Search

Top 60 Oracle Blogs

Recent comments

Using dbca to create a physical standby database

While investigating new options I discovered with dbca for the previous article I noticed that it’s now possible to use Oracle’s Database Creation Assistant (dbca) to create a physical standby database using the -createDuplicateDB flag.

I wanted to know how easily this can be done on my Oracle 18.5.0 system. I have 2 VMs, server3 and server4 running Oracle Linux 7.6. I created the primary database on server3. The database name is NCDB with the db_unique_name set to SITEA. I also use SITEA as the database SID to prevent confusion. My physical standby database with db_unique_name SITEB will reside on server4.

The setup is kept as simple as possible to avoid any unintended complications. In other words, this is a lab setup, not something you’d entertain using for any other purpose. As always, this is a tech post and not a piece about licencing. It is up to you to ensure you aren’t performing any tasks that you aren’t licensed for.

Creating the primary database

The primary database can be created using dbca as always. Unlike the last post, this one uses ASM, not a filesystem to show how easy it is to do both.

The primary database is a non-Container Database (NCDB). I intend to create the primary instance as SITEA (Site A) with the standby named SITEB. This way no one gets confused when the database runs in primary role in the designated disaster recovery location. I haven’t seen gdbName <> sid/db_unique_name in many single instance deployments and appreciate that might be unfamiliar but bear with me. I’ll try to make this as transparent as possible.

Invoking dbca

Enough talking, let’s create the database:

[oracle@server3 ~]$ dbca -silent -createDatabase \
> -gdbName NCDB \
> -templateName lab_template.dbc \
> -sid SITEA \
> -enableArchive true -archiveLogMode auto \
> -memoryMgmtType auto_sga \
> -createAsContainerDatabase false \
> -recoveryAreaDestination +RECO \
> -totalMemory 2048 \
> -datafileDestination +DATA \
> -initParams db_unique_name=SITEA,dg_broker_start=true \
> -customScripts add_srls_dbca.sql

[...]

Prepare for db operation
10% complete
Registering database with Oracle Restart
14% complete
Copying database files
43% complete
Creating and starting Oracle instance
45% complete
49% complete
53% complete
56% complete
62% complete
Completing Database Creation
68% complete
70% complete
71% complete
Executing Post Configuration Actions
100% complete
Database creation complete. For details check the logfiles at:
/u01/app/oracle/cfgtoollogs/dbca/SITEA.
Database Information:
Global Database Name:SITEA
System Identifier(SID):SITEA
Look at the log file "/u01/app/oracle/cfgtoollogs/dbca/SITEA/SITEA1.log" for further details.

A few words of explanation: As I said I use SITEA as db_unique_name and the database’s SID. In Oracle Managed Files (OMF) deployments a file name references the db_unique name, and I wanted my data files, temp files and log files all to be in +DATA/SITEA/. db_unique_name defaults to db_name, and not changing it would have resulted in the database being created in +DATA/NCDB and that’s not what I wanted.

I decided to include starting the Data Guard broker as part of the initialisation parameters. Archiving is enabled right from the start as well.

The custom script you see referenced creates the necessary number of standby redo logs. This will save me a bit of work, because RMAN is smart. When duplicating the database for use as a standby it creates standby redo logs (SRL) for me if if finds SRLs in the target database. Yet another thing less for me to worry about.

Implications of using gdbname and sid (+ db_unique_name) in dbca

If you haven’t used a SID <> gDBName combination before, you might feel about unsure about how Oracle creates things. First of all, it created the database with the db_name you passed to dbca as gdbname. In this post that’s NCDB. Normally, if you grep’d the operating system’s process table for NCDB you would find all the instance’s background processes. With my dbca command however (where an Oracle SID <> gdbname is provided) you won’t find anything though. Grep for the ORACLE_SID “SITEA” instead. Likewise, if you want to create a bequeath connection to the database, set your ORACLE_SID to SITEA.

Since this is Oracle Restart the database is registered with Grid Infrastructure under its db_unique_name/ORACLE_SID combination in the RDBMS home:

[oracle@server3 ~]$ srvctl config database
SITEA
[oracle@server3 ~]$ srvctl config database -db SITEA
Database unique name: SITEA
Database name: NCDB
Oracle home: /u01/app/oracle/product/18.0.0/dbhome_1
Oracle user: oracle
Spfile: +DATA/SITEA/PARAMETERFILE/spfile.273.1003226559
Password file:
Domain:
Start options: open
Stop options: immediate
Database role: PRIMARY
Management policy: AUTOMATIC
Disk Groups: DATA,RECO
Services:
OSDBA group:
OSOPER group:
Database instance: SITEA
[oracle@server3 ~]$

Interacting with the database via srvctl requires you to use the db_unique_name SITEA as shown.

This is my preferred way to set things up these days. I’d like to invite you to critically reflect about the approach I took. I prefer it over other designations such as “PROD” and “STDBY”.

Creating the standby database

With the primary database created, archivelog mode enabled plus a working Fast Recovery Area I can move on to creating the physical standby database.

Before I do this a few small changes have to be made to my network configuration. I always add TNS entries for both primary and standby database in the “primary” host’s $ORACLE_HOME/network/admin/tnsnames.ora file and ship this across to the “standby” host. They are needed later. With the preparation done, I proceed to calling dbca:

[oracle@server4 ~]$ dbca -silent -createDuplicateDB \
> -gdbName NCDB \
> -sid SITEB \
> -initParams db_create_file_dest=+DATA,db_recovery_file_dest=+RECO,dg_broker_start=true \
> -primaryDBConnectionString server3:1521/SITEA \
> -createAsStandby \
> -dbUniqueName SITEB

[ ... ]

Prepare for db operation
22% complete
Listener config step
44% complete
Auxiliary instance creation
67% complete
RMAN duplicate
89% complete
Post duplicate database operations
100% complete
Look at the log file "/u01/app/oracle/cfgtoollogs/dbca/SITEB/SITEB1.log" for further details.

So far so good. What did dbca do? And more importantly, how did it create the standby database? I didn’t have a backup of NCDB/SITEA at the time I called dbca on server4 so it must have performed a duplicate … from active database. Looking at the dbca trace in $ORACLE_BASE/cfgtoollogs/dbca/SITEB I can confirm this. After wading through a lot of messages, here’s the relevant bit:

INFO: Mar 11, 2019 12:14:04 PM oracle.assistants.dbca.driver.backend.steps.RmanDuplicateStep executeImpl
INFO: Running rman duplicate command:run {
set newname for datafile 1 to new;
set newname for datafile 3 to new;
set newname for datafile 4 to new;
set newname for datafile 7 to new;
duplicate target database
for standby
from active database
dorecover
nofilenamecheck;
}

This might be an issue for some users because it requires a direct connection to the source database. I don’t know if RMAN might use the “from service … using backupset …” clause if you provided an offline backup.

The files I care about are all in the right place:

SQL> select name from v$datafile union all
2 select name from v$tempfile union all
3 select member from v$logfile union all
4 select value from v$parameter2 where name = 'control_files';

NAME
-------------------------------------------------------
+DATA/SITEB/DATAFILE/system.268.1003228125
+DATA/SITEB/DATAFILE/sysaux.273.1003228127
+DATA/SITEB/DATAFILE/undotbs1.272.1003228129
+DATA/SITEB/DATAFILE/users.271.1003228129
+DATA/SITEB/TEMPFILE/temp.262.1003228149
+DATA/SITEB/ONLINELOG/group_3.266.1003228137
+RECO/SITEB/ONLINELOG/group_3.310.1003228137
+DATA/SITEB/ONLINELOG/group_2.267.1003228137
+RECO/SITEB/ONLINELOG/group_2.258.1003228137
+DATA/SITEB/ONLINELOG/group_1.270.1003228135
+RECO/SITEB/ONLINELOG/group_1.259.1003228135
+DATA/SITEB/ONLINELOG/group_4.265.1003228137
+RECO/SITEB/ONLINELOG/group_4.311.1003228137
+DATA/SITEB/ONLINELOG/group_5.264.1003228139
+RECO/SITEB/ONLINELOG/group_5.312.1003228139
+DATA/SITEB/ONLINELOG/group_6.274.1003228139
+RECO/SITEB/ONLINELOG/group_6.313.1003228141
+DATA/SITEB/ONLINELOG/group_7.263.1003228147
+RECO/SITEB/ONLINELOG/group_7.314.1003228147
+DATA/SITEB/CONTROLFILE/current.260.1003228117
+RECO/SITEB/CONTROLFILE/current.260.1003228117

21 rows selected.

As reported by other bloggers, dbca creates a “temporary” listener that it doesn’t remove after the database creation process is completed. I specifically wanted to see if this is the case with Oracle Restart as well and yes, it is:

[oracle@server4 SITEB]$ ps -ef | grep tnslsnr
oracle 1125 1 0 10:27 ? 00:00:00 /u01/app/oracle/product/18.0.0/dbhome_1/bin/tnslsnr LISTENER20190318102721 -inherit
grid 5837 1 0 09:01 ? 00:00:00 /u01/app/grid/product/18.0.0/grid/bin/tnslsnr LISTENER -no_crs_notify -inherit
oracle 7588 31809 0 10:44 pts/1 00:00:00 grep --color=auto tnslsnr

The extra listener didn’t survive a reboot though and the processes should disappear eventually. dbca persists the configuration into $ORACLE_HOME/network/admin/listener.ora though, and those entries should probably be removed. This is especially true with Oracle Restart (and RAC for that matter) because I want all listeners to be controlled from the GRID home, not the RDBMS home.

There are other leftover files in $ORACLE_HOME/dbs as well for which I’d hope dbca removes them one day.

On the plus side dbca registers the database in Oracle Restart for me as part of the work it did:

[oracle@server4 ~]$ srvctl config database -db SITEB
Database unique name: SITEB
Database name: NCDB
Oracle home: /u01/app/oracle/product/18.0.0/dbhome_1
Oracle user: oracle
Spfile: +DATA/SITEB/PARAMETERFILE/spfile.275.1003228201
Password file: +DATA/SITEB/orapwsiteb
Domain:
Start options: open
Stop options: immediate
Database role: PRIMARY
Management policy: AUTOMATIC
Disk Groups: DATA,RECO
Services:
OSDBA group:
OSOPER group:
Database instance: SITEB

It’s good to see that I don’t need to worry about storing password file and spfile in ASM myself, so that job is already taken care off. A few other things need to be changed though. The database is still considered a primary (of which I know it isn’t) and the start option therefore is wrong, too. There are licensing implications opening a standby read only rather than just mounting it when turning on redo apply. Refer to the documentation for more details.

Out of habit I changed the role to PHYSICAL_STANDBY and the start option to MOUNT. This is what I ended up with in my lab, YMMV and depends on how you are licensed.

[oracle@server4 ~]$ srvctl config database -db SITEB
Database unique name: SITEB
Database name: NCDB
Oracle home: /u01/app/oracle/product/18.0.0/dbhome_1
Oracle user: oracle
Spfile: +DATA/SITEB/PARAMETERFILE/spfile.275.1003228201
Password file: +DATA/SITEB/orapwsiteb
Domain:
Start options: mount
Stop options: immediate
Database role: PHYSICAL_STANDBY
Management policy: AUTOMATIC
Disk Groups: DATA,RECO
Services:
OSDBA group:
OSOPER group:
Database instance: SITEB

Creating the Broker configuration

The standby redo logs have been created by RMAN and with dg_broker_start set to true by dbca for both databases I can now create a Data Guard Broker configuration. With Oracle Restart I don’t have to statically register the databases with the listener anymore. Clusterware takes care of starting/stopping them during role changes.

[oracle@server3 ~]$ dgmgrl
DGMGRL for Linux: Release 18.0.0.0.0 - Production on Mon Mar 18 11:11:55 2019 Version 18.5.0.0.0

Copyright (c) 1982, 2018, Oracle and/or its affiliates. All rights reserved.

Welcome to DGMGRL, type "help" for information.
DGMGRL> connect sys@SITEA
Password:
Connected to "SITEA"
Connected as SYSDBA.
DGMGRL> create configuration autoconf
> primary database is 'SITEA'
> connect identifier is 'SITEA';
Configuration "autoconf" created with primary database "SITEA"
DGMGRL> add database 'SITEB' as connect identifier is 'SITEB';
Database "SITEB" added
DGMGRL> edit database 'SITEA' set property standbyfilemanagement=auto;
Property "standbyfilemanagement" updated
DGMGRL> edit database 'SITEB' set property standbyfilemanagement=auto;
Property "standbyfilemanagement" updated
DGMGRL> enable configuration
Enabled.
DGMGRL> show configuration

Configuration - autoconf

Protection Mode: MaxPerformance
Members:
SITEA - Primary database
SITEB - Physical standby database

Fast-Start Failover: DISABLED

Configuration Status:
SUCCESS (status updated 17 seconds ago)
DGMGRL>

You might want to increase the protection level from MaxPerformance to something that better suits your requirements. Three might be additional changes you want to do within the broker as well. The code output you just saw is the bare minimum I put into place.

Summary

All in all this has been pretty straight forward without too many hiccups along the way. If you want to you can put a lot of those steps into your favourite automation toolkit and you’ll end up with a standby database reasonably quickly provided your environment supports this kind of setup.