 places) is now: TIC_UTIL=100* ((((BYTSENT*TRTPBYTS)+(BYTRECV*TRTPBYTR))/1000000000)+ (((TRTFRAMS*TRTPFRMS)+(TRTFRAMR*TRTPFRMR))/1000000)) /DURATM; Thanks to Janice Radel, General American Life Insurance Co., USA. Change 12.183 Change 11.201 intended to change the spelling of variable VMAC90 NEWDSN to NEWDSNSM in TYPE90, but it was misspelled as Oct 19, 1994 NEWSDNSM. It is now correctly spelled NEWDSNSM. Thanks to Graeme Yeandle, British Telecom, UK. Change 12.182 UTILCICS fails with syntax error, because line 342 is a UTILCICS /* */ line which prematurely terminates the comment Oct 19, 1994 block around the new text. Delete that line and UTILCICS will execute. Read the comments and you can understand how to use the output to diagnose MXG error messages in in your CICS type 110 SMF records. Thanks to Norbert Korsche, OMV, AUSTRIA.
BUILDPDB actual change was not noted in member CHANGES. While not Oct 19, 1994 usually needed, it was added to the PDB for completeness, since resources for tasks which start before SMF has come (called "early address spaces" like CATALOG, PCAUTH, GRS, ALLOCAS, etc.) may have their resources recorded in the TYPE30_6 instead of TYPE30_V (and since these ASIDs are not normally terminated, even at shutdown, there may not ever be a TYPE30_4 record written for them). Thanks to Norbert Korsche, OMV, AUSTRIA. Change 12.180 Type 42 subtype 5 (TYPE42SR) and subtype 6 (TYPE42DS) VMAC42 variables STARTIME/ENDTIME can be erratically wrong. The Oct 19, 1994 timestamps are in GMT instead of LOCAL if TIMEZONE has a Nov 9, 1994 non-zero GMT offset specified in SYS1.PARMLIB(CLOCKnn), (but SMFTIME is always on the LOCAL clock). For the TYPE42SR and for the TYPE42DS INTVCLOS=0 (CLOSE) records, MXG did not attempt to convert to GMT, so the only error in those records was that the STARTIME/ENDTIME are GMT. However, MXG did use the difference between SMFTIME and ENDTIME for the TYPE42DS INTVCLOS=1 (INTERVAL) record to determine the GMT offset and then convert START/ENDTIME to LOCAL, but I now have data to prove it is impossible to convert, because the SMF record timestamp can be many hours later than the end of interval (for swapped out tasks, the SMF record is not timestamped until the task swaps back in!). (In MXG 12.04A only, I actually added GMT conversion logic to the INTVCLOS=0 in TYPE42DS and in TYPE42SR, but now this revised change removes conversion of GMT to local for all type 42 subtypes. Until IBM gives us a GMT offset in these records, MXG cannot convert GMT to local in any safe fashion. In addition, some site's data show SMF42PTE later than SMFTIME, which really messes up any conversion algorithm, so the only safe strategy is to leave the timestamps in GMT as provided by IBM - you, knowing your own GMT offset must then convert STARTIME/ENDTIME back to LOCAL in your report programs using TYPE42SR or TYPE42DS.) See also the MVS Technical Notes for additional type 42 problems and concerns. P.S. APAR OW05631 added the DEVNR and VOLSER to TYPE42DS. Thanks to Diane Eppestine, Southwestern Bell, USA. Change 12.179 SAP Accounting record under IMS was not properly decoded. VMACIMS Change the INPUT of SAPTIMTR from "TU4." to "&PIB.4.3", Oct 19, 1994 change SAPCPUT (three times) and SAPELTI (once) from "&PIB.4.1" to "&PIB.4.3". Also, the HH, MM, SS and TH fields used for construction of SAPTIME probably need to be changed from "&PIB.1" to "&PK.1", but that has not yet been validated with a hex dump of a record. Thanks to ???, SJ ECONOMISERVICE, SWEDEN. Thanks to Boerje Edlund, SAS Institute, SWEDEN.
RMFINTRV labeled consistently as "TOTAL*CPU TIME*CAPTURED" in all VMAC30 of the SMF/RMF datasets created by MXG. Previously, it VMAC32 had different labels to describe its contents (TCB+SRB in VMAC434 some records, TCB+SRB+HPT+... in RMF, etc.) but when you VMAC435 mixed SMF records (like in BUILDPDB) you ended up with an VMAC7072 inaccurate label. To know the components of CPUTM in a VMACDB2 particular record, read the ADOCxxxx description. VMACORAC VMACTSOM
Oct 19, 1994 Thanks to Waldemar Schneider, SAS EUROPE, GERMANY. Change 12.177 Archaic Landmark 1.7 only. Variable STARTIME in MONIDBDS TYPEMONI was missing, because its creation (five lines, starting Oct 18, 1994 with IF CREATIME and ending with RESPONSE=) should have been located immediately after APPLID=SYSID; statement. Thanks to Warren Tanaka, Farmers Insurance Group, USA.
TRNDTALO No test data was available when the trending member was Oct 3, 1994 written; test data exposed the error.
ANALTMDB code of 4, a note that ANALTMDB has no lines, and there Oct 3, 1994 are 2396 members unloaded (instead of the expected 2379)! This does not really cause a problem, but the error is because member ANALTMDB contains an unloaded PDS, but I failed to change the "./" IEBUPDTE control statements to "XY", in my source PDS, so they were propagated into the tape, and so when IEBUPDTE unloaded the tape, those "./" caused members (all starting with RTMD) to be created! Changing the "./" in my ANALTMDB to XY avoids the error, but you need to take no action.
VMACDB2 DB2ACCTB dataset by Change 12.169 should have been Oct 3, 1994 spelled QLACLOCN. The text of 12.169 was changed in MXG 12.04A to avoid further confusion. Change 12.173 MXG 12.04 only. New NPM dataset NPMLINE did not have all VMAC28 of the variables it should have, due to spelling error. Oct 3, 1994 Change the one occurrence of _VA28NLF to _VA28NLV to fix!
ANALAPAF that sums MDF domain busy to provide a total platform Oct 3, 1994 busy is added by this user contribution. Note that the report code requires the number of real engines so that it can calculate correctly. Thanks to Will Evans, NIKE Information Services, USA. Change 12.171 Cosmetic changes. In members VMAC110,VMAC74,VMACSTX, and EXTY99U2 VMACTMNT "3TH" was changed to "3RD". In EXTY99U2 the VMAC110 comment was corrected to TYPE99_2, and in member VMACTLMS VMAC74 the missing "%%INCLUDE SOURCLIB(EXTYTLMS);" statement was VMACSTX added at the end (but its absence caused no error, as VMACTLMS there is only one dataset in the DATA statement!). Oct 3, 1994 Thanks to Freddie Arie, Lone Star Gas, TEXAS. Change 12.170 MXG 12.04 only. The last twelve lines (starting with ANALCISH "options ..." in lower case) must be deleted; this was a Oct 1, 1994 testing invocation within the member that doesn't belong!
ASUMDB2A variables QWACBSC, QWACESC, and QLACLOCN added to its ASUMDB2B KEEP= list in VMACDB2, so that new summary and trending TRNDDB2A members ASUMDB2B and TRNDDB2B could be added to MXG. In TRNDDB2B addition, variables QWACPKGN QLACRRRC and QLACRRSE were VMACDB2 added to the SUM= operand in existing ASUMDB2A/TRNDDB2A Sep 29, 1994 to support future report enhancements.
EXTELEAP new datasets: EXTELESH TELEAPPL - TELEVIEW Application Logon/Logoff EXTELETE TELETERM - TELEVIEW Terminal Logon/Logoff IMACTELE TELESHUT - TELEVIEW Shutdown. TYPETELE See Change 12.229 update. VMACTELE Sep 29, 1994 Thanks to Tom Parquette, MONY, USA.
VMACOMCI occurs when those supposedly packed decimal fields have Sep 29, 1994 hex zeroes! Each occurrence of &PD. is now preceded by the double question mark modifier to suppress the message and the hex dump of the offending record. Thanks to Sudie Wulfert-Schickedanz, Edward D. Jones, USA. Change 12.166 Support for CICS/ESA 4.1.0 is added by this change. EXCICDL3 The changes are compatible, except that new Statistics EXCICFEC Subtypes will generate messages on the SAS log if you EXCICFEP do not have MXG 12.04 or later installed. EXCICFET EXCICPGG 1.Major enhancements to the CICSTRAN dataset include: EXCICUSG a. Of real significance, nine new sets of counts and EXCICXMC durations of long-wanted delays and waits: EXCICXMG DSPDIOcn/tm - First Dispatch Total Count/Delay EXCICXMR ENQDIOcn/tm - KC Enqueue Delay Count/Duration IMACCICS LU61IOcn/tm - LU 6.1 I/O Wait Count/Duration VMAC110 LU62IOcn/tm - LU 6.2 I/O Wait Count/Duration Apr 28, 1994 MXTDIOcn/tm - First Dispatch MAXTASK Delay Cnt/Dur May 17, 1994 RMISIOcn/tm - RMI Suspend Count/Duration Jun 3, 1994 RMITIOcn/tm - RMI Elapsed Count/Duration Jun 23, 1994 SZWAIOcn/tm - FEPI Suspend Count/Duration Jun 30, 1994 TCLDIOcn/tm - First Dispatch TCLASS Delay Count/Dur Aug 19, 1994 (This delay is especially important Sep 29, 1994 for sites that have had to throttle DB2 resource consumption by using CICS TCLASS limits - finally you can show your DB2 guy how much delay is being caused by his poorly designed DB2 transactions!). b. GMT offset, so that MXG can finally correct start/stop times from GMT to local for sites which have non-zero GMT offset in member CLOCKnn. c. Replacement of one-byte TCLASS with 8-byte TCLSNAME. d. Removal of variables OPERATOR TCSTG PC31UHWM PC24UHWM e. Addition of resource variables TCM62IN2 TCC62IN2 TCM62OU2 TCC62OUT for LU 6.1/6.2 messages/characters. f. Addition of resource variables PC24RHWM PC24SHWM and PC31SHWM for read-only and shared program storage. g. Addition of resource variables SZALLOCT SZRCVCT SZENDCT SZSTRTCT SZCHROUT SZCHRIN SZALLCTO SZRCVTO and SCTOTCT for FEPI activity. h. TASKNR contains special character transaction numbers that MXG decodes as special values of variable TASKNR: Characters MXG TASKNR Description '00'X,'III' 13,224,393 System Initialization Task '00'X,'J00' 13,758,704- Journal Control, Jnr=00- '00'X,'J99' 13,761,017 Journal Control, Jrn=99 '00'X,'JBS' 13,746,914 Journal Control '40'X,'TCP' 14,926,807 Terminal Control (so that these values do not set the "INVALID TASKNR" error condition in MXG processing). 2.Major enhancements to the subtype 2 Statistics record: a. Nine new statistics datasets are created: CICDLIG CICFEPIC CICFEPIP CICFEPIT CICPAUTO CICUSG CICXMC CICXMG CICXMR b. Six existing statistics datasets will now have zero observations with CICS/ESA 4.1: CICDLIT CICDMG CICDMR CICTC CICTCLR CICTSR 3. Complete list of all new variables in old datasets and new datasets can be found in member DOCVER12. Thanks to Edward McCarthy, Health Insurance Commission, AUSTRALIA. Thanks to Scott Wiig, First Bank Systems, USA. Thanks to JP Meijer, Standard Bank of South Africa Limited, RSA. 4. The following table lists all CICS Statistics datasets now created by MXG (note that those with * after Name will have zero observations with CICS/ESA 4.1 records):
Dataset IBM Exit sfx Name STID Description of Data Set Name Name
CICAUTO 24 Autoinstall Global STIAUTO AUT A04 CICCONMR 76 ISC/IRC Mode Entry Specific STICONMR CO3 A20 CICCONSR 52 ISC/IRC System Entry Specific STICONSR CO1 A14 CICCONSS 54 ISC Connection - System Security STICONSS CON A21 CICDBUSS 28 DBCTL Global Unsolicited STIDBUSS DBU DBU CICDLIG 72 DL/I (Global) STIDLIG DL3 A25 CICDLIR 70,7 DL/I Local Specific STIDLIR DL1 A18 CICDLIT* 71 DL/I Local Totals STIDLIT DL2 A18 CICDMG* 15 Domain Manager Global STIDMG DMG DMG CICDMR* 13 Domain Manager Specific STIDMR DMR DMR CICDQG 45 TDQUEUE Transient Data Global STITDQG DQG A11 Yüklə 28,67 Mb. Dostları ilə paylaş:
|