High Version Count Issues（SQL高Version Count） (文件 ID 296377.1)

Troubleshooting: High Version Count Issues (文件 ID 296377.1)

In this Document

Applies to:

Purpose

The Troubleshooting Guide is provided to assist in debugging SQL sharing issues. When possible, diagnostic tools are included in the document to assist in troubleshooting problems. This document does not contain bugs/patches as these topics are addressed in the articles referenced at the bottom of this document.

Ask Questions, Get Help, And Share Your Experiences With This Article

Would you like to explore this topic further with other Oracle Customers, Oracle Employees, and Industry Experts?

Click here to join the discussion where you can ask questions, get help from others, and share your experiences with this specific article.
Discover discussions about other articles and helpful subjects by clicking to access the main My Oracle Support Community page for Database Tuning.

Troubleshooting Steps

What is a 'High' Version Count? 

There is no definitive definition of what a 'High' number of versions for a particular cursor is, different systems may be able to deal with different ranges of versions, However, AWR reports start reporting versions over 20 for a particular cursor and so that is as good an indicator of a potential problem as any.



Once you start getting into the hundreds or thousands range, then these are definitely 'High' counts and the causes should be examined and the numbers reduced so as to encourage the SQL to be shared. It is important to understand that sometimes high version counts are expected and not a result of any defect.

What is shared SQL ? 

The first thing to remember is that all SQL is implicitly sharable. When a SQL statement is entered, the RDBMS will create a hash value for text of the statement and that hash value then helps the RDBMS to easily find SQL already in the shared pool. It is not in the scope of this article to discuss this in any great detail, so let's just assume entering a series of text results in a hash value being created.
所有SQL是隱式共享。當輸入SQL語句時,資料庫將會為文字的語句建立一個hash value，該值會幫助RDBMS輕易找到SQL在共享池中的位置。

For instance :- 'select count(*) from emp' hashes to the value 4085390015

We now create a parent cursor for this sql and a single child. It does not matter that a SQL statement may never be shared - when it is first parsed a parent and a single child are created. The easy way to think of this is that the PARENT cursor is a representation of the hash value and the child cursor(s) represent the metadata for that SQL

What is 'SQL Metadata'?

Metadata is all the information which enables a statement to run. For instance, in the example I have given EMP is owned by scott and therefore has an OBJECT_ID which points to the EMP table owned by this user. When the user SCOTT logged in, optimizer parameters are initialised in that session for use by the statement, so this too is used by the optimizer and is therefore metadata. There are other examples of Metadata which will be mentioned further in this document.
Metadata包含執行SQL的所有資訊。

Let's say this session logs out and back in again now. It then runs the same command again (as the same user). This time we already have the SQL in the shared pool (but we don't know this yet). What we do is hash the statement and then search for that hash value in the shared pool. If we find it, we can then search through the children to determine if any of them are usable by us (ie the metadata is the same). If it is, then we can share that SQL statement
I would still have one version of that SQL in the shared pool because the metadata enabled me to share the statement with the already existent child. The fundamentals are that the parent is not shared, it is the children which determine shareability.

Now - another user 'TEST' has it's own version of EMP. If that user was to now run the select statement above then what would happen is :-

1. The statement is hashed - it is hashed to the value 4085390015
2. The SQL will be found in the shared pool as it already exists
3. The children are scanned (at this point we have one child)
4. Because the OBJECT_ID of the EMP table owned by TEST is different the OBJECT_ID owned by scott we have a 'mismatch'

(Essentially, what happens here is that we have a linked list of children which we move through in turn, comparing the metadata of the current SQL with that of all the children. If there were 100 children then we would scan each of them (looking for a possible mismatch and moving on) until we found one we could share. If we cannot share any (ie. have exhausted the list of children) then we need to create a new child)

5. We therefore have to create a new child - we now have 1 PARENT and 2 CHILDREN.

Why should I be concerned about 'High' Versions?

Unnecessary non-sharing of SQL, and the resultant versions of SQL, is a primary cause of library cache contention. Contention reduces the performance of your database and, in extreme cases, can cause it to appear to 'hang'. When you have unnecessary versions of a cursor, each time that cursor is executed, the parse engine has to search through the list of versions to see which is the cursor that you want. This wastes CPU cycles that you could be using on something else.

How do I see the versions and why they are not shared ?

The easiest way to get version information in a clear format is to use the script in the following article:

To find the reasons for mismatches see the following section: What do the reasons given in v$SQL_SHARED_CURSOR mean?

If you are unable to use that script then you can select the same information from the base views as illustrated in the examples below.

Lets use the example above and take a look at what SQL we can use to see this in the shared pool.

SCOTT runs select count(*) from emp

I can now run the following to see the PARENT statement and it's hash value and address

To see the CHILDREN (I expect to see 1 at this point) :-

• Version 9.2.X.X and below :
  
  select * from v$sql_shared_cursor where kglhdpar = '0000000386BC2E58'
• Version 10.0.X.X and above:
  
  select * from v$sql_shared_cursor where address = '0000000386BC2E58'

Output:

ADDRESS          KGLHDPAR         U S O O S L S E B P I S T A B D L T R I I R L I O S M U T N F 
---------------- ---------------- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 
0000000386BC2D08 0000000386BC2E58 N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N 

We can see we have a single child (ADDRESS 0000000386BC2D08).
The mismatch information (U S O O S L etc) is all N because this is the first child. Now, if I log in as another user and run the same select (select count(*) from emp) and look again I will get the following output:-

ADDRESS          KGLHDPAR         U S O O S L S E B P I S T A B D L T R I I R L I O S M U T N F 
---------------- ---------------- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 
0000000386BC2D08 0000000386BC2E58 N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N 
0000000386A91AA0 0000000386BC2E58 N N N N N N N N N N N N N Y N N N Y N N N N N N N N N N N N N 

We can now see the 2nd child ( 0000000386A91AA0) and also the reasons why it could not be shared with the first (The 'Y's denote a mismatch). The reasons are:

(1) AUTH_CHECK_MISMATCH and
(2) TRANSLATION_MISMATCH

This is because the objects under my new user do not map to those of SCOTT (the current child). A mismatch occurs because I cannot access SCOTTs objects and translation fails since we have different object_ids for the objects in each of our schemas. 

What do the reasons given in v$SQL_SHARED_CURSOR mean?

Below are the list of reasons as well as some worked examples (Those denoted by ** are the ones most often seen) :-

• UNBOUND_CURSOR

  The existing child cursor was not fully built (in other words, it was not optimized)

• SQL_TYPE_MISMATCH

  The SQL type does not match the existing child cursor

• **OPTIMIZER_MISMATCH

  The optimizer environment does not match the existing child cursor (The optimizer mode has changed and therefore the existing child cannot be re-used).
  For example:
  
  
  select count(*) from emp;      ->> 1 PARENT, 1 CHILD
  alter session set optimizer_mode=ALL_ROWS
  select count(*) from emp;      ->> 1 PARENT, 2 CHILDREN 
   
  
  Note: The behavior applies with the setting of trace events. for example, if I turned on tracing with 10046 than I would get an OPTIMIZER_MISMATCH and another child cursor

• OUTLINE_MISMATCH

  The outlines do not match the existing child cursor. For example, if a user had created stored outlines previously for this command and they were stored in separate categories (say "OUTLINES1" and "OUTLINES2"), if they then executed the following:
  
  alter session set use_stored_outlines = OUTLINES1;
  select count(*) from emp;
  
  alter session set use_stored_oulines= OUTLINES2;
  select count(*) from emp;
  
  The second execution of the select from emp would create another child since the outline used is different than the first run. This child would be marked as an OUTLINE_MISMATCH.

• STATS_ROW_MISMATCH

  The existing statistics do not match the existing child cursor. Check that 10046/sql_trace is not set on all sessions as this can cause this.

• LITERAL_MISMATCH

  Non-data literal values do not match the existing child cursor

• SEC_DEPTH_MISMATCH

  Security level does not match the existing child cursor

• EXPLAIN_PLAN_CURSOR

  The child cursor is an explain plan cursor and should not be shared. Explain plan statements will generate a new child by default - the mismatch will be this.

• BUFFERED_DML_MISMATCH

  Buffered DML does not match the existing child cursor

• PDML_ENV_MISMATCH

  PDML environment does not match the existing child cursor

• INST_DRTLD_MISMATCH

  Insert direct load does not match the existing child cursor

• SLAVE_QC_MISMATCH

  The existing child cursor is a slave cursor and the new one was issued by the coordinator (or, the existing child cursor was issued by the coordinator and the new one is a slave cursor).

• TYPECHECK_MISMATCH

  The existing child cursor is not fully optimized

• AUTH_CHECK_MISMATCH

  Authorization/translation check failed for the existing child cursor
  The user does not have permission to access the object in any previous version of the cursor. A typical example would be where each user has it's own copy of a table

• **BIND_MISMATCH

  The bind metadata does not match the existing child cursor. For example, in the following, the definition of the bind variable 'a' has changed between the 2 statements:
  
  
  variable a varchar2(100);
  select count(*) from emp where ename = :a ->> 1 PARENT, 1 CHILD
  
  variable a varchar2(400);
  select count(*) from emp where ename = :a ->> 1 PARENT, 2 CHILDREN

• DESCRIBE_MISMATCH

  The type-check heap is not present during the describe for the child cursor

• LANGUAGE_MISMATCH

  The language handle does not match the existing child cursor
  

  
  

• TRANSLATION_MISMATCH

  The base objects of the existing child cursor do not match.
  The definition of the object does not match any current version. Usually this is indicative of the same issue as "AUTH_CHECK_MISMATCH" where the object is different.

• ROW_LEVEL_SEC_MISMATCH

  The row level security policies do not match

• INSUFF_PRIVS

  Insufficient privileges on objects referenced by the existing child cursor

• INSUFF_PRIVS_REM

  Insufficient privileges on remote objects referenced by the existing child cursor

• REMOTE_TRANS_MISMATCH

  The remote base objects of the existing child cursor do not match. For example:
  USER1:
  
  select count(*) from table@remote_db  
  USER2:
  
  select count(*) from table@remote_db  
  Although the SQL is identical, the dblink pointed to by remote_db may be a private dblink which resolves to a different object altogether

• LOGMINER_SESSION_MISMATCH

• INCOMP_LTRL_MISMATCH

• OVERLAP_TIME_MISMATCH

  Error_on_overlap_time mismatch

• SQL_REDIRECT_MISMATCH

  SQL redirection mismatch

• MV_QUERY_GEN_MISMATCH

  Materialized view query generation

• USER_BIND_PEEK_MISMATCH

  User bind peeking mismatch

• TYPCHK_DEP_MISMATCH

  Cursor has type-check dependencies

• NO_TRIGGER_MISMATCH

  No trigger mismatch

• FLASHBACK_CURSOR

  No cursor sharing for flashback

• ANYDATA_TRANSFORMATION

  Anydata transformation change

• INCOMPLETE_CURSOR

  Incomplete cursor. When bind length is upgradeable (i.e. we found a child cursor that matches everything else except that the bind length is not long enough), we mark the old cursor is not usable and build a new one.  This means the version can be ignored.

• TOP_LEVEL_RPI_CURSOR

  Top level/rpi cursor. In a Parallel Query invocation this is expected behaviour (we purposely do not share)

• DIFFERENT_LONG_LENGTH

  Different long length

• LOGICAL_STANDBY_APPLY

  Logical standby apply mismatch

• DIFF_CALL_DURN

  Different call duration

• BIND_UACS_DIFF

  Bind uacs mismatch

• PLSQL_CMP_SWITCHS_DIFF

  PL/SQL compiler switches mismatch

• CURSOR_PARTS_MISMATCH

  Cursor "parts executed" mismatch

• STB_OBJECT_MISMATCH

  STB object different (now exists). For explanation of STB_OBJECT_MISMATCH, please read following blog: https://blogs.oracle.com/optimizer/entry/my_cursor_wasn_t_shared

• ROW_SHIP_MISMATCH

  Row shipping capability mismatch

• PQ_SLAVE_MISMATCH

  PQ slave mismatch If you encounter this reason code and you are using parallel execution (PX), then check you really want to be using it. This mismatch can be caused by running lots of small SQL statements which do not really need PX. Also, if you are on versions prior to 11g you may be hitting

• TOP_LEVEL_DDL_MISMATCH

  Top-level DDL cursor

• MULTI_PX_MISMATCH

  Multi-px and slave-compiled cursor

• BIND_PEEKED_PQ_MISMATCH

  Bind-peeked PQ cursor

• MV_REWRITE_MISMATCH

  MV rewrite cursor

• ROLL_INVALID_MISMATCH

  Rolling invalidation window exceeded. This is caused by the rolling invalidation capability in DBMS_STATS. The child cannot be shared as it's invalidation window is exceeded. See:
  
  
  Document 557661.1  Rolling Cursor Invalidations with DBMS_STATS in Oracle10g

• OPTIMIZER_MODE_MISMATCH

  Optimizer mode mismatch

• PX_MISMATCH

  Parallel query execution mismatch. Refer to the following for known issues where this reason is shown:
  
  Document 1629107.1 Common Bugs Associated with PX_MISMATCH
   

• MV_STALEOBJ_MISMATCH

  Materialixed View stale object mismatch

• FLASHBACK_TABLE_MISMATCH

  Flashback table mismatch

• LITREP_COMP_MISMATCH

  Literal replacement compilation mismatch 

New in 11g :

• PLSQL_DEBUG

  Debug mismatch Session has debugging parameter plsql_debug set to true

• LOAD_OPTIMIZER_STATS

  Load optimizer stats for cursor sharing

• ACL_MISMATCH

  Check ACL mismatch

• FLASHBACK_ARCHIVE_MISMATCH

  Flashback archive mismatch

• LOCK_USER_SCHEMA_FAILED

  Failed to lock user and schema

• REMOTE_MAPPING_MISMATCH

  Remote mapping mismatch

• LOAD_RUNTIME_HEAP_FAILED

  Runtime heap mismatch

• HASH_MATCH_FAILED

  Hash mismatch. Set to "Y" if sharing fails due to a hash mismatch, such as the case with mismatched histogram data or a range predicate marked as unsafe by literal replacement (See )

New in 11.2:

• PURGED_CURSOR

  Cursor marked for purging. The cursor has been marked for purging with dbms_shared_pool.purge

• BIND_LENGTH_UPGRADEABLE

  Bind length upgradeable and could not be shared because a bind variable size was smaller than the new value being inserted (marked as BIND_MISMATCH in earlier versions).

• USE_FEEDBACK_STATS

  Cardinality feedback. Cardinality feedback is being used and therefore a new plan could be formed for the current execution.

• BIND_EQUIV_FAILURE

  The bind value's selectivity does not match that used to optimize the existing child cursor. When adaptive cursor sharing is used and the cursor is bind aware, then if the selectivity is outside of the current ranges and a new plan is desirable then a new child is raised with this as the reason code for non-sharing of the previous plan. For an example, see Document 836256.1. After each execution in the example, run:
  select sql_id, address, child_address, child_number, BIND_EQUIV_FAILURE from v$sql_shared_cursor where sql_id='19sxt3v07nzm4';
  ... once the cursor is marked as bind aware and a second plan is seen then the following will be the resultant output:
  SQL_ID        ADDRESS          CHILD_ADDRESS    CHILD_NUMBER B
  ------------- ---------------- ---------------- ------------ -
  19sxt3v07nzm4 000000007A1C0DE0 000000007A1BF980            0 N
  19sxt3v07nzm4 000000007A1C0DE0 000000007A10DDB0            1 Y
  
  
  As can be seen, the new version is created due to BIND_EQUIV_FAILURE

There is no longer  ROW_LEVEL_SEC_MISMATCH in 11.2.

Version_rpt script:

The script version_rpt can also be run to produce a summary report of the v$sql_shared_cursor view with additional diagnostic information. The script can be found in:

Running the Script:
Generate reports for all cursors with more than 100 versions using SQL_ID (10g and up):

Generate reports for all cursors with more than 100 versions using HASH_VALUE:

Generate the report for cursor with sql_id cyzznbykb509s:

What further tracing is available.

In 10G it is possible to use CURSORTRACE to aid the investigation of why cursors are not being shared. This event should only be used under the guidance of support and the resultant trace file is undocumented. To get the trace for a particular SQL statement you first of all need to get the hash_value (See the above select from v$sqlarea). You then set the trace on using:-  

(levels 578-580 can be used for high level tracing (577=level 1, 578=level 2, 580=level 3)

This will write a trace file to user_dump_dest each time we try to reuse the cursor.
To turn off tracing use:-

Please note: exists in 10.2 (fixed in 10.2.0.4) where cursor trace cannot fully be turned off and single line entries will still be made to the trace file as a result. The w/a is to restart the instance. How invasive this BUG is depends on the executions of the cursor (and the size of the resultant trace file additions)

In 11.2 there is also cursordump:

 (please ensure system , not session, is used as the level meaning changes)

This dumps some additional information such as expanding on the parameters for 'optimizer_mismatch' issues.

In later versions of the RDBMS there are also enhancements which dump more information as to the actual reason a child cursor could not share (ie the parameter differences). This information can be found in the REASON column of v$sql_shared_cursor and is in XML format. See for example.

Are there any times when a high version count is expected even though BINDS are being used?

Consider the following where cursor_sharing=SIMILAR

You will see several versions , each with no obvious reason for not being shared

One of the cursor sharing criteria when literal replacement is enabled with cursor_sharing as similar is that bind value should match initial bind value if the execution plan is going to change depending on the value of the literal. The reason for this is we _might_ get a sub optimal plan if we use the same cursor. This would typically happen when depending on the value of the literal optimizer is going to chose a different plan. Thus in this test case we have a predicate with > , if this was a equality we would always share the same child cursor. If application developers are ready to live with a sub-optimal plan and save on memory , then they need to set the parameter to force.

"The difference between SIMILAR and FORCE is that SIMILAR forces similar statements to share the SQL area without deteriorating execution plans.
Setting CURSOR_SHARING to FORCE forces similar statements to share the SQL area potentially deteriorating execution plans."

It is also possible to tell from 10046 trace (level 4/12 - BINDS) if a bind is considered to be unsafe
The flag oacfl2 in 9i and fl2 in 10g will show if a variable is unsafe.

In 10g (10.2.0.5) and 11g using the example query above this looks like:

The "fl2=0300" entry indicates that this is and Unsafe literal and the bind was generated by replacement :

The 0x200 entry being the important flag for determination of literal 'safety'.
For additional details on this topic see:

Enhancement to obsolete parent cursors if Version Count exceeds a threshold

In 11gr2, an issue of Child cursors growing very long was introduced.  An enhancement request was filed to address this issue Bug 10187168. When the child cursors grow beyond certain count be it 20 or 100, it obsoletes the parent cursors.  In order to activate this enhancement bug set following:

1. If 11.2.0.3 and above, set the following parameters:

2. If  11.2.0.2.2, then set: 

To show obsoleted cursor, run following query:

For more information, please read the following article regarding the enhancement involved:

High Version Count with Adaptive Cursor Sharing

With introduction of adaptive cursor sharing in 11g, there may be increased version count due to more child cursors.  The adaptive cursor sharing is meant to adapt execution plans, depending on the selectivity of the bind variable.  For more information on adaptive cursor sharing, please review following note:

Some known issues with Adaptive Cursor Sharing Overview:

Known Issues

A list of known issues can be found in:

Troubleshooting Other Issues

For guidance troubleshooting other performance issues take a look at: