Misconception of Archive Log Sequences in Data Guard

I have read in OTN forums and personally went though such misconception of archive log sequences in primary, We will go in detail to understand why sequence number is showing higher value(Ex: 50) in “v$archived_log” even though the current log sequence(SQL> archive log list) is very low  Ex: 5-6.

This behavior because of registering the old incarnation sequences with the recent control file, If you are using FRA or try to register the archive log location when there are old incarnation sequences exist in the same location. The old incarnation sequences can persist if we haven’t cleaned up after several Drills(Fail-over/Switchover). I did several drills before writing this small article how the log sequences will be updated with control file. By thus you cannot estimate or use views to check the synchronization between primary and standby database(s). So consider to cleanup them by several procedures. We will discuss available scenarios.

Now lets check what is the maximum sequence archived on primary with the first query  and second query output shows what is the log sequence applied on standby with the GAP.

PRIMARY
SQL>  select max(sequence#) from v$archived_log;
MAX(SEQUENCE#)
--------------
            56
SQL>
DB_NAME    HOSTNAME       LOG_ARCHIVED LOG_APPLIED APPLIED_TIME   LOG_GAP
 ---------- -------------- ------------ ----------- -------------- -------
ORCL       ORACLE-PRIMARY           56          48 03-FEB/19:17         8

So the maximum archived sequence on primary is “56” and log applied on standby is “48” from primary and the GAP stated as “8”. Initially we start troubleshoot to fix the archive gaps of “8” and after all the troubleshooting part found that there is no archive with “56” on primary and the current log sequences is just “7” from the below output.

SQL> archive log list
Database log mode              Archive Mode
Automatic archival             Enabled
Archive destination            USE_DB_RECOVERY_FILE_DEST
Oldest online log sequence     5
Next log sequence to archive   7
Current log sequence           7
SQL>

Here the problem is only from the primary database and you can see from below output what is the maximum sequence applied on the standby database.

STANDBY
SQL>  select max(sequence#) from v$archived_log where applied='YES';
MAX(SEQUENCE#)
--------------
             6
SQL>

So what’s wrong with the view “v$archived_log” ? Of course if you check the “RESETLOGS_CHANGE#” from the “v$archived_log” then you can see multiple “RESETLOGS_CHANGE#” this is because of previous fail-overs the old incarnation sequence numbers are still registered too. So how to cleanup those archives? You have several ways

1) Re-create Control file

— This is not always possible because you need downtime of primary/production database.

2) Uncatalog archive log files.

— By this procedure you can uncatalog all of the archives and after removing old incarnation archives again we can register them as below steps.

RMAN> change archivelog sequence 1 uncatalog;
RMAN> change archivelog all uncatalog;
RMAN> catalog start with ‘archive log location’;

3) Clear the section in the controlfile which contains data referencing v$archived_log, May be in view “v$archived_log” it can contain information of destinations 1 to 30, This below procedure describes process of only keeping entries from one distinct location. The package “dbms_backup_restore.resetCfileSection” refers to different sections such as cleanup of “v$archived_log” and cleanup of backups so on.

Now gather information to verify the number of sequences after executing the package.

SQL> select count(*) from v$archived_log;
COUNT(*)
----------
237 
SQL>

Run the below DBMS package from “SYS” user.

SQL> execute sys.dbms_backup_restore.resetCfileSection( 11);
PL/SQL procedure successfully completed.
SQL>

Check again how many entries exist in v$archived_log?

SQL> select count(*) from v$archived_log;
COUNT(*)
----------
0 
SQL> select sequence#,applied from v$archived_log;
no rows selected
SQL>

Register existing archive log files with Control file as below. Before this step ensure there are no archive log sequences of old incarnation. 🙂
RMAN> catalog start with ‘Archive Log Location’;

Note:- Prior to this entire operation, Recommended to take Full backup.

After registering the archive sequences of current “RESETLOGS_CHANGE#” then you can check the sequence numbers of primary and standby database(s) and also synchronization between primary and standby databases.

— Happy Reading–

 

 

 

 

Website Comments

  1. Raj
    Reply

    Hi Basha,

    What is number 11 here?

    execute sys.dbms_backup_restore.resetCfileSection( 11);

    Raj

    • nassyambasha@gmail.com
      Reply

      Hi Raj,
      The entry 11 refers to v$archived_log entries, so that it clears the history and again you have to catalog new archive log files after clearing unnecessary old incarnation files. This number vary from each view, so for v$archived_Log it is 11..
      Well as i read from many sources, each number refers to each view, like wise 19 refers to deleted objects so on.
      I hope you got it.

      Thank you

  2. Ahamed Sulthan
    Reply

    Hi Basha –

    we faced the same issues on production version 11.2.finally fixed issues

    This is very nice articles

    Thanks

Post a comment

*