* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
Oct 30, 1998 (for concatenation). I use exclamation points instead of long vertical bars because exclamation points translate between ASCII and EBCDIC in all languages and long bars did not translate from UK to USA (hence this change). Thanks to David Nuechterlein, Nissan Motor Corporation, USA.
TYPETMS5 volumes; some variables from TMSRECS dataset were blank Oct 30, 1998 in the output DSNBRECD dataset, because it turns out that the RETAIN statement doesn't work like I thought it did! See SAS Technical Note 1., Newsletter 35 about RETAIN. The SET statement to combine TMSRECS and DSNBRECS was changed back to MERGE, the IF FIRST.VOLSER THEN was changed to IF FIRST.VOLSER and INTMS THEN DO;, and the subsequent adjacent END; and IF INTMS THEN DO; statements were deleted (so that now the logic is FIRST.VOLSER to do all the INTMS processing followed by IF INDSNB to do the DSNBREC processing, and then the if LAST.VOLSER to output the volume record into TMS.TMS). Thanks to Chuck Hopf, MNBA, USA. Change 16.250 -MXG 16.04 only. Zero observations in dataset SYNCSORT if ASUMPRTR you used IMACSYNC from 16.04. Syncsort's ID macro name is IMACRMDS _SYNCID, but the example in IMACSYNC had _IDSYNC. MXG's IMACRTE convention now is for the name to be _IDxxxx, but three IMACSYNC early names were spelled _xxxxID ( _SYNCID in IMACSYNC, Oct 27, 1998 _RMDSID in IMACRMDS, and _RTEID in IMACRTE ) and all are now correctly spelled in their IMAC comments. These IMAC members: CADM, CIMS, DIOS, HO15, OLDV, SAM, SFS, TAND had incorrect comments that have been revised -Member ASUMPRTR deletes datasets with PROC DATASETS, but it was missing the NOLIST option, so it printed lines of unneeded/unwanted messages on your SAS log. Now it don't. Thanks to Glenn Harper, Memorial HealthCare System, USA.
EXILKA23 -New subtypes and new headers for new subtypes will cause EXILKAAP the old MXG code to fail. EXILKAIU -New variables were added to the ILKAST20 dataset, and it EXILKAOE now contains both subtype 20 and subtype 22 events; the EXILKAPR variable SMFACTYP identifies which subtype created the EXILKASP observation (20 is end of ftp, 22 is end-of-volume ftp). EXILKAVS -New ILKAST23 dataset tracks TELNET sessions from st 23. FORMATS -Subtypes 110 thru 124 create new ILKASTPR protocols VMACILKA dataset. Note: Nov 4: The IF 150 LE ... statement was VMXGINIT wrong; it should have been ELSE IF 150 LE ... and the Oct 27, 1998 IN (110-112,118,123-124) should have been written as Nov 4, 1998 IN (110,111,112,118,123,124) to process 111 records. -Subtypes 150,151,152 create new ILKASTAP, ILKASTIU, and ILKASTOE for API, IUCV, or Open Edition Close events. -Subtype 80 creates ILKASTVS with global virtual storage and creates ILKASTSP with individual subpool virtual storage; only subpools with non-zero allocation/usage are output in ILKASTSP. -Subtype 100 is not yet decoded, no test record yet. -However the new subtypes 80 and 150-152 have severe data errors. The UOF and DOF offsets are off by one byte due to SMFACACB being only 7 instead of 8 bytes. (MXG can protect that offset). In the subtype 152, the counters appear to be off by one byte (open problem with vendor). In the subtype 80, the subpool sections are not all 40 bytes long - some are 39 (open problem with vendor). Thanks to Will Evans, CNF Transportation, USA. Change 16.248 The two ELSE IF PCTRDYWT=.; statements should have been VMAC7072 written ELSE PCTRDYWT=.; but very fortunately, this Oct 23, 1998 typo-that-became-a-logic-error has never been executed. Only when there are OS/390 systems with 15 or more CPUs would this typo have become an execution error, and then the results (if not the cause) would have been obvious: zero obs would have been created in dataset TYPE70! Thanks to Harry Olschewski, DeTeCSM/SLM, GERMANY Change 16.247 Support for MainView Compressed history files. Beginning ASMMNVW with MainView for MVS 2.5, and soon with other MainView TYPEBBMQ products, Boole history files are compressed (although TYPECMFV specifying HISTCOMP=N as a parm on the EXEC card in the VMACBBMQ MMR PAS JCL will turn compression off). Boole's SAMPLIB VMACCMFV member MMRHUNC contains a standalone decompression job, Oct 23, 1998 but this change creates a new SAS INFILE exit "MNVW" that decompresses the compressed data during INPUT on the fly! You assemble the ASMMNVW member to create the load module named MNVWIFUE and store that member in a load library that you concatenate to the //STEPLIB for the SAS step that will read the compressed data. You then copy the INFILE-macro definition from the VMAC into the IMACxxxx member for the specific product: IMACxxxx INFILE-macro INFILE Product IMACCMFV _CMFV CMFVSAM CMF/MMR IMACBBMQ _BBMQ BBMQVSAM MQ Series and add the string "MNVW" (which is the exit name) after the INFILE name: MACRO _CMFV INFILE CMFVSAM MNVW STOPOVER .... ; % MACRO _BBMQ INFILE BBMQVSAM MNVW STOPOVER .... : %
macros _CMFV and _BBMQ that are defined in their VMACxxxx member, so you can enable the MNVW INFILE exit either by copying the _CMFV or _BBMQ macro definition into the IMACxxxx for that product, or you can use the MACKEEP architecture to change the definitions in your SYSIN. As long as you have the load module in the //STEPLIB concatenation, you can enable the INFILE exit now as it will read any mixture of compressed or uncompressed data. The Assembly code in ASMMNVW was written by Ray Revis. Thanks to Ray Revis, Boole and Babbage, USA. Change 16.246 Tandem variable RESPTM in dataset TANDLINE was too large VMACTAND by a million; it should be INPUT as @119 RESPTM &PIB.8.6 Oct 22, 1998 instead of just &PIB.8. Thanks to Raff Rushton, Kaiser Permanente, USA. Change 16.245 Support for the RACF "installation-defined data field" in VMAC80A SMF80DTP='07'x data segment is now INPUT as new variable Oct 20, 1998 RACF07DT with maximum length of 32 bytes. -replace
WHEN (7) DO; INPUT +RACFDLN @; END; with
WHEN (7) DO; IF 0 LT RACFDLN LE 32 THEN INPUT RACF07DT $VARYING32. RACFDLN @; ELSE IF RACFDLN GT 32 THEN DO; INPUT RACF07DT $EBCDIC32. @; SKIP=RACFDLN-32; INPUT +SKIP @; END;
END; -add variable name RACF07DT to the KEEP= list for datasets TYPE8008,TYPE8009,TYPE8010,TYPE8011,TYPE8012,TYPE8013, TYPE8020 and TYPE8021, as this field can exist in all of those RACF events. Thanks to Albert Venetitay, CTICEP, FRANCE. Change 16.244 Notes amplifying what is captured in which TCP records ADOCTCP from an IBM Support person were added to documentation. Oct 20, 1998 The TYPETCPA "APICALLS" event appears to capture all bytes of all TCP/IP traffic, and even includes FTP counts so it appears using TYPETCPA for accounting bytes in/out may be feasible. Thanks to Xiaobo Zhang, ISO, USA.
VMACCMFV VMACCMFV, but it was missing from the KEEP= list for some Oct 20, 1998 datasets. Thanks to Neil Ervin, Huntington Services, USA. ======Changes thru 16.242 were in MXG 16.04 dated Oct 19, 1998====== Change 16.242 MXG expected the new TCP Statistics record to be created IMACTCP as SUBTYPE=5, but TCP lets the TCP installer set whatever VMACTCP value they want for each subtype. MXG has always been Oct 19, 1998 able to look inside the TCP record to determine what it was without using the subtype, but the Statistics record support does depend on the SUBTYPE value. This change defines the new MACRO _SUBTCP5 5 % with its default of 5 than can be changed with MACKEEP/IMACKEEP/IMACTCP if your subtype is not five. Thanks to Ben Coonfield, Browning-Ferris Services, Ind., USA. Change 16.241 MXG 16.04 only. Dataset VXMTRDEV was not created, so the VMACVMXA PROC MEANS caused ERROR: THERE ARE NO ANALYSIS VARIABLES. Oct 19, 1998 The correction was inside MACRO _SIODDEV definition; the SORT was changed to OUT=_LMTRDEV instead of ZZMTRDEV, and the SET ZZMTRDEV was changed to SET _LMTRDEV; Thanks to Steve Clark, California Federal Bank, USA. Change 16.240 Counter variables MROTRAN/DB2TRAN were set to zero, but ASUMUOW this should be done only if the observation did not come Oct 19, 1998 from the SPIN library, so IF NOT INSPIN THEN DO; ... END; was added around the two statements. After IF LAST. logic, new variables WTRLIOCN/WTRLIOTM were not reset, but now are equated to HTRLIOCN/HTRLIOTM. Thanks to Harry Olschewski, DeTeCSM/SLM, GERMANY.
VMAC30 times by 128, and variables SMF30EIC/SMF30EID/SMF30EIW Oct 19, 1998 were not multiplied at all; the last three lines should have been for EIC/EID/EIW instead of repeat AIC/AID/AIW. These new address space level connect/disconnect/pend durations were added in OS/390 Release 2.4. Thanks to Martin Peck, iT-AUSTRIA, AUSTRIA.
IMAC115 Series are now as they were before 16.04. The correct VMAC115 names of MQMLOG, MQMMSGDM, and MQMBUFER are created, Oct 19, 1998 instead of TYPE1151, TYPE1152, and TYPE115B. Thanks to Brian Crow, British Telecom, ENGLAND.
VMXGOPTR &OPTNAME=S. This had no effect in any delivered code, Oct 19, 1998 but if you had used %VMXGOPTR to reset the S= option, it did not work. Thanks to Tom Parker, CSC/Hogan Systems, USA.
EXTY8026 dataset TYPE8026. This event is new in RACF 2.2. This VMAC80A new record causes MXG 14.14 to fail, but MXG 15.15 and Oct 19, 1998 later did not fail with either TYPE80A or TYPE80. Thanks to Roger Lowe, CITIBANK Asia Pacific Center, SINGAPORE.
VMAC110 with JCRLL value of zero caused MXG to loop on the same Oct 19, 1998 record. MXG now protects for this invalid record, prints an error message and deletes the bad records. Thanks to Tom Wieland, Phoenix Home Life Mutual Insurance Co., USA.
VMAC57 INVALID DATA FOR JESNR because the old 4-byte location is Oct 19, 1998 now hex zeros. Adding the ?? modifier between JESNR and &NUM.4. eliminates the message. There actually was no error in the TYPE57 dataset, since the real JESNR is read from a 5-digit field later in the record, but this will eliminate an unnecessary and confusing message. Thanks to Bendt Wiberg, CSC Computer Management A/S, DENMARK. Change 16.233 MXG 16.04 only. Two Sort Macros, _SDB2STB and _SDB2STR, VMACDB2 referenced &PDB2STB..DB2STATB and &PDB2STR..DB2STATR Oct 19, 1998 instead of _LDB2STB and _LDB2STR, causing ANALDB2R to fail when PDB=SMF was specified, if there was no PDB DD. Thanks to Chuck Hopf, MBNA, USA.
NTINTRV to be added at the end of the BY statement. Oct 14, 1998 Thanks to Howard Glastetter, Weyerhaeuser Company, USA. ======Changes thru 16.231 were in MXG 16.04 dated Oct 9, 1998====== Change 16.231 Variables STDUPLEX, FCB, OVLYLOAD, and OVLYUSED are added BUILD005 to MACRO _PDB30_6 so that they will exist in PDB.PRINT BUIL3005 dataset (and hence will be in ITSV's XPRINT table). These Oct 8, 1998 variables were needed because they had been kept in MICS. Thanks to Theo Peelen, SHELL, THE NETHERLANDS.
VMAC30 have job records with MULTIDD='Y' (there were so many DDs BUILD005 that SMF had to write multiple physical type 30 records), BUIL3005 the PROC SORT of TYPE30_5 was not robust enough to delete Oct 8, 1998 those MULTIDD duplicates, which causes variable RESTARTS in PDB.JOBS to be greater than 1 for job/TSO/STCs that were not restarted. Fortunately, nothing else was bad! Member VMAC30 was changed to KEEP variable EXTRADD in the TYPE30_5 dataset, and the BY statement in macro _STY30U5 was changed by adding MULTIDD EXTRADD at its end. Members BUILx005 had the BY statement for the PROC SORT of DATA=TYPE30_5 changed by adding MULTIDD EXTRADD also. Thanks to Chuck Hopf, MBNA, USA.
VMACTPMX typos and token id typos in the format table. Message is Oct 9, 1998 now printed on log if an unknown token is encountered. Oct 19, 1998 The token finding logic was revised to work under ASCII. Variables PERFORM and INSYSAF no longer have $HEX format. Labels were "split" with * characters and ending parens. IMACTPMX's comments were corrected. Thanks to Lawrence Jermyn, Fidelity Investments, USA. Change 16.228 NTSMF Version 2.2 records with Record Header 2.2.1 VMACNTSM cause NEW OBJECT errors, OBJECT='SERVICE PACK 3', because Oct 8, 1998 I did not understand the new Record Header Values: Header Description 2.0 Original Header format, no 0.0 record 2.0.h Extended Header format, 0.0 record created 2.1 Compressed Header format. 2.2.1 0.1 record created, 2.0.H Extended Format 2.2.2 0.1 record created, 2.1 Compressed Format
reduces the amount of data in each record header and hence the volume of data sent from the monitored to the monitor. However, if you have not installed 2.2 on all of your machines, NTSMF 2.2 will use the old spec on the monitored machines, and if they contains HEADER FORMAT 0, you will write the old, longer records until you change that parameter. The MXG variable, VERSION, is used not only to determine the format of the record, but whether there will be a 0.0 record (used to calculate STARTIME if it exists), whether there are CPU fields in the 0.0 record, whether there will be a 0.1 record, and whether it will precede (prior to NTSMF 2.2) or follow (NTSMF 2.2+) the 0.0 record! Thanks to Howard Glastetter, Weyerhaeuser Company, USA.
VMACCIMS error, because MXG only input the new segment for 6.1. Oct 8, 1998 Change the IMSLEVEL test from 61 to 51 so it will read: IF IMSLEVEL GE '51' AND IMFLEVEL GE '32' THEN INPUT @373+OFFIMS UTCA .... Thanks to Harmut Peter, R & V Allgemeine Versicherung AG, GERMANY Thanks to Don Cleveland, Wellpoint, USA.
VMAC110 CICSSAP from the optional SAP journal segment. In 16.04, Oct 8, 1998 the _LCICSAP and _WCICSAP datasets had incorrectly spelt the dataset CICSAP instead of CICSSAP. Thanks to Theo T. Peelen, Shell, THE NETHERLANDS.
VMACHSM DFSMS 1.4 records caused no error message. MXG Change Oct 8, 1998 16.136 incorrectly located the INPUT for new variables WFSCPUTM and WFSACCT to before the INPUT of the VOLSER segments. If there were 36 bytes of VOLSER data, the new logic INPUT them as WFSCPUTM/WFSACCT, and then when the old logic tried to read the VOLSERs, there were none!. In addition, only the first VOLSER segment would have been input. This change moves the new logic until after any VOLSERs were input, and also iterates on WFSRNENT. Thanks to Michael E. Friske, Fidelity Investments, USA.
VMACOMVT they are not from OMEGAMON for VTAM. (If you have OMVT Oct 6, 1998 records and have overridden the _IDOMVT macro in either your IMACOMVT or IMACKEEP, you don't see this error - it is only if you don't have OMVT but some other product is writing type 220 records). Change the "220" in _IDOMVT to "512", which is the impossible default SMF number that should have been in VMACOMVT. Thanks to Edd Carter, General Mills Restaurant, USA.
RMFINTRV changed from "PROCESSOR" to "PROCESSORS" to be consistent Oct 6, 1998 with variable CPUUPTTM and because all three variables do contain the total time that all processors in the CEC were active, uncaptured, or "up" (i.e., available). Thanks to Edd Carter, General Mills Restaurant, USA. Change 16.222 -First MXG 16.04 only. The new _STYTCPA "Sort" macro, VMACTCP used only by the new TYPSTCP member, was missing an VMXGINIT ending parenthesis in its definition. The DATA statement Oct 6, 1998 that follows "MACRO _STYTCPA" statement should be: DATA _LTYTCPA (LABEL='TYTCPA: .... API CALLS (SOCKETS)'); -Macro variable PTCPTCPS was not GLOBALed nor %LET in the VMXGINIT member, causing unresolved macro reference. Thanks to Chuck Hopf, MBNA, USA. ======Changes thru 16.221 were in MXG 16.04 dated Sep 30, 1998====== Change 16.221 Trending of Tape Allocations could overcount in a last ASUMTALO interval that had data spread across two week's input TRNDTALO with overlap during the same hour (i.e., last weeks last Sep 30, 1998 allocation was from 0300-0315, this week has an allocate from 0330-0345). This change corrects the allocate tape count in that interval, but MXG's revisions in ML-17 of ASMTAPES/MXGTMNT automatically synchronizes its interval with SMF interval, so the exposure is reduced further. Thanks to Jim Stevens, MBNA, USA.
VMACNTSM records was revised. For NTSMF 2.0 and 2.1, the 0.1 was Sep 28, 1998 written first and then the 0.0, so MXG could OUTPUT to NTCONFIG from the 0.0 record (including the three fields that were RETAINed from the preceding 0.1 record). Now, the output for NTCONFIG was located until after both the 0.0 and 0.1 were read and all their fields RETAINed, and then OUTPUT from the 0.1 if NTSMF is 2.2 or later, or OUTPUT from the 0.0 if NTSMF is Version 2.1 or earlier. This really has minimal impact, as the NTCONFIG dataset is only a log of configuration status, and not critical, but now as new configuration fields are added to the 0.1 record, they will be added to dataset NTCONFIG.
ASMIMSL6 -The TIME DEC call returns a 0CYYDDDF format that is used Sep 28, 1998 as a reference date to decide whether to delete or ignore "old" INQUEUE records, but that value should have been converted to the YYYYDDDF format of the INQUEUE record. -ASMIMSL6 was enhanced to prevent U0012 abends if there were insufficient slots being allocated on the initial table load. While the number of slots can be provided to ASMIMSL6 thru the //SLOTS DD statement, this enhancement eliminates the need to specify slots in advance; the INQUEUE is read (fast) to count records and NUMSLOTS is reset to records+1000 if the slots are too few. Thanks to Alan Green, British American Financial Services, ENGLAND.
VMXGTAPE is on tape, but when run under ASCII SAS, produced notes Sep 30, 1998 the SAS log. Additionally, when VMXGTAPE was used with a DISK dataset, the second execution failed with an error message UNABLE TO CLEAR LIBNAME. That error is fixed by relocating the RUN; statement to be before the LIBNAME. The error was not reported, because the only member to use VMXGTAPE twice, TYPECIMS, uses tape rather than disk for those large IMSTRAN files from IMF, and VMXGTAPE did work fine when the MVS SAS dataset was on tape. Thanks to Freddie Arie, Lone Star Gas, USA.
EXNTNSHS NetShow Unicast Service creates NETSHOWS and NETSHOWU EXNTNSHU datasets, respectively. VMACNTSM -Logic for NTCONFIG was revised to read CPU Speed field Sep 23, 1998 when VERSION GE '2.0.H' AND NTVERSN GE '3.51' because the segment exists in the 2.0.h records. Additionally, VERSION=UPCASE(VERSION); was inserted after its INPUT and before the test to avoid case sensitivity. Thanks to Jim Quigley, Con Ed, USA. Thanks to Denny C. Wong, New York Life, USA. ======Changes thru 16.216 were in MXG 16.03B6 dated Sep 23, 1998====== Change 16.216 Support for OS/390 2.6 (COMPATIBLE) added new data: VMAC73 -Dataset TYPE73 new var SMF73ACR, Channel Path Acronym VMAC74 -Dataset TYPE74 new var DEVDISNV, Device Disconnect Time VMAC79 Is Not Valid flag, Y or Blank. Sep 23, 1998 -Dataset TYPE74. IBM renamed field SMF74NID to SMF74DCT and increased it from 26 to 28 bytes, but MXG continues to name the variable SFM74NID and it contains either the Node Descriptor ID for self-describing devices, or the four-byte Device Number. -Dataset TYPE74CA new var R745CCMT, Hardware Type and Model of the Control Unit, 28 byte character variable. -Dataset TYPE79C new var R79CACR, Channel Path Acronym. -Changes to TYPE99 subtype 6 were already in MXG. Change 16.215 ACF2 Version 6.2 INCOMPATIBLY changed the type 'V' record VMACACF2 causing INPUT STATEMENT EXCEEDED RECORD LENGTH error. Sep 23, 1998 Fortunately, most sites don't use the ACF2VR dataset that is built from this record, so you can process their other 6.2 data by inserting the new line that tests LENGTH-COL so that member VMACACF2 then contains: IF ACVMFIDC GT 0 THEN DO _I_=3 TO ACVMFIDC; IF LENGTH-COL+1 GE 8 THEN INPUT ACVMFSEC $EBCDIC8. @; /*SECONDARY*AUTHID*/ CA provided same day documentation and technical support of the changes in their record, which inserted an offset field to the optional DB2 data fields, which is now used and tested by MXG. No other new data was added. Thanks to Frank d'Hoine, National Bank of Belgium, BELGIUM. Yüklə 28,67 Mb. Dostları ilə paylaş:
|