Rman-10038: Database Session For Channel D1 Terminated Unexpectedly

Applies to:

Oracle Database - Enterprise Edition - Version 10.2.0.1 and later
Information in this document applies to any platform.
***Checked for relevance on 14-Oct-2013***

Symptoms

On 10.2.0.1 in Production:
When attempting to backup the database with RMAN the following error occurs.

ERROR
Rman-10038: database session for channel d1 terminated unexpectedly rman-10038

STEPS
The issue can be reproduced at will with the following steps:

1. RMAN > run {
allocate channel d1 device type disk maxpiecesize 4G format
'/db01/backup/rman/ebusp/$today/db_%U';
backup database;
release channel d1;
}

2. When run the backup terminates after completing the last backup piece with the error stack as below:

c

hannel d1: starting piece 1 at 11-NOV-06
channel d1: finished piece 1 at 11-NOV-06
piece handle=/db01/backup/rman/ebusp/20061111/db_17i23l29_1_1 tag=TAG20061111T040008 comment=NONE
channel d1: starting piece 2 at 11-NOV-06
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03009: failure of backup command on d1 channel at 11/11/2006 04:07:41
RMAN-10038: database session for channel d1 terminated unexpectedly

3. In the alert.log for the target database there are also ORA-07445 errors such as below:

ORA-07445: exception encountered: core dump [kzldbind()+180] [SIGSEGV] [Address not mapped to object] [0x0] [] []
Errors in file /app01/oracle/db/ebusp/10.2.0/admin/ebusp1_wn0lxofdbx01/udump/ebusp1_ora_31140.trc:
ORA-07445: exception encountered: core dump [_intel_fast_memcpy.A()+14] [SIGSEGV] [Address not mapped to object] [0x0] [] []

BUSINESS IMPACT:
Due to this issue, users are at risk due to the backup failing to complete successfully.

Cause

The issue was reported in bug : Bug 5195481 ORA-7445 IN RMAN AFTER CHANGING COMPATIBLE TO 10.2.0.

Bug has been closed as Not a Bug.

The issue appears when customers do not have the autobackup of the controlfile switched to ON in the RMAN
configuration which is similar to the bug reported.

The ORA-07445 errors ONLY occur after completing the backup and writing the final backup piece in the data collected RMAN output. The ORA-07445 error : [_intel_fast_memcpy.A()+14] also match the bug details as above.

The issue can occure aswell when the spfile is not in the $ORACLE_HOME/dbs directory or a symbolic link to the spfile in the $ORACLE_HOME/dbs directory.

Solution

To implement the solution, please execute the following steps:

1. Ensure controfile autobackups are set to ON in the RMAN configuration.

To check, invoke rman connect to target and catalog as appropriate and run:

RMAN>show all;

if you see this: CONFIGURE CONTROLFILE AUTOBACKUP OFF;

then run:

RMAN> CONFIGURE CONTROLFILE AUTOBACKUP ON;

And try the RMAN backup again, if still an issue then please confirm the below in regards to the spfile.

3. Check the location and presence of the spfile?

SQL> show parameter spfile;

If an spfile is being used, then make sure it is in the $ORACLE_HOME/dbs directory.
If not, create a symbolic link to it from $ORACLE_HOME/dbs:

$ ln -s <path>/<spfile> $ORACLE_HOME/dbs/<spfile>