Interpreting Raw SQL_TRACE output (Doc ID 39817.1)

PPLIES TO: Oracle Database Cloud Schema Service - Version N/A and later Oracle Database Exadata Express Cloud Service - Version N/A and later Oracle Database Exadata Cloud Machine - Version N/A and later Oracle Cloud Infrastructure - Database Service - Version N/A and later Oracle Database Backup Service - Version N/A and later Information in this document applies to any platform. PURPOSE This is a reference document which summarizes the output format of the raw SQL_TRACE output file. NOTE: The format may vary slightly between releases. SCOPE DBA(s) and SQL Performance Tuning experts who need additional information on how to interpret a raw SQL_TRACE output. DETAILS NOTE: Square brackets : ( [ and ] ) are used to enclose some sections where specific text has been removed. These are not in the original text and are included for visibility. So, for example : WAIT #1: nam="event name" ela=0 p1=0 p2=0 p3=0 in the trace would be shown as : WAIT #[CURSOR]: nam="event name" ela=0 p1=0 p2=0 p3=0 in this document with [CURSOR] replacing the cursor number. Standard Output APPNAME mod='%s' mh=%lu act='%s' ah=%lu APPNAME Application name setting. This only applies to Oracle 7.2 and above. This can be set by using the DBMS_APPLICATION_INFO package. See <Note:30366.1> mod Module name. mh Module hash value. act Action. ah Action hash value. PARSING IN CURSOR #[CURSOR] len=X dep=X uid=X oct=X lid=X tim=X hv=X ad='X' [statement] END OF STMT [CURSOR] Cursor number. len Length of SQL statement. dep Recursive depth of the cursor. uid Schema id under which SQL was parsed. oct Oracle command type. lid Privilege user id. tim Timestamp. Pre-Oracle9i, the times recorded by Oracle only have a resolution of 1/100th of a second (10mS). As of Oracle9i some times are available to microsecond accuracy (1/1,000,000th of a second). The timestamp can be used to determine times between points in the trace file. The value is the value in V$TIMER when the line was written. The timer has platform-specific implementation differences. If there are TIMESTAMPS in the file you can use the difference between 'tim' values to determine an absolute time. hv Hash id. ad SQLTEXT address (see views V$SQLAREA <Note:43761.1> and V$SQLTEXT <Note:43764.1>). [statement] The actual SQL statement being parsed. PARSE ERROR #%d:len=%ld dep=%d uid=%ld oct=%d lid=%ld tim=%lu err=%d [statement] ... PARSE ERROR In Oracle 7.2+ parse errors are reported to the trace file. len Length of SQL statement. dep Recursive depth of the statement uid User id. oct Oracle command type (if known). lid Privilege user id. tim Timestamp. err Oracle error code (e.g. ORA-XXXXX) reported [statement] The SQL statement that errored. If this contains a password, the statement is truncated as indicated by '...' at the end. PARSE #[CURSOR]:c=0,e=0,p=0,cr=0,cu=0,mis=0,r=0,dep=0,og=4,tim=0 EXEC #[CURSOR]:c=0,e=0,p=0,cr=0,cu=0,mis=0,r=0,dep=0,og=4,tim=0 FETCH #[CURSOR]:c=0,e=0,p=0,cr=0,cu=0,mis=0,r=0,dep=0,og=4,tim=0 UNMAP #[CURSOR]:c=0,e=0,p=0,cr=0,cu=0,mis=0,r=0,dep=0,og=4,tim=0 - OPERATIONS: PARSE Parse a statement. EXEC Execute a pre-parsed statement. FETCH Fetch rows from a cursor. UNMAP If the cursor uses a temporary table, when the cursor is closed you see an UNMAP when we free up the temporary table locks.(Ie: free the lock, delete the state object, free the temp segment) In tkprof, UNMAP stats get added to the EXECUTE statistics. SORT UNMAP As above, but for OS file sorts or TEMP table segments. c CPU time (centiseconds prior to 9i, microseconds rounded to centiseconds granularity on 9i and above) e Elapsed time (centiseconds prior to 9i, microseconds thereafter) p Number of physical reads. cr Number of buffers retrieved for CR reads. cu Number of buffers retrieved in current mode. mis Cursor missed in the cache. r Number of rows processed. dep Recursive call depth (0 = user SQL, >0 = recursive). og Optimizer goal: 1=All_Rows, 2=First_Rows, 3=Rule, 4=Choose tim Timestamp (100th's of sec < Oracle9i, Microseconds thereafter) Use this to determine the time between any 2 operations. CLOSE #[CURSOR]:c=%u e=%u dep=%d type=%u tim=%u CLOSE cursor is closed c CPU time (centiseconds prior to 9i, microseconds rounded to centiseconds granularity on 9i and above) e Elapsed time (centiseconds prior to 9i, microseconds thereafter) dep Recursive depth of the cursor type type of close operation tim Timestamp (100th's of sec < Oracle9i, Microseconds thereafter) Use this to determine the time between any 2 operations. ERROR #%d:err=%d tim=%lu ERROR Error shown after an execution or fetch error. err Oracle error code (e.g. ORA-XXXXX) at the top of the stack. tim Timestamp. STAT #[CURSOR] id=N cnt=0 [pid=0 pos=0 obj=0 op='SORT AGGREGATE '] STAT Lines report explain plan statistics for the numbered [CURSOR]. [CURSOR] Cursor which the statistics apply to. id Line of the explain plan which the row count applies to (starts at line 1). This is effectively the row source row count for all row sources in the execution tree. cnt Number of rows for this row source. As of 7.3.3 the items in '[...]' are also reported: pid Parent id of this row source. pos Position in explain plan. obj Object id of row source (if this is a base object). op='...' The row source access operation. These let you know the 'run time' explain plan. 11g and above: In 11g the "op=" section has been expanded so that STAT lines contain more information. For example: STAT #[CURSOR] id=1 cnt=1 pid=0 pos=1 obj=0 op='SORT AGGREGATE (cr=6 pr=0 pw=0 time=483 us)' STAT #[CURSOR] id=2 cnt=14 pid=1 pos=1 obj=75510 op='TABLE ACCESS FULL EMP (cr=6 pr=0 pw=0 time=432 us cost=2 size=0 card=14)' cr Consistent reads. pr Physical reads. pw Physical writes. time Elapsed time taken by the operation (in microseconds). cost Optimizer cost of the operation. size Estimated size (in bytes). card Estimated Optimizer Cardinality. XCTEND rlbk=%d rd_only=%d XCTEND A transaction end marker. rlbk 1 if a rollback was performed, 0 if no rollback (commit). rd_only 1 if transaction was read only, 0 if changes occurred. RPC Remote Procedure Calls These are logged and interpreted in a similar manner, here is an example: RPC CALL:FUNCTION APPS.ARP_TAX_VENDOR.IS_GEOCODE_VALID(P_GEOCODE IN VARCHAR2) RETURN BOOLEAN; RPC BINDS: bind 0: dty=1 bfp=60000000005e4db8 flg=08 avl=09 mxl=30 val="161630280" bind 1: dty=3 bfp=60000000005e4df8 flg=02 avl=04 mxl=04 val=00 RPC EXEC:c=0,e=162 Additional Wait/Bind Tracing Output The items below are only output if WAITS or BINDS are being traced. The following notes explain how to get a sql trace file with WAIT and/or BIND information: Document 376442.1 How To Collect 10046 Trace (SQL_TRACE) Diagnostics for Performance Issues Document 21154.1 EVENT: 10046 "enable SQL statement tracing (including binds/waits)" BINDS #[BIND]: bind 0: dty=2 mxl=22(22) mal=00 scl=00 pre=00 oacflg=03 oacfl2=0 size=24 offset=0 bfp=02fedb44 bln=22 avl=00 flg=05 value=10 [BIND] Variables bound to a cursor. bind N The bind position being bound. dty Data type mxl Maximum length of the bind variable (private max len in paren). mal Array length. scl Scale. pre Precision. oacflg Special flag indicating bind options oacflg2 Continuation of oacflg size Amount of memory to be allocated for this chunk offset Offset into this chunk for this bind buffer bfp Bind address. bln Bind buffer length. avl Actual value length (array length too). flg Special flag indicating bind status value The actual value of the bind variable. Numbers show the numeric value, strings show the string etc... If a message such as : "bind 6: (No oacdef for this bind)" is seen then this means that there is no bind data defined at the time of the dump. This can mean that the cursor is being dumped before the cursor has been bound, for example. It is also possible to see "bind 6: (No oacdef for this bind)", if no separate bind buffer exists for some other reason. WAIT #[CURSOR]: nam="event name" ela=0 p1=0 p2=0 p3=0 WAIT An event that we waited for. nam What was being waited for . The wait events here are the same as are seen in view V$SESSION_WAIT <Note:43718.1>. For any Oracle release a full list of wait events and the values in P1, P2 and P3 below can be seen in view V$EVENT_NAME <Note:43721.1> ela Elapsed time for the operation. p1 P1 for the given wait event. p2 P2 for the given wait event. p3 P3 for the given wait event. NOTE: Starting with 11g, Traces have started to replace P1, P2 and P3 with more meaningful phrases related to the actual event in question 11g WAIT Examples (Index Scan) from Oracle11g (elapsed time in microseconds): WAIT #6: nam='db file sequential read' ela= 8458 file#=110 block#=63682 blocks=1 obj#=221 tim=506028963546 Explanation: Completed WAITing under CURSOR no 6 for "db file sequential read" We waited 8458 microseconds i.e. approx. 8.5 milliseconds For a read of: File 110, start block 63682, for 1 Oracle block of Object number 221. Timestamp was 506028963546 WAIT #1: nam='library cache: mutex X' ela= 814 idn=3606132107 value=3302829850624 where=4 obj#=-1 tim=995364327604 Explanation: Completed WAITing under CURSOR no 1 for "library cache: mutex X" We waited 814 microseconds i.e. approx. 0.8 milliseconds To get an eXclusive library cache latch with Identifier 3606132107 value 3302829850624 location 4 It was not associated with any particular object (obj#=-1) Timestamp 995364327604 9i WAIT Example (Full Table Scan) from Oracle9i (elapsed time in microseconds): WAIT #1: nam="db file scattered read" ela=5010 p1=4 p2=1435 p3=25 Explanation: Completed WAITing under CURSOR no 1 for "db file scattered read". We waited 5010 microseconds i.e. approx. 5 milliseconds for a read of: File 4, start block 1435, for 25 Oracle blocks 8i WAIT Example (Index Scan) from Oracle8i (elapsed time in centiseconds): WAIT #1: nam="db file sequential read" ela=4 p1=4 p2=1224 p3=1 Explanation: Completed WAITing under CURSOR no 1 for "db file sequential read". We waited 0.04 seconds for a single block read (p3=1) from file 4, block 1224 NOTE: WAIT lines in trace refer to completed waits. The current WAIT cannot be seen in the standard SQL_TRACE trace output. A trace providing a PROCESS STATE or similar is required to see the current wait. REFERENCES NOTE:21154.1 - EVENT: 10046 "enable SQL statement tracing (including binds/waits)" NOTE:30366.1 - PACKAGE DBMS_APPLICATION_INFO Specification NOTE:376442.1 - * How To Collect 10046 Trace (SQL_TRACE) Diagnostics for Performance Issues NOTE:43718.1 - VIEW: "V$SESSION_WAIT" Reference Note NOTE:43721.1 - VIEW: "V$EVENT_NAME" Reference Note NOTE:43761.1 - VIEW: "V$SQLAREA" Reference Note NOTE:43764.1 - VIEW: "V$SQLTEXT" Reference Note