A Troubleshooting Data Guard

A.1.1 Renaming Datafiles with the ALTER DATABASE Statement

You cannot rename the datafile on the standby site when the STANDBY_FILE_MANAGEMENT initialization parameter is set to AUTO. When you set the STANDBY_FILE_MANAGEMENT initialization parameter to AUTO, use of the following SQL statements is not allowed:

• ALTER DATABASE RENAME

• ALTER DATABASE ADD/DROP LOGFILE

• ALTER DATABASE ADD/DROP STANDBY LOGFILE MEMBER

• ALTER DATABASE CREATE DATAFILE AS

If you attempt to use any of these statements on the standby database, an error is returned. For example:

SQL> ALTER DATABASE RENAME FILE '/disk1/oracle/oradata/payroll/t_db2.log' to 'dummy';

alter database rename file '/disk1/oracle/oradata/payroll/t_db2.log' to 'dummy' 
* 
ERROR at line 1: 
ORA-01511: error in renaming log/datafiles 
ORA-01270: RENAME operation is not allowed if STANDBY_FILE_MANAGEMENT is auto

See Section 9.3.1 to learn how to add datafiles to a physical standby database.

A.1.2 Standby Database Does Not Receive Redo Data from the Primary Database

If the standby site is not receiving redo data, query the V$ARCHIVE_DEST view and check for error messages. For example, enter the following query:

SQL> SELECT DEST_ID "ID", -
> STATUS "DB_status", -
> DESTINATION "Archive_dest", -
> ERROR "Error" -
> FROM V$ARCHIVE_DEST WHERE DEST_ID <=5;

ID DB_status Archive_dest                   Error   
-- --------- ------------------------------ ------------------------------------
 1  VALID    /vobs/oracle/work/arc_dest/arc                          
 2  ERROR    standby1                       ORA-16012: Archivelog standby database identifier mismatch  
 3  INACTIVE                            
 4  INACTIVE                    
 5  INACTIVE                                           
5 rows selected.

If the output of the query does not help you, check the following list of possible issues. If any of the following conditions exist, redo transport services will fail to transmit redo data to the standby database:

• The service name for the standby instance is not configured correctly in the tnsnames.ora file for the primary database.

• The Oracle Net service name specified by the LOG_ARCHIVE_DEST_n parameter for the primary database is incorrect.

• The LOG_ARCHIVE_DEST_STATE_n parameter for the standby database is not set to the value ENABLE.

• The listener.ora file has not been configured correctly for the standby database.

• The listener is not started at the standby site.

• The standby instance is not started.

• You have added a standby archiving destination to the primary SPFILE or text initialization parameter file, but have not yet enabled the change.

• Redo transport authentication has not been configured properly. See section 3.1.2 for redo transport authentication configuration requirements.

• You used an invalid backup as the basis for the standby database (for example, you used a backup from the wrong database, or did not create the standby control file using the correct method).

A.1.3 You Cannot Mount the Physical Standby Database

You cannot mount the standby database if the standby control file was not created with the ALTER DATABASE CREATE [LOGICAL] STANDBY CONTROLFILE ... statement or RMAN command. You cannot use the following types of control file backups:

• An operating system-created backup

• A backup created using an ALTER DATABASE statement without the PHYSICAL STANDBY or LOGICAL STANDBY option

A.4.1 Switchover Fails Because Redo Data Was Not Transmitted

If the switchover does not complete successfully, you can query the SEQUENCE# column in the V$ARCHIVED_LOG view to see if the last redo data transmitted from the original primary database was applied on the standby database. If the last redo data was not transmitted to the standby database, you can manually copy the archived redo log file containing the redo data from the original primary database to the old standby database and register it with the SQL ALTER DATABASE REGISTER LOGFILE file_specification statement. If you then start apply services, the archived redo log file will be applied automatically. Query the SWITCHOVER_STATUS column in the V$DATABASE view. A switchover to the primary role is now possible if the SWITCHOVER_STATUS column returns TO PRIMARY or SESSIONS ACTIVE.

SQL> SELECT SWITCHOVER_STATUS FROM V$DATABASE;

SWITCHOVER_STATUS 
----------------- 
TO PRIMARY 
1 row selected 

See Chapter 17 for information about other valid values for the SWITCHOVER_STATUS column of the V$DATABASE view.

To continue with the switchover, follow the instructions in Section 8.2.1 for physical standby databases or Section 8.3.1 for logical standby databases, and try again to switch the target standby database to the primary role.

A.4.2 Switchover Fails Because SQL Sessions Are Still Active

If you do not include the WITH SESSION SHUTDOWN clause as a part of the ALTER DATABASE COMMIT TO SWITCHOVER TO PHYSICAL STANDBY statement, active SQL sessions might prevent a switchover from being processed. Active SQL sessions can include other Oracle Database processes.

When sessions are active, an attempt to switch over fails with the following error message:

SQL> ALTER DATABASE COMMIT TO SWITCHOVER TO PHYSICAL STANDBY;
 
ALTER DATABASE COMMIT TO SWITCHOVER TO PHYSICAL STANDBY * 
ORA-01093: ALTER DATABASE CLOSE only permitted with no sessions connected

Action: Query the V$SESSION view to determine which processes are causing the error. For example:

SQL> SELECT SID, PROCESS, PROGRAM FROM V$SESSION -   
> WHERE TYPE = 'USER' -
>  AND SID <> (SELECT DISTINCT SID FROM V$MYSTAT);

SID        PROCESS   PROGRAM 
---------  --------  ------------------------------------------------ 
        7      3537  oracle@nhclone2 (CJQ0)
       10
       14
       16
       19
       21
 6 rows selected.

In the previous example, the JOB_QUEUE_PROCESSES parameter corresponds to the CJQ0 process entry. Because the job queue process is a user process, it is counted as a SQL session that prevents switchover from taking place. The entries with no process or program information are threads started by the job queue controller.

Verify the JOB_QUEUE_PROCESSES parameter is set using the following SQL statement:

SQL> SHOW PARAMETER JOB_QUEUE_PROCESSES; 

NAME                           TYPE      VALUE
------------------------------ -------   -------------------- 
job_queue_processes            integer   5

Then, set the parameter to 0. For example:

SQL> ALTER SYSTEM SET JOB_QUEUE_PROCESSES=0; 
Statement processed.

Because JOB_QUEUE_PROCESSES is a dynamic parameter, you can change the value and have the change take effect immediately without having to restart the instance. You can now retry the switchover procedure.

Do not modify the parameter in your initialization parameter file. After you shut down the instance and restart it after the switchover completes, the parameter will be reset to the original value. This applies to both primary and physical standby databases.

Table A-1 summarizes the common processes that prevent switchover and what corrective action you need to take.

A.4.3 Switchover Fails with the ORA-01102 Error

Suppose the standby database and the primary database reside on the same site. After both the ALTER DATABASE COMMIT TO SWITCHOVER TO PHYSICAL STANDBY and the ALTER DATABASE COMMIT TO SWITCHOVER TO PRIMARY statements are successfully executed, shut down and restart the physical standby database and the primary database.

However, the startup of the second database fails with ORA-01102 error "cannot mount database in EXCLUSIVE mode."

This could happen during the switchover if you did not set the DB_UNIQUE_NAME parameter in the initialization parameter file that is used by the standby database (that is, the original primary database). If the DB_UNIQUE_NAME parameter of the standby database is not set, the standby and the primary databases both use the same mount lock and cause the ORA-01102 error during the startup of the second database.

Action: Add DB_UNIQUE_NAME=unique_database_name to the initialization parameter file used by the standby database, and shut down and restart the standby and primary databases.

A.4.4 Redo Data Is Not Applied After Switchover

The archived redo log files are not applied to the new standby database after the switchover.

This might happen because some environment or initialization parameters were not properly set after the switchover.

Action:

• Check the tnsnames.ora file at the new primary site and the listener.ora file at the new standby site. There should be entries for a listener at the standby site and a corresponding service name at the primary site.

• Start the listener at the standby site if it has not been started.

• Check if the LOG_ARCHIVE_DEST_n initialization parameter was set to properly transmit redo data from the primary site to the standby site. For example, query the V$ARCHIVE_DEST fixed view at the primary site as follows:

  SQL> SELECT DEST_ID, STATUS, DESTINATION FROM V$ARCHIVE_DEST;
  

  If you do not see an entry corresponding to the standby site, you need to set LOG_ARCHIVE_DEST_n and LOG_ARCHIVE_DEST_STATE_n initialization parameters.

• Set the STANDBY_ARCHIVE_DEST and LOG_ARCHIVE_FORMAT initialization parameters correctly at the standby site so that the archived redo log files are applied to the desired location. (Note that the STANDBY_ARCHIVE_DEST parameter has been deprecated and is supported for backward compatibility only.)

• At the standby site, set the DB_FILE_NAME_CONVERT and LOG_FILE_NAME_CONVERT initialization parameters. Set the STANDBY_FILE_MANAGEMENT initialization parameter to AUTO if you want the standby site to automatically add new datafiles that are created at the primary site.

A.4.5 Roll Back After Unsuccessful Switchover and Start Over

For physical standby databases in situations where an error occurred and it is not possible to continue with the switchover, it might still be possible to revert the new physical standby database back to the primary role by using the following steps. (This functionality is available starting with Oracle Database 11g Release 2 (11.2.0.2).)

1. Shut down and mount the new standby database (old primary).

2. Start Redo Apply on the new standby database.

3. Verify that the new standby database is ready to be switched back to the primary role. Query the SWITCHOVER_STATUS column of the V$DATABASE view on the new standby database. For example:

   SQL> SELECT SWITCHOVER_STATUS FROM V$DATABASE;
    
   SWITCHOVER_STATUS 
   ----------------- 
   TO_PRIMARY 
   1 row selected
   

   A value of TO PRIMARY or SESSIONS ACTIVE indicates that the new standby database is ready to be switched to the primary role. Continue to query this column until the value returned is either TO PRIMARY or SESSIONS ACTIVE.

4. Issue the following statement to convert the new standby database back to the primary role:

   SQL> ALTER DATABASE COMMIT TO SWITCHOVER TO PRIMARY WITH SESSION SHUTDOWN;
   

   If this statement is successful, the database will be running in the primary database role, and you do not need to perform any more steps.

   If this statement is unsuccessful, then continue with Step 5.

5. When the switchover to change the role from primary to physical standby was initiated, a trace file was written in the log directory. This trace file contains the SQL statements required to re-create the original primary control file. Locate the trace file and extract the SQL statements into a temporary file. Execute the temporary file from SQL*Plus. This will revert the new standby database back to the primary role.

6. Shut down the original physical standby database.

7. Create a new standby control file. This is necessary to resynchronize the primary database and physical standby database. Copy the physical standby control file to the original physical standby system. Section 3.2.2 describes how to create a physical standby control file.

8. Restart the original physical standby instance.

   If this procedure is successful and archive gap management is enabled, the FAL processes will start and re-archive any missing archived redo log files to the physical standby database. Force a log switch on the primary database and examine the alert logs on both the primary database and physical standby database to ensure the archived redo log file sequence numbers are correct.

   See Section 6.4.3.1 for information about archive gap management and Appendix F for information about locating the trace files.

9. Try the switchover again.

   At this point, the Data Guard configuration has been rolled back to its initial state, and you can try the switchover operation again (after correcting any problems that might have led to the initial unsuccessful switchover).

A.5.1 Failures During the Prepare Phase of a Switchover Operation

If a failure occurs during the preparation phase of a switchover operation, you should cancel the switchover and retry the switchover operation from the very beginning.

SQL> ALTER DATABASE PREPARE TO SWITCHOVER TO LOGICAL STANDBY CANCEL;

A.5.2 Failures During the Commit Phase of a Switchover Operation

Although committing to a switchover involves a single SQL statement, internally a number of operations are performed. The corrective actions that you need to take depend on the state of the commit to switchover operation when the error was encountered.

A.10.1 Recovering from Errors

Logical standby databases maintain user tables, sequences, and jobs. To maintain other objects, you must reissue the DDL statements seen in the redo data stream.

If SQL Apply fails, an error is recorded in the DBA_LOGSTDBY_EVENTS table. The following sections demonstrate how to recover from two such errors.

SQL> SET LONG 1000
SQL> ALTER SESSION SET NLS_DATE_FORMAT  = 'DD-MON-YY HH24:MI:SS';

Session altered.

SQL> SELECT EVENT_TIME, COMMIT_SCN, EVENT, STATUS FROM DBA_LOGSTDBY_EVENTS;

EVENT_TIME              COMMIT_SCN
------------------ ---------------
EVENT
-------------------------------------------------------------------------------
STATUS
-------------------------------------------------------------------------------
22-OCT-03 15:47:58

ORA-16111: log mining and apply setting up

22-OCT-03 15:48:04          209627
insert into "SCOTT"."EMP"
values
   "EMPNO" = 7900,
   "ENAME" = 'ADAMS',
   "JOB" = 'CLERK',
   "MGR" IS NULL,
   "HIREDATE" = TO_DATE('22-OCT-03', 'DD-MON-RR'),
   "SAL" = 950,
   "COMM" IS NULL,
   "DEPTNO" IS NULL
ORA-01653: unable to extend table SCOTT.EMP by %200 bytes in tablespace T_TABLE

SQL> ALTER TABLESPACE t_table ADD DATAFILE '/dbs/t_db.f' SIZE 60M;
Tablespace altered.

SQL> ALTER DATABASE START LOGICAL STANDBY APPLY IMMEDIATE;
Database altered.

A.10.2 Troubleshooting SQL*Loader Sessions

Oracle SQL*Loader provides a method of loading data from different sources into the Oracle Database. This section analyzes some of the features of the SQL*Loader utility as it pertains to SQL Apply.

Regardless of the method of data load chosen, the SQL*Loader control files contain an instruction on what to do to the current contents of the Oracle table into which the new data is to be loaded, via the keywords of APPEND and REPLACE. The following examples show how to use these keywords on a table named LOAD_STOK:

• When using the APPEND keyword, the new data to be loaded is appended to the contents of the LOAD_STOK table:

  LOAD DATA
  INTO TABLE LOAD_STOK APPEND
  

• When using the REPLACE keyword, the contents of the LOAD_STOK table are deleted prior to loading new data. Oracle SQL*Loader uses the DELETE statement to purge the contents of the table, in a single transaction:

  LOAD DATA
  INTO TABLE LOAD_STOK REPLACE
  

Rather than using the REPLACE keyword in the SQL*Loader script, Oracle recommends that prior to loading the data, issue the SQL*Plus TRUNCATE TABLE command against the table on the primary database. This will have the same effect of purging both the primary and standby databases copy of the table in a manner that is both fast and efficient because the TRUNCATE TABLE command is recorded in the online redo log files and is issued by SQL Apply on the logical standby database.

The SQL*Loader script may continue to contain the REPLACE keyword, but it will now attempt to DELETE zero rows from the object on the primary database. Because no rows were deleted from the primary database, there will be no redo recorded in the redo log files. Therefore, no DELETE statement will be issued against the logical standby database.

Issuing the REPLACE keyword without the SQL statement TRUNCATE TABLE provides the following potential problems for SQL Apply when the transaction needs to be applied to the logical standby database.

• If the table currently contains a significant number of rows, then these rows need to be deleted from the standby database. Because SQL Apply is not able to determine the original syntax of the statement, SQL Apply must issue a DELETE statement for each row purged from the primary database.

  For example, if the table on the primary database originally had 10,000 rows, then Oracle SQL*Loader will issue a single DELETE statement to purge the 10,000 rows. On the standby database, SQL Apply does not know that all rows are to be purged, and instead must issue 10,000 individual DELETE statements, with each statement purging a single row.

• If the table on the standby database does not contain an index that can be used by SQL Apply, then the DELETE statement will issue a Full Table Scan to purge the information.

  Continuing with the previous example, because SQL Apply has issued 10,000 individual DELETE statements, this could result in 10,000 Full Table Scans being issued against the standby database.

A.10.3 Troubleshooting Long-Running Transactions

One of the primary causes for long-running transactions in a SQL Apply environment is because of Full Table Scans. Additionally, long-running transactions could be the result of SQL statements being replicated to the standby database, such as when creating or rebuilding an index.

Identifying Long-Running Transactions

If SQL Apply is executing a single SQL statement for a long period of time, then a warning message similar to the following is reported in the alert log of the SQL Apply instance:

Mon Feb 17 14:40:15 2003
WARNING: the following transaction makes no progress
WARNING: in the last 30 seconds for the given message!
WARNING: xid =
0x0016.007.000017b6 cscn = 1550349, message# = 28, slavid = 1
knacrb: no offending session found (not ITL pressure)

Note the following about the warning message:

• This warning is similar to the warning message returned for interested transaction list (ITL) pressure, with the exception being the last line that begins with knacrb. The final line indicates:

  • A Full Table Scan may be occurring

  • This issue has nothing to do with interested transaction list (ITL) pressure

• This warning message is reported only if a single statement takes more than 30 seconds to execute.

It may not be possible to determine the SQL statement being executed by the long-running statement, but the following SQL statement may help in identifying the database objects on which SQL Apply is operating:

SQL> SELECT SAS.SERVER_ID -
>  , SS.OWNER -
>  , SS.OBJECT_NAME -
>  , SS.STATISTIC_NAME -
>  , SS.VALUE -
>  FROM V$SEGMENT_STATISTICS SS -
>  , V$LOCK L -
>  , V$STREAMS_APPLY_SERVER SAS -
>  WHERE SAS.SERVER_ID = &SLAVE_ID -
>  AND L.SID = SAS.SID -
>  AND L.TYPE = 'TM' -
>  AND SS.OBJ# = L.ID1;

Additionally, you can issue the following SQL statement to identify the SQL statement that has resulted in a large number of disk reads being issued per execution:

SQL> SELECT SUBSTR(SQL_TEXT,1,40) -
>  , DISK_READS -
>  , EXECUTIONS -
>  , DISK_READS/EXECUTIONS -
>  , HASH_VALUE -
>  , ADDRESS -
>  FROM V$SQLAREA -
>  WHERE DISK_READS/GREATEST(EXECUTIONS,1) > 1 -
>  AND ROWNUM < 10 -
>  ORDER BY DISK_READS/GREATEST(EXECUTIONS,1) DESC;

Oracle recommends that all tables have primary key constraints defined, which automatically means that the column is defined as NOT NULL. For any table where a primary-key constraint cannot be defined, an index should be defined on an appropriate column that is defined as NOT NULL. If a suitable column does not exist on the table, then the table should be reviewed and, if possible, skipped by SQL Apply. The following steps describe how to skip all DML statements issued against the FTS table on the SCOTT schema:

1. Stop SQL Apply:

   SQL> ALTER DATABASE STOP LOGICAL STANDBY APPLY;
   Database altered
   

2. Configure the skip procedure for the SCOTT.FTS table for all DML transactions:

   SQL> EXECUTE DBMS_LOGSTDBY.SKIP(stmt => 'DML' , -
   >  schema_name => 'SCOTT' , -
   >  object_name => 'FTS');
   PL/SQL procedure successfully completed
   

3. Start SQL Apply:

   SQL> ALTER DATABASE START LOGICAL STANDBY APPLY IMMEDIATE;
   Database altered
   

Troubleshooting ITL Pressure

Interested transaction list (ITL) pressure is reported in the alert log of the SQL Apply instance. Example A-3 shows an example of the warning messages.

Tue Apr 22 15:50:42 2003
WARNING: the following transaction makes no progress
WARNING: in the last 30 seconds for the given message!
WARNING: xid =
0x0006.005.000029fa cscn = 2152982, message# = 2, slavid = 17

Real-Time Analysis

The messages shown in Example A-3 indicate that the SQL Apply process (slavid) #17 has not made any progress in the last 30 seconds. To determine the SQL statement being issued by the Apply process, issue the following query:

SQL> SELECT SA.SQL_TEXT -
>  FROM V$SQLAREA SA -
  >  , V$SESSION S -
  >  , V$STREAMS_APPLY_SERVER SAS -
  >  WHERE SAS.SERVER_ID = &SLAVEID -
  >  AND S.SID = SAS.SID -
  >  AND SA.ADDRESS = S.SQL_ADDRESS

SQL_TEXT
------------------------------------------------------------
insert into "APP"."LOAD_TAB_1" p("PK","TEXT")values(:1,:2)

An alternative method to identifying ITL pressure is to query the V$LOCK view, as shown in the following example. Any session that has a request value of 4 on a TX lock, is waiting for an ITL to become available.

SQL> SELECT SID,TYPE,ID1,ID2,LMODE,REQUEST -
> FROM V$LOCK -
> WHERE TYPE = 'TX'

SID        TY ID1        ID2        LMODE      REQUEST
---------- -- ---------- ---------- ---------- ----------
         8 TX     327688         48          6          0
        10 TX     327688         48          0          4

In this example, SID 10 is waiting for the TX lock held by SID 8.

Post-Incident Review

Pressure for a segment's ITL is unlikely to last for an extended period of time. In addition, ITL pressure that lasts for less than 30 seconds will not be reported in the standby databases alert log. Therefore, to determine which objects have been subjected to ITL pressure, issue the following statement:

SQL> SELECT SEGMENT_OWNER, SEGMENT_NAME, SEGMENT_TYPE -
>  FROM V$SEGMENT_STATISTICS -
>  WHERE STATISTIC_NAME = 'ITL WAITS' -
>  AND VALUE > 0 -
>  ORDER BY VALUE

This statement reports all database segments that have had ITL pressure at some time since the instance was last started.

Resolving ITL Pressure

To increase the INITRANS integer for a particular database object, it is necessary to first stop SQL Apply.

The following example shows the necessary steps to increase the INITRANS for table load_tab_1 in the schema app.

1. Stop SQL Apply:

   SQL> ALTER DATABASE STOP LOGICAL STANDBY APPLY;
   Database altered.
   

2. Temporarily bypass the database guard:

   SQL> ALTER SESSION DISABLE GUARD;
   Session altered.
   

3. Increase the INITRANS on the standby database. For example:

   SQL> ALTER TABLE APP.LOAD_TAB_1 INITRANS 30;
   Table altered
   

4. Reenable the database guard:

   SQL> ALTER SESSION ENABLE GUARD;
   Session altered
   

5. Start SQL Apply:

   SQL> ALTER DATABASE START LOGICAL STANDBY APPLY IMMEDIATE;
   Database altered.
   

Also, consider modifying the database object on the primary database, so in the event of a switchover, the error should not occur on the new standby database.

A.10.4 Troubleshooting ORA-1403 Errors with Flashback Transactions

If SQL Apply returns the ORA-1403: No Data Found error, then it may be possible to use Flashback Transaction to reconstruct the missing data. This is reliant upon the UNDO_RETENTION initialization parameter specified on the standby database instance.

Under normal circumstances, the ORA-1403 error should not be seen in a logical standby database environment. The error occurs when data in a table that is being managed by SQL Apply is modified directly on the standby database and then the same data is modified on the primary database. When the modified data is updated on the primary database and is subsequently received on the logical standby database, SQL Apply verifies the original version of the data is present on the standby database before updating the record. When this verification fails, the ORA-1403: No Data Found error is returned.

The Initial Error

When SQL Apply verification fails, the error message is reported in the alert log of the logical standby database and a record is inserted in the DBA_LOGSTDBY_EVENTS view.The information in the alert log is truncated, while the error is reported in it's entirety in the database view. For example:

LOGSTDBY stmt: UPDATE "SCOTT"."MASTER"
  SET
    "NAME" = 'john'
  WHERE 
    "PK" = 1 and 
    "NAME" = 'andrew' and 
    ROWID = 'AAAAAAAAEAAAAAPAAA'
LOGSTDBY status: ORA-01403: no data found
LOGSTDBY PID 1006, oracle@staco03 (P004)
LOGSTDBY XID 0x0006.00e.00000417, Thread 1, RBA 0x02dd.00002221.10

The Investigation

The first step is to analyze the historical data of the table that caused the error. This can be achieved using the VERSIONS clause of the SELECT statement. For example, you can issue the following query on the primary database:

SELECT VERSIONS_XID
      , VERSIONS_STARTSCN
      , VERSIONS_ENDSCN
      , VERSIONS_OPERATION
      , PK
      , NAME
   FROM SCOTT.MASTER
        VERSIONS BETWEEN SCN MINVALUE AND MAXVALUE
  WHERE PK = 1
  ORDER BY NVL(VERSIONS_STARTSCN,0);

VERSIONS_XID     VERSIONS_STARTSCN VERSIONS_ENDSCN V  PK NAME
---------------- ----------------- --------------- - --- -------
03001900EE070000           3492279         3492290 I   1 andrew
02000D00E4070000           3492290                 D   1 andrew

Depending upon the amount of undo retention that the database is configured to retain (UNDO_RETENTION) and the activity on the table, the information returned might be extensive and you may need to change the versions between syntax to restrict the amount of information returned.From the information returned, it can be seen that the record was first inserted at SCN 3492279 and then was deleted at SCN 3492290 as part of transaction ID 02000D00E4070000.Using the transaction ID, the database should be queried to find the scope of the transaction. This is achieved by querying the FLASHBACK_TRANSACTION_QUERY view.

SELECT OPERATION
     , UNDO_SQL
  FROM FLASHBACK_TRANSACTION_QUERY
 WHERE XID = HEXTORAW('02000D00E4070000');

OPERATION  UNDO_SQL
---------- ------------------------------------------------
DELETE     insert into "SCOTT"."MASTER"("PK","NAME") values
           ('1','andrew');
BEGIN

Note that there is always one row returned representing the start of the transaction. In this transaction, only one row was deleted in the master table. The UNDO_SQL column when executed will restore the original data into the table.

SQL> INSERT INTO "SCOTT"."MASTER"("PK","NAME") VALUES ('1','ANDREW');SQL> COMMIT;

When you restart SQL Apply, the transaction will be applied to the standby database:

SQL> ALTER DATABASE START LOGICAL STANDBY APPLY IMMEDIATE;