Online Redo Logs and TEMP File Locations with Graceful Switchover

Follow
Problem Description

Standby version 7.0 with 2 redo logs directories /redoA/PRD and /redoB/PRD on primary and standby server (same structure).

Graceful Switchover is successful, but several problems occurred after the switchover, in that online redo logs and tempfiles were created on the new primary system, but in new locations:

  1. The redo logs were created in a new structure under /oradata/PRD/onlinelogs (but /redoA/PRD and /redoB/PRD are available).

  2. Under /oradata/PRD/datafile/ Dbvisit Standby created an OMF Temp Tablespace. But the original was in /oradata/PRD.

Why is this, as we have no OMF definitions and the setup was clean? Is there a parameter I can check?

Solution

When we look at managing redo and temp locations for the standby database there are a large number of factors to take into account. Things such as:

  • Is ASM used?
  • Is OMF enabled?
  • Are RAW devices used?
  • Is Filesystem used?
  • Is Primary ASM and Standby ASM?
  • Is Primary ASM and standby Filesystem?
  • Is Primary Filesystem but standby ASM?
  • Is primary and standby locations the same?

and a few more.

So managing this is a complex process as each site can be different and we attempt to cater for as many scenarios as possible.

By default we favour the use of OMF files when the creation of OMF files are enabled. By this we mean if parameters such as: db_create_file_dest and db_create_online_log_dest_n are used. We pick this up and during the CSD and Graceful Switchover (GS) processes, these values will be taken into account and OMF files will be created for TEMP and REDO (which by default is not used or required, on a standby database).

What happened in your case was that these parameters were set for the standby database, and on Graceful Switchover this was picked up and the redo/temp files were created as OMF files.

If you want to disable the use of OMF in this case, prior to GS make sure these parameters are not enabled (set to null), for example:

alter system reset db_create_file_dest scope=spfile;
alter system reset db_create_online_log_dest_1 scope=spfile;

Note: Using the reset options requires the standby to be restarted for the spfile changes to take affect.

Following the above changes, during Graceful Switchover, OMF will not be used for the redo/temp files. However, the redo and temp files will take the following naming convention:

DEFAULTSTDBY_REDO_NAME = 'redo%G%N.log';

Name pattern used when creating stdby redo once a stdby ctl file recreated, %N is counter starting from 1, %G is group number.

DEFAULT_STDBY_TEMP_NAME = '%T%N.dbf';

Name pattern used when creating stdby tempfiles once a stdby ctl file recreated, %N is counter starting from 1, %T is tablespace name.

These parameters can be updated in the DDC file if required.

Additional Information

For more detail on Redo and Temp file management during the CSD and GS process please review this section in the online user guide:

https://dbvisit.atlassian.net/wiki/display/UGDS7/Redo+and+Tempfile+Management+during+CSD+and+GS

Mike Donovan February 12, 2015 12:15

Have more questions? Submit a request

Comments