DUPLICATE from ACTIVE Database Using RMAN, a Step-by-Step Guide

Very useful description – thanks!

Beside the “Two bold disadvantages of ACTIVE database” you’ve mentioned I see an other points: the whole directory structure has to be created before.

Regards, Heiko

This is an awesome post Kamran. In my case, I found that a few extra steps were needed however because I got this error after RMAN dup completed (only HRPRD1 will start (in my experience). Here is the error message:

[[email protected] ~]$ srvctl start database -d RATIONAL
PRCR-1079 : Failed to start resource ora.rational.db
CRS-5017: The resource action “ora.rational.db start” encountered the following error:
ORA-01618: redo thread 2 is not enabled – cannot mount
. For details refer to “(:CLSN00107:)” in “/u01/app/11.2.0/grid/log/cgi-ma-uat-db2v/agent/crsd/oraagent_oracle/oraagent_oracle.log”.

CRS-2674: Start of ‘ora.rational.db’ on ‘cgi-ma-uat-db2v’ failed
CRS-2632: There are no more servers to try to place resource ‘ora.rational.db’ on that would satisfy its placement policy

Following is needed to fix and convert single-instance cloned DB to RAC 2-node (more steps would be needed for 3-node etc) but similar to below.

Login to the HRPRD1 up and running instance.

alter database add logfile thread 2 group 4;
alter database add logfile thread 2 group 5;
alter database add logfile thread 2 group 6;

Then also:

create undo tablespace UNDOTBS2 datafile size 99614720;

(I used size identical to UNDOTBS1).

Now 2nd instance startup fine:

srvctl start instance -d HRPRD -i HRPRD2

Thanks again for this awesome post Kamran!!!

Forgot one command in my post above. After the add of the logfiles on thread 2, you have to run this command also:

alter database enable public thread 2;

Thanks

Hi

The script catclust.sql is not needed? I mean, if the source is single instance, all GV$ views are missing, am i right?

Regards

As spelled above about the bold disadvantages for duplicating database from target (Implies HRDEV in current case) , can you please detail on the tools that can be used to “gauge” the high network traffic connection prob and the impending performance impact on the source before ahead.

Hi Kamran Bakhshandeh,

Another worthy point that I’m persuaded to emphasize is regards the “Directory Structure” and the “Host” wherein which you are been planning to DUPLICATE the database.

CASE -1
=======
Duplicating a Database on a “Remote Host” with the ‘Same Directory Structure’ as source (HRDEV), then NO changes is required to be made for PFILE.

DUPLICATE DATABASE TO HRPRD
FROM ACTIVE DATABASE
PFILE (if starting the auxiliary instance with a client-side parameter file)
NOFILENAMECHECK; (if duplicated datafiles location= Source data file’s location and database’s are in DIFFERENT Host’s)

CASE-2
======
Duplicating a Database on a Remote Host with a Different Directory Structure:

Here you MUST change the PFILE as the duplicate database filenames structure is DIFFERENT from the Source filenames.

DUPLICATE DATABASE TO HRPRD
FROM ACTIVE DATABASE
DB_FILE_NAME_CONVERT=(/oracle/oradata/HRDEV/,/dup/oracle/oradata/HRPRD/)
LOGFILE (names and sizes for the duplicate database redo logs)
‘/dup/oracle/oradata/HRPRD/redo01.log’ SIZE 200K,
‘/dup/oracle/oradata/HRPRD/redo02.log’ SIZE 200K;

Surmise : If we DON’T specify the LOGFILE option, then RMAN after having done with Incomplete Recovery, opens up the database with RESETLOGS option to create the online logs.

CASE 3
======
Creating a Duplicate Database on the Local Host wherein the SOURCE (HRDEV) is resident.

Same procedure need to be followed as spelled in CASE 2.

NOTE : Do not use the NOFILENAMECHECK option in here as doing so will lead to overwriting the SOURCE Datafiles.

SO the above enunciation by Kamran Bakhshandeh will falls in CASE 2.

Hi,

there is a typo for the listener config in step 2 – extra bracket:

This ->

SID_LIST_LISTENER =
(SID_LIST =
(SID_DESC =
(SID_NAME = HRPRD1)
(ORACLE_HOME = /apps/oracle/product/11.2.0/db_1)
(GLOBAL_DBNAME = HRPRD)
)
)

Should be ->

SID_LIST_LISTENER =
(SID_LIST =
)
(SID_DESC =
(SID_NAME = HRPRD1)
(ORACLE_HOME = /apps/oracle/product/11.2.0/db_1)
(GLOBAL_DBNAME = HRPRD)
)
)

this command is wrong for creating an undo ts:
create tablespace UNDOTBS2 datafile ‘+HR’ size 1g;
===> correct one would be:
create undo tablespace UNDOTBS2 datafile ‘+HR’ size 1g;

cheers
Adrian

Hi, thanks for the beautiful explanation here. My doubt, is it possible to have same database name but having diff DBID and DB_UNIQUE_NAME while duplicating DB using RMAN. If yes what should I use with RMAN duplicate command dbname or db_unique_name

Hi Kamran

Great post, very useful and professional. Just wondering if all recommended changes are in the same post or different one?!

ba tashakoor faravan

Hormoz

Hi Kamran

Helpful post, What should be the process if I want to duplicate from RAC database to single instance database without ASM. I think in the auxiliary parameter file need to set
cluster_database=false;
db_create_file_dest=non_asm_file_location;
alter system set db_create_online_log_dest_1=non_asm_file_location;
DB_FILE_NAME_CONVERT and LOG_FILE_NAME_CONVERT parameters

Do I need any command regarding logfile thread and undo tablespace after issuing

. oraenv
HRPRD1
rman target sys/[email protected] auxiliary sys/[email protected]
RMAN>run{
DUPLICATE TARGET DATABASE TO HRPRD
FROM ACTIVE DATABASE;
}

or

DUPLICATE DATABASE TO HRPRD
FROM ACTIVE DATABASE
PFILE (if starting the auxiliary instance with a client-side parameter file)
NOFILENAMECHECK; (if duplicated datafiles location= Source data file’s location and database’s are in DIFFERENT Host’s)

or is there any article DUPLICATE from ACTIVE Database Using RMAN from rac to single instance database:

Hello all,
first thank you for this article.
I need your help …. I installed oracle 11g on my pc “desktop”, with a database that used by an application that contain 2 parts ” the 2 parts use the same tables of the database” . now I want to separate the 2 parts of the application and I have to duplicate my database on the same oracle installation, in way that I can use the 2 parts of the application in the same time separately and the changes makes by the part 1 on database 1 don’t affect the database 2.
is this possible ? if yes how ? …. please help me …
THANK YOU….

Could you please explain how duplicate works internally?Does it copy the data-files using any operating system copy utility like rsync or scp in unix.

Interesting and useful information shared by you. If you want to remove duplicate data, then it would be great idea to use a data deduplication software. It helps to find and remove duplicate data easy and fast way.

Hi..,

This is an interesting feature, though it has its own disadvantages.

I have tried to do the same in 11.2.0.3 databases, but during duplication, the RMAN did not consider all the datafiles – i have created 4 datafiles before i started the duplication, and during the set up i did not consider these 4 datafiles, any particular reason for this not to consider these files ???

Because of this the recovery stopped. Can you please let me know the reason for this ?

Hi there,

I guess all of this is not possible to be done between one single instance on SunSolaris to RAC instances on Linux because of endian. Which RMAN does not support. Right ?

Pierre-Jean.

Hi Kamran,
in my lab I was doing the opposite, from a rac db as target and clone to a single db.
I have executed the steps accordingly, however at the end of the process I got an error that sounds different from what it should be.
Please take a look at this:
—
contents of Memory Script:
{
Alter clone database open resetlogs;
}
executing Memory Script

RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03002: failure of Duplicate Db command at 03/06/2014 18:55:47
RMAN-03015: error occurred in stored script Memory Script
RMAN-06136: ORACLE error from auxiliary database: ORA-01152: file 2 was not restored from a sufficiently old backup
ORA-01110: data file 2: ‘/oracle/oradata/usa/sysaux01.dbf’
—

Do you have any suggestion what should I do in order to open the cloned database??
Thank you and congrats for the great article.

Interesting post Kamran! thanks

DUPLICATE TARGET DATABASE TO HRPRD
FROM ACTIVE DATABASE;

What happens internally in detail

Great post Kamran ..

Quick question –

In Active clone, say application/processes are up and running on source DB and it’s a busy DB with high DML’s and I start active clone say at 12:30 p.m and this clone takes 4 hours to complete. So what time data I’ll see at target side? Will it be as of 12:30 p.m or as of 4:30 p.m? Or it’ll be inconsistent?

Thanks in advance for answering on this.

— Ankit

Starting Duplicate Db at 11-MAY-15
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03002: failure of Duplicate Db command at 05/11/2015 22:52:59
RMAN-05501: aborting duplication of target database
RMAN-06403: could not obtain a fully authorized session
RMAN-04006: error from auxiliary database: ORA-01034: ORACLE not available

I have tested the sys connection on both the target and aux database and it fine. I have started up the aux database in nomount state. Kindly share your thoughts
ORA-27101: shared memory realm does not exist
IBM AIX RISC System/6000 Error: 2: No such file or directory

Hi Kamran, great post.

Do you have another one to make the duplicate active database for standby?

Thank you.

Regards.
Antranik.-

I have very large database which is around 6 TB running on 11g R2 , I need to complete the duplicate database with in 12 Hours so that i can start the 12c Upgrade .

how to increase the rman channels to duplicate database from active database .

how to increase the performance to duplicate database from active database .

Hi
Many thanks for our post.
Saved me lot of trouble and the process worked alright .

Thanks
Amit

Reviving an old post, but I found something missing this afternoon.
If the source database is on ASM, the TNS aliases should be added to both the tnsnames.ora of the ASM and database OH, lest you get a spurious ORA-12154 – even though you’ve tested connectivity both ways with the database OH

Hi,

Thanks for your post. Its quite helpful. I have a question about applying redo logs on duplicate database. Our DB size is 2 TB and we wish to enable active duplicate and then keep rolling forward for X days until we decide to break it and then run application upgrade on duplicate database. Is it possible to apply redo logs continuously on auxiliary database ?

I am in a mess. I am trying to clone my database. But somehow, after connecting to RMAN, both, connected to target database and connected to auxiliary database name turns out to be same. I have checked both listener and tns service details carefully. Is there any way I can send the screen shot of it?

4- Ceate a password file for auxiliary instance HRPRD1 on prd-db-01:
— Spelling is incorrect, Kindly rectify.

Thanks for the posting.
Can you control to what point in time this duplicate would restore the database?

Thanks for the details Kamran and everyone who given his valuable feedback. I have one question on active duplicate DB.
– What about if tablespace altered(DataFile resize,Add Datafile..etc) performed on target database during the active duplicate process.
– What about the recent changes on target db.. as i understand during the whole process CF will be locked on certain point recovery will happened till that point, but still like to understand in brief.

Request you to please share your input.

Thanks.
Raj Gupta

One more requirement to keep in mind: Both databases must use the same password.

This is an amazing post Kamran. For my situation, I found that a couple of additional means were required anyway on the grounds that I got this mistake after RMAN dup finished (just HRPRD1 will begin (in my experience). Here is the mistake message:

does rman duplicate uses port 22 to copy data between databases.