* copyright (C) 1984-2019 merrill consultants dallas texas usa
Yüklə 28,67 Mb.
|
|
DATA _VAR120 /view=typ1209r _SMF _CDE120
run; _st1209r
run; macro _st1209R % _S120; RUN;
The DATA/SORT took 6 hours, the VIEW/SORT took 2. Thanks to Joe Faska, DTCC, USA. Thanks to Michael Oujesky, DTCC, USA.
VMXGFIND when multiple input PDBs were to be read; while all were May 11, 2017 read, only one was output. Thanks to David A. Sadler, Optum, USA. Change 35.116 35.04 only. The TYPSMVJE test was in TESSUSR1 but should JCLTEST9 have been in TESSOTHR member, so TESSUSR1 step failed May 16, 2017 because //MVJEIN DD was not found in that step's JCL. Add //MVJEIN DD DUMMY to the TESSUSR1 step. TYPSQACS replaced TESSQACS in //TESTQAPM step.
Thanks to Rod Feak, MIB, Inc., USA. Change 35.115 Reserved Change Number. Change 35.114 New variables added to TYPE749 PCIE data found in SMF VMAC74 manual refresh: May 10, 2017 R749FLAG='VALIDITY*FLAG' R749NET1='1ST*PORT*PNET ID' R749NET2='2ND*PORT*PNET ID' R749DBYX='BYTES*TRANSMITTED*BY PCIE*FUNCTION' Change 35.113 MXG 35.04 only, TYPE70 SHARE weights wrong, although the VMAC7072 PDB.ASUMCELP values were correct and recommended for the May 10, 2017 analysis of LPAR weights. Thanks to Andrew Petersen, CSC, AUSTRALIA. Change 35.112 -MXG 35.04 Only, only with variable names longer than 32 VGETSORT bytes. ERROR Truncated SORTBY variable name not found. May 9, 2017 The LENGTH for the new SORT variables is $32 now. May 11, 2017 VGETSORT is used in BLDSMPDB, UTILROLL and MULTIPDB. -VGETSORT: Cosmetic, UNINIT variable NOBS message because it was not in the KEEP list, but had no impact. Change 35.111 DB2 12.1, INVALID QLAC SEG ERROR, LENQLAC=218, new VMACDB2 field QLACPRLV was inserted by DB2 CONTINUOUS DELIVERY, VMACDB2H but was unknown to MXG as there was no notification by May 9, 2017 IBM that a field was inserted. MXG detected the change, May 11, 2017 printed the ERROR message, and deleted the record, so some observations in DB2ACCT were not output. The error led to the discovery of an updated DB2 MACLIB with this text in DSNDQLAC member: e26995 Continuous Delivery. Product functional/build level. QLACPRLV. s28617 but a search for s28617 discovered nothing. Only a search for the new field, QLACPRLV found it was added. But there were no other references to the s28617 nor e26995 tokens in the other MACLIB members. Of course, now that I know this new field name, Google found both fields referenced in APAR PI74456: "IFCID 3 accounting information will now provide the partner's functional/service/build level in a new QLACPRLV field." "IFCID 365 location statistics information will now provide the partner's functional/service/build level in a new QLSTPRLV field." The real issue raised with DB2 support and unanswered as of this writing is: HOW AM I SUPPOSED TO KNOW that fields were inserted by Continuous Delivery.
QWHS_MOD_LVL='MOD LEVEL FOR*CONTINUOUS*DELIVERY' QWHS_REC_INCOMPAT='INCOMPATIBLE*CHANGE*VALUE' QWHS_REC_COMPAT='COMPATIBLE*CHANGE*VALUE' QWHS_REC_VALIDITY='CHECK*NEEDED FOR*INCOMPAT*COMPAT' The current MOD_LVL is V12R1M100 and MXG's COMPAT and INCOMPAT count of changes is zero before and after this INCOMPAT change. It is unclear how these fields could be used, since they are after the record was changed. Thanks to Dennis Gaetner, Fiduciagad, GERMANY. Thanks to Sieghart Seith, Fiduciagad, GERMANY.
VMACPRPR FAILURE because the DCB attributes were set for SMF, but May 9, 2017 PRISMAPR input records are FB/256/27904. Thanks to Gene Heikkinen, Blue Cross Minnesota, USA. Change 35.109 Variables SM1209EX/EY/EZ/FA were accidentally dropped by VMAC120 Change 35.024 from dataset TYP120R; you can correct with May 5, 2017 MACRO _KT1209R SM1209EX SM1209EY SM1209EZ SM1209FA % in your IMACKEEP tailoring member until you update MXG. Thanks to Larry A. Gray, Lowes, USA.
ANALID argument, if you invoke %ANALID yourself, but BUILDPDB's VMXGINIT invocation is internal, so this new macro variable May 5, 2017 %LET MXGTITLEANALID=SMF RECORDS AUDIT REPORT; Thanks to Robert Chavez, Florida Power and Light, USA.
VMACIAM
May 4, 2017 Change 35.106 Adds an array of system IDs SYS1-SYS10 and creates new VMXGUOW variables SYSTEMCICS (system of origin of the 110) and May 4, 2017 SYSTEMDB2 (system where the 101 was found) to the PDB.ASUMUOW dataset. Change 35.105 The CICS duration fields are now formatted TIME16.6 VMAC110 to show the full resolution to the microsecond. FORMATS May 4, 2017 only impact the printed/displayed value of the variable.
VMACEDGR for RVCOMPRAT and RVPHYUSED when values can't be derived. May 2, 2017 Only warning and hex dumps were printed; the output data sets were correctly built; this change suppressed the log messages when the values are asterisks. Thanks to Craig Collins, State of Wisconsin, USA. Change 35.103 If you specified "defer=yes" in lower case and the input VMXGSET was on tape you got a 413 ABEND because the compare to Apr 28, 2017 defer= was comparing to upper case. Not reported, found.
VMACVMXA INPUT STSI $VARYING255 STSILEN @; failed when STSILEN was Apr 28,2017 greater than 255; increased to 512 in VXMTRTOP. -Also 6.3, VXIODVSW code didn't protect the undocumented extra 4 bytes. And these new variables are now created: LANFORW ='LAN*FORWARDING*FLAGS' OASPORTN ='OSA*PORT*NUMBER' ACCTYPE ='ACCESS*LIST*TYPE*FLAG' -New variables in VXSTSYG dataset: RCCSCAPF='BFP*ZIP*CAPABILITY' RCCCCAPF='BFP*CP*CAPABILITY' RCCNCAPF='NOMINAL*CP*CAPABILITY' SSI1PCPS='CP CORE*SPEED*CYCLE*PER MICRO' SSI1SCPS='ZIP CORE*SPEED*CYCLE*PER MICRO' -New variables in VXSTSYG dataset: RSAWRTHROTS='TIMES*LIMITED*PAGING*BANDWIDTH' RSAPRTHROTS='TIMES*PARTIAL*WRITE*THROTTLE' RSANDMREC='GLOBAL*RECLAIM*TASK*INITIATED' RSANDMRND='NDMBKS*RETURNED*GLOBAL*RECLAIM' -New variables in VXSTORSP dataset: PLSNDMRQ='FROM*RECYCLE' PLSNDMLO='NDMBK REQS*RECYCLE*LOCAL*SUPPLIED' PLSNDMGL='NDMBK REQS*RECYCLE*GLOBAL*SUPPLIED' PLSNDMG2L='NDMBKS MOVED*GLOBAL*TO LOCAL' PLSNDMDX='TIMES*TASK*RETURNED*CHAIN*TO LOCAL' PLSNDMRET='NDMBKS*RETURNED*TO LOCAL' PLSNDML2G='NMDBKS*MOVED*LOCAL TO*GLOBAL' PLSNDMREL='NMDBKS*RETURNED*TO FREE' PLSNDMREC='TIME*LOCAL*RECLAIM*INITIATED' PLSNDMRND='NDMBKS*RETURNED*TO FREE*LCL RECLAIM' PFXCLPLCNT='FRAMES*CLEARED*LOCAL*AVAIL*PLUS' PLSCLALLO='CLEARED*LOCAL*AVAILABLE*LLOW THRESH' PLSCLALHI='CLEARED LOCAL AVAILABLE*LHIGH THRESH' PLSCLALADDED='FRAMES*ADDED TO*CLEARED*LOCAL' PLSCLALFWREMOVED='FRAMES*ADDED TO*PROCESSED*LIST' PLSCLALTRIMMED='FRAMES*TRIMMED*CLEARED*LOCAL' PLSFPPFENTERED='TIMES*FPP FAULT*ENTERED*FOR GUEST' PLSFPPFSUCCESS='TIMES*FPP FAULT*SUCCESSFUL' PLSCPPFENTERED='TIMES*FPP FAULT*ENTERED*FOR CP' PLSCPPFSUCCESS='TIMES*FPP FAULT*SUCCESSFUL*FOR CP' PLSCPPFMDC='TIMES*FPP FAULT*EXITED*CACHE' PLSCLALREQUESTS1='TIMES*FRAME REQ*CLEARED*AVAIL LIST' PLSCLALUNFILLED1='TIMES*REQUEST*FOUND LAL*EMPTY' PLSCLALREPLENOD='TIMES*REQUEST*FOUND CLA*EMPTY*DEMAND' PLSCLALUNFILLED2='TIMES*REQUEST*FOUND CLA*EMPTY*BEFORE' PLSCLALREQUESTS2='TIMES*REQUEST*FROM*CLA LIST' PLSCLALUNFILLED3='TIMES*REQUEST*FOUND CLA*EMPTY*BEFORE3' PLSBGCTM='CPU TIME*BACKGROUND*REPLEN*CL GAL' PLSCGALREQUESTS='TIMES*WANTED*MOVE FRAMES*CGAL TO CLLA' PLSCGALFRAMESR='FRAMES*WANTED*MOVE*CGAL TO CLLA' PLSCGALMOVED='FRAMES*MOVED*CGAL TO CLLA' PLSCGALNOLOCK='TIMES NOT*MOVED*CGAL TO CLLA*LOCK' PLSCGALUNFILLEDN='FRAMES NOT*MOVED*CGAL TO CLLA*FILL' PLSCGALWANTING='TIMES*CGLA*INSUFFICIENT*FRAMES' PLSCGALUNFILLEDW='FRAMES NOT*MOVED*CGAL TO CLAL*DW' PLSULALCNT='FRAMES ON*UNCLEARED*LAL' PLSULALLO='UNCLEARED*LOCAL*AVAILABLE*LIST LOW TH' PLSULALHI='UNCLEARED*LOCAL*AVAILABLE*LIST HI TH' PLSULALREQUESTS1='FRAME REQ*UNCLEARED*LAL' PLSULALUNFILLED1='TIMES REQ*UNCLEAR LAL*BEFORE*ATTEMPT' PLSULALREPLENOD='TIMES REQ*UNCLEAR LAL*AND*ATTEMPT' PLSULALUNFILLED2='TIMES REQ*UNCLEAR LAL*BEFORE*FILLED' PLSSWPROCLCNT='FRAMES*ON THE*SOFTWARE*PROCESSED LIST' Thanks to James T. Barton, Veterans Administration, USA. Change 35.101 New parameter OUTCODE= lets you insert some code just VMXGGETM prior to the end of VMXGGETM. Should be complete data Apr 28,2017 or PROC STEPS. Thanks to Craig Collins, State of Wisconsin, USA. Change 35.100 SYSLOG code has been updated; the old code was 2016. SYSLOG
May 2, 2017 Thanks to Karl Lasecki, Chemical Abstracts Service, USA. Change 35.099 ANALID relocate. Change 35.098 RMF III Filter enhancements. ADOCRMFV -Enhancements for character data filtering for RMF Monitor ASMRMFV III CPU (Processor Data Control Block), CPC (CPC Data VMACRMFV Control Block), improved message RMFV029* DEAD SPACE Apr 26, 2017 controls, better error message formats and content, and other usability and performance gains. -1 new character filter is added to support CPU entry selection from this table to the RMFBSAM output file. This filter is effective only if the CPU or CPC table is selected and applies to BOTH tables. New Keyword Aliases ------------ ------------------------------------------ CPUSYSTEM= CPUSYSID=, CPUSID=
CPUAND/CPUOR parameters. Syntax and usage mirrors that used for the existing SYSTEM= parameter for selection by SMF Sysid using ranges and/or patterns.
from this table to the RMFBSAM output file. This filter applies ONLY to the CPC (not the CPU) table.
------------ ------------------------------------------ CPCSYSTEM= CPCSYSID=, CPCSID=
CPCAND/CPCOR parameters. Syntax and usage mirrors that used for the existing SYSTEM= parameter for selection by SMF Sysid using ranges and/or patterns.
The new CPCSYSTEM= parameter may appeal to large installations running multiple z/OS LPARs on a CEC (Central Electronic Complex) and using RMF Monitor III gathering data on several of them.
Complex Data Block) table for each instance of RMF Monitor III on a given CEC except if the LPAR is a z/VM guest.
Processor data for every image on the CEC whether it be a z/OS LPAR or not. There is no RMF III option to turn off CPC data collection, so this parameter is an alternative. As long as the RMF Monitor III MINTIME, CYCLE, and SYNC options are identical redundant CPC tables for every RMF Monitor III native (non-guest) LPAR on the CEC are created at every MINTIME interval. The MXG PDB build will create a ZRBLCP observation for each Logical Processor for each LPAR for every MINTIME interval. For installations with several RMF Monitor III LPARs on a CEC this can result in a lot of extra, but not useful duplicate SAS ZRBLCP observations.
Technique" for more details on use of CPCSYSTEM=. It is a user responsibility to set up CPCSYSTEM= for each CEC configuration properly and to track any LPAR SYSID changes as they occur. An incorrect CPCSYSTEM= specification will result in loss of data in the MXG ZRBLCP SAS data set should the SYSID no longer exist or be misspelled. -CSR (Common Storage Remaining) processing now moves CSR entries in blocks for as many entries that fit to the RMFBSAM output buffer when NO CSR character data filters are used. When these filters were added in MXG Change 34.373 processing changed to move one CSR entry at a time to the output buffer. However, this is an unnecessary overhead if no CSR filtering is in effect. The earlier processing technique is restored for this case.
35.098 level showed about a 1% CPU reduction for CSR processing with no filters used compared to the 34.373 level. This will vary with the number of CSR entries and RMF III VSAM data sets processed.
New fields added to the MXG00 record include: CPUSYSTEM= and CPCSYSTEM= Range/Pattern maximums CPUSYSTEM= and CPCSYSTEM= Range/Pattern table sizes % of available TIOT entries used ASMCPCRX='MAXIMUM*CPCSYSTERM*RANGES' ASMCPCPX='MAXIMUM*CPCSYSTERM*PATTERNS' ASMCPURX='MAXIMUM*CPUSYSTERM*RANGES' ASMCPUPX='MAXIMUM*CPUSYSTERM*PATTERNS' ASMSHSPL='WARNING*LIMIT*PCT SPACE*USE' ASMCPCRS='SIZE*CPCSYSTEM*RANGE*TABLE' ASMCPCPS='SIZE*CPCSYSTEM*PATTERN*TABLE' ASMCPURS='SIZE*CPUSYSTEM*RANGE*TABLE' ASMCPUPS='SIZE*CPUCSYSTEM*PATTERN*TABLE' -TIOT entries used percentage is added to the RMFV000I message. TIOT usage information is grouped on a single report line.
SLIM=) keywords is added to control appearance of the RMFV029* DEAD SPACE message (*=I,W,E,S).
exhausted RMF III indexes even when the VSAM data set usage was relatively high. In this case re-allocating the VSAM data set to make it smaller is not productive. ILIMIT= specifies a percentage in the range of 0 to 100 as a threshold for RMF III VSAM Data Set Header (DSH) indexes usage. The default is 100.
as a threshold for RMF III VSAM Data Set space usage. The default is 95.
SPACE options in effect mean that if all 1110 sample indexes are exhausted in the Data Set Header (DSH) record, but the RMF Monitor III VSAM data set is 95% or more utilized no DEAD SPACE condition is flagged.
of NOINDEXES/INDEXES, NOSPACE/SPACE, ILIMIT=, and SLIMIT= parameters interact.
can suppress it completely with ILIMIT=0 and SLIMIT=0. ILIMIT= and SLIMIT= values in effect are displayed in message RMFV037I. Section 22 RMF III VSAM Data Set Index Usage and Sizing is updated to discuss use of ILIMIT= and SLIMIT= options. -There are now 4 distinct levels for messages that can have variable severity based on the settings of the various existing *ERR= condition keywords:
Setting Suffix Meaning Code ------- ------ ------------------------- --------- IGNORE I Ignore error/continue No change WARN W Warn error/continue 0004 ERROR E Issue error/may continue 0008 ABEND S Issue error/Abend U0998 N/A (1)
Code.
inconsistently may or may not have resulted in an Abend.
--------- ------- ------------------------- ALLOCERR= WARN DYNAMIC ALLOCATION ERRORS ATTRERR= WARN DATA SET ATTRIBUTE ERRORS CATERR= WARN CSI CATALOG LOOKUP ERRORS DEADERR= WARN VSAM DEAD SPACE ERRORS DSIGERR= WARN DSIG ID (DSH) ERRORS DSNERR= WARN CSI DSNAME LOOKUP ERRORS DUPERR= WARN DUPLICATE DSNAME ERRORS EMPTYERR= IGNORE EMPTY VSAM DATA SET ERRORS PATTERR= ABEND PATTERN AND/OR RANGE ERRORS READERR= ABEND VSAM READ I/O ERRORS TABERR= WARN RMF III TABLE VALIDATE ERRORS TYPEERR= WARN DATA SET TYPE ERRORS RCERR= WARN REPORT CLASS FIND ERRORS RGERR= WARN RESOURCE GROUP FIND ERRORS SCERR= WARN SERVICE CLASS FIND ERRORS WLERR= WARN WORKLOAD FIND ERRORS -ASMRMFV will now generate the correct message format for variable severity messages during assembly based on the defaults above. This avoids unnecessary tailoring during ASMRMFV initiation. Any user overrides of the above settings will still require tailoring of related messages by ASMRMFV during start up. -Internal error message generation interface updated for following messages: RMFV004E, RMFV005*, RMFV007S, RMFV034S, RMFV035*, RMFV056* (*=I,W,E,S) Improvements for these messages include: 3 separate error messages subroutines replaced by one for code path length reduction. Extraneous blanks in these error messages eliminated for better legibility. Clearer and less cryptic error descriptions. -RMFV006E message had incorrect timestamp when FROMDATE= exceeded TODATE= -RMFV007S message missing DDNAME when RMFBSAM DD was not present.
not available for a failed function or service.
range, i.e. DOW=day1:day2 to examine the last half of the day of week range if an error is found in the first half. Also leading and trailing colons are stripped before length checking.
and former Section 31 Bibliography is now Section 32. -Several documentation Sections are updated to support the above changes: Section 0 "Contents" Section 5 "Input Data Selection Parameters" Section 6 "Report Control Parameters" Section 8 "Error Handling Parameters" Section 12 "Messages" Section 13 "Filtered Records" Section 16 "Return Codes" Section 25 "Ranges and Patterns" Section 30 "CPC Data Relief Technique" Section 31 "Summary" Section 32 "Bibliography"
VMACCIMS formatted DATETIME25.6 to display full microseconds. The Apr 25, 2017 other datetimes are limited to DATETIME21.2 resolution. Thanks to Jan Tielemans, KBC, BELGIUM. Change 35.096 New UTILROLL utility combines all SAS datasets from one UTILROLL or more SAS data libraries (think "hourly" PDB runs) into JCLROLL1 another LIBNAME (think "daily" PDB), either concatenating JCLROLL2 the new data, or interleaving to preserve the sort order. VGETSORT If the ROLLTO LIBNAME is empty or the dataset being added Apr 24, 2017 added to does not exist the code will ignore the ROLLTO LIBNAME and use only the ROLLFROM to create the ROLLTO datasets. -VGETSORT could return bad information if there was a variable name longer than 8 bytes. Length was increased to 32 and NOBS ENG were added to the output.
VMAC7072 RMF Workload report values: Apr 24, 2017 APPLCP ='APPL PCT*OF 1 CPU*CPU TIME*ON CP' APPLAAPCP='APPL PCT*OF 1 CPU*ZAAP ELIG*ON CP' APPLIIPCP='APPL PCT*OF 1 CPU*ZIIP ELIG*ON CP' APPLAAP ='APPL PCT*OF 1 CPU*CPU TIME*ON ZAAP' APPLIIP ='APPL PCT*OF 1 CPU*CPU TIME*ON ZIIP' with this code: IF R723MCF GT 0 THEN APPLCP= 100*CPUTM/(DURATM*R723MCF); IF R723MCFI GT 0 THEN DO; APPLAAPCP=100*CPUIFETM/(DURATM*R723MCFI); APPLAAP= 100*(256*CPUIFATM/R723NFFI)/(DURATM*R723MCFI); END;
IF R723MCFS GT 0 THEN DO; APPLIIPCP=100*CPUZIETM/(DURATM*R723MCFS); APPLIIP= 100*(256*CPUZIPTM/R723NFFS)/(DURATM*R723MCFS); END;
Thanks to Ray Bole, IBM Global Services, USA. Change 35.094 Support for BMC Mainview for Java Environment creates EXMVJE01 DDDDDD MXG MXG EXMVJE02 DATASET DATASET DATASET EXMVJE04 SUFFIX NAME LABEL EXMVJE07 EXMVJE08 MVJE01 MVJE01 JMX MEMORY SUMMARY EXMVJE12 MVJE02 MVJE02 JMX THREAD SUMMARY IMACMVJE MVJE04 MVJE04 JMX CLASS LOADING SUMMARY TYPEMVJE MVJE07 MVJE07 JMX GARBAGE COLLECTION SUMMARY TYPSMVJE MVJE08 MVJE08 JMX MEMORY POOLS VMACMVJE MVJE12 MVJE12 JMX CPU USAGE VMXGINIT
Apr 25, 2017 Change 35.093 MXG 35.03 only, variable PLATBUSY in TYPE70/RMFINTRV and VMAC7072 PCTOFHDW in RMFINTRV were incorrect after Change 35.064 Apr 22,2017 revised the SHARE calculations. Thanks to Paul Naddeo, FISERV, USA. Thanks to Robin Hanley, FISERV, USA. Thanks to David Bixler, FISERV, USA. Change 35.092 Support for z/VM 64 (INCOMPATIBLE). VMACVMXA -Dataset VXSYTCUG new variables Apr 20,2017 SSI2MTIF='MULTITHREADING*CONFIGURATION' SSI2MTGF='MULTITHREAD*GENERAL*PROC*CONFIG' SSI2MTID='MULTITHREAD*MAX*TID' LCUTCTOD='DATETIME*WHEN CORE*INFO*FETCHED' -Dataset VXMTRSYS new variable CALFLAG3-'MISCELLANEOUS*FLAGS*/ -Datasets VXUSEACT and VXUSELOF new variables VMDTTIME_MT1='TOTAL*MT-1*EQUIVALENT*TIME' VMDVTIME_MT1='RUN*MT-1*EQUIVALENT*TIME' VMDVTMP_MT1 ='EQUIV*MT-1*VTIME*ON PRIMARY' VMDTTTP_MT1 ='EQUIV*MT-1*VTIME+SIM*PRIMARY' VMDVTMS_MT1 ='EQUIV*MT-1*VTIME*ON SECNDRY' VMDTTMS_MT1 ='EQUIV*MT-1*VTIME+SIM*SECNDRY' VMAVTMP_PRO ='TOTAL*MT-1*PRORATED*CORE*PRI' VMATTMP_PRO ='RUN*MT-1*PRORATED*CORE*PRI' VMAVTMS_PRO ='TOTAL*MT-1*PRORATED*CORE*SEC' VMATTMS_PRO ='RUN*MT-1*PRORATED*CORE*SEC' PROBITS='PRORATED*CORE*TIME*BITS' -Datasets VXPRCPUP new variables WHIOCAPV='MAX*CORES*PERMITTED' WHIOCTVR='CALCULATED*T/V*RATIO' WHIOPTVR='CEILING*PROJECTED*T/V*RATIO' SRXTVCNF='CONFIDENCE*PERCENTAGE' CALTVALG='CEILING*PROJECTION*ALGORITHM' WHIOTVCT='CEILING*PROJECTION*VALID*SAMPLES*/ WHIOGCPV='MAXIMUM*AGGREGATE*CORES' -Datasets VXIODVSW new variables VQSAFLAG ='LACP*CONFIGURATION' NIDLAPRE ='PREVIOUS*LOAD*BALANCE' NIDLACUR ='CURRENT*LOAD*BALANCE' NID_TOTPFCNT='PREVIOUS*LOAD*BALANCE*FRAMES' -New segments IODPAD,IODPON,MTRPCI are not yet supported. -These 6.4 segments don't exist in my test data so these Yüklə 28,67 Mb. Dostları ilə paylaş:
|