Oracle 10G R2. Duplicate database with RMAN fails

[caesardutta]

The basic problem is given below

Code: Select all

channel ORA_AUX_DISK_1: reading from backup piece /archive/backup/rman/cscdb1/CSCDB_data_full_14423_1_2nnvu37h
ORA-19870: error reading backup piece /archive/backup/rman/cscdb1/CSCDB_data_full_14423_1_2nnvu37h
ORA-19505: failed to identify file "/archive/backup/rman/cscdb1/CSCDB_data_full_14423_1_2nnvu37h"
ORA-27037: unable to obtain file status
Linux-ia64 Error: 2: No such file or directory
Additional information: 3
failover to previous backup


But I see that the file "/archive/backup/rman/cscdb1/CSCDB_data_full_14423_1_2nnvu37h" very much exists.

Previously I found that there were backup pieces which does not exists. So I issued

Code: Select all

RMAN> DELETE NOPROMPT OBSOLETE;

RMAN> DELETE NOPROMPT EXPIRED BACKUP;

RMAN> CROSSCHECK BACKUP;


After issuing the above sequence of commands I am getting the current error as mentioned above.

From the target database I have connected as

Code: Select all

[linux_prompt] rman target / auxiliary sys/passwd@NEWDB

RMAN> duplicate target database to NEWDB;


Please help

[Tim...]

Hi.

If Oracle is reporting the file is missing, this means one of the following:

- The file is actually missing.
- The file is present, but Oracle does not have permissions to access it, or the directory structure leading up to it.
- Some mysterious but has occurred.

Remember, if you are doing a backup-based duplication, the backup files need to be on the server where the duplicate will run. Access to the files on the primary server is irrelevant.

Cheers

Tim...

[caesardutta]

Hi Tim:

I want to keep this very simple. First let me mention the steps instead of providing RMAN logs and increase size of your database.

Oracle 10GR2 on RHEL IA-64

Step 1

AUXILIARY
-------------------
>> Single instance
>> Init file created with DB_FILE_NAME_CONVERT and LOG_FILE_NAME_CONVERT
>> started instance in nomount mode

TARGET
-------------------------
>> Two Node RAC Cluster
>> RMAN runs every night at 8:00PM in increment 0 mode i.e. full database backup
>> Backup is stored at /archive/backup/rman/node1 --> SAN mount point
>> RMAN target / auxiliary sys/pwd@NEWDB
RMAN> DUPLICATE TARGET DATABASE TO NEWDB
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
Backup piece not found
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::
Restoring
Restoring
backup piece /bbbb/mmmm/lllllll --> At this point I checked that database files appeared on AUXILIARY instance

Restoring
Restoring
backup piece /aaa/sss/sss --> At this point I checked that database files appeared on AUXILIARY instance
:::::::::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::::::

Backup piece not found

Then a series of errors mentioning datafile <number> not found in backup piece

RMAN aborts

Step 2

While still connected to TARGET and AUXILIARY, and the AUXILIARY still in nomount mode

RMAN> CROSSCHECK BACKUP;
RMAN> DELETE NOPROMPT EXPIRED BACKUP;
RMAN> DELETE NOPROMPT OBSOLETE;

RMAN> DUPLICATE TARGET DATABASE TO NEWDB

::::::::::::::::::::::::::::::Not a single backup piece found, identified:::::::::::::::::::::::::::::::

Strange thing is that all the backup piece (which are mentioned as not found) do exists

warm regards,

Caesar

[Tim...]

Hi.

Don't know what to say. Maybe you should raise an SR with Oracle support. If the files are present and accessible by oracle this should not happen.

Cheers

Tim...