Multiple Oracle DB/Instances on single machine, how to include in one script.

NetRestore, I you can easily  create another script which will copy template policy (as you know policy is just a folder), rename it and add correct script location.

Hi,

Gavrilov, does this mean I will be able to include multiple Database's into one policy, will doing your suggested method enable us to backup all the DB's in one go?

Unfortunately, no. I don't think it's a good idea to add multiple scripts into one policy cause if one DB backup fails you need to restart whole policy. I just wanted to say that if you have alredy created all necessary backup scripts for all DBs you can create a template policy and write a script which will copy the policy, rename it ( for example add SID name) and after that add script location into backup selection. I think it's quite simple. I do understand that you'll have a lot of policies but it will give you flexability and you'll be able to to run individual instancse backup.

Yep, I believe you confirmed my fears relating to this, i know i am going to have to do a policy per DB instance.

Thanks Gavrilov.

You can do it by determining running instance inside script like below. However, as already noticed here, you should not configure single big backup. When backup fails, all DB will backuped up again.

#! /bin/sh

#
DBA_USER=oracle
DEFAULT_ORACLE_HOME=/u01/app/oracle/product/11.2.0/dbhome_1
TARGET_CONNECT_STR=/
LOGFILE=${0}.`date +%Y%m%d`.out
#


SYSNAME=`uname -s`
if [ "${SYSNAME}" = "HP-UX" ];then UNIX_STD=2003; export UNIX_STD; fi
if [ "${SYSNAME}" = "SunOS" ];then
	ORATAB=/var/opt/oracle/oratab
else
	ORATAB=/etc/oratab
fi
if [ "${NB_ORA_POLICY}" = "" ];then
	echo "This script must be kicked by NetBackup Oracle Agent. Do not run manually."
	exit 1
fi
if [ "${NB_ORA_FULL}" = "1" ];then
	BACKUP_TYPE="INCREMENTAL LEVEL=0"
elif [ "${NB_ORA_CINC}" = "1" ];then
	BACKUP_TYPE="INCREMENTAL LEVEL=1 CUMULATIVE"
elif [ "${NB_ORA_INCR}" = "1" ];then
	BACKUP_TYPE="INCREMENTAL LEVEL=1"
else
	BACKUP_TYPE="INCREMENTAL LEVEL=0"
fi

SIDS_SUCCEED=
SIDS_FAIL=
SIDS=`ps -eo args | grep "^ora_dbw0_" | cut -c 10- | grep -v "^+"`
if [ -z "${SIDS}" ];then
	echo "No instance running on this host."
	exit 0
fi
echo "### `date` Backup session started" >> ${LOGFILE}
for sid in ${SIDS}; do

	ORACLE_SID=${sid}

	# deterimine ORACLE_HOME from oratab
	ORACLE_HOME=`grep "^${sid}:" ${ORATAB} | head -n 1 | tr : " " 2>/dev/null`
	set -- ${ORACLE_HOME}
	ORACLE_HOME=$2
	if [ "${ORACLE_HOME}" = "" ];then ORACLE_HOME=${DEFAULT_ORACLE_HOME}; fi

	RMAN=${ORACLE_HOME}/bin/rman
	CMD_STR="
env ORACLE_SID=${ORACLE_SID} ORACLE_HOME=${ORACLE_HOME} ${RMAN} target ${TARGET_CONNECT_STR} nocatalog <<EOM
RUN {
ALLOCATE CHANNEL t1 TYPE 'SBT_TAPE';
ALLOCATE CHANNEL t2 TYPE 'SBT_TAPE';
BACKUP
	${BACKUP_TYPE}
	SKIP INACCESSIBLE
	FILESPERSET 5
	FORMAT 'bk_${ORACLE_SID}_%s_%p_%t'
	DATABASE;
sql 'alter system archive log current';
RELEASE CHANNEL t1;
RELEASE CHANNEL t2;
ALLOCATE CHANNEL t1 TYPE 'SBT_TAPE';
ALLOCATE CHANNEL t2 TYPE 'SBT_TAPE';
BACKUP
	FILESPERSET 20
	FORMAT 'arch_${ORACLE_SID}_%s_%p_%t'
	ARCHIVELOG ALL DELETE INPUT;
RELEASE CHANNEL t1;
RELEASE CHANNEL t2;
ALLOCATE CHANNEL t1 TYPE 'SBT_TAPE';
BACKUP
	FORMAT 'ctrl_${ORACLE_SID}_%s_%p_%t'
	CURRENT CONTROLFILE;
RELEASE CHANNEL t1;
}
EOM
"
	echo "==== ${ORACLE_SID} backup started on `date` ====" >> ${LOGFILE}
	su - ${DBA_USER} -c "${CMD_STR}" >> ${LOGFILE} 2>&1
	RC=$?
	echo "==== ${ORACLE_SID} backup ended on `date` : rc=${RC} ====" >> ${LOGFILE}
	if [ ${RC} -ne 0 ];then
		SIDS_FAIL="${SIDS_FAIL} ${ORACLE_SID}"
	else
		SIDS_SUCCEED="${SIDS_SUCCEED} ${ORACLE_SID}"
	fi
done

RC=`echo "${SIDS_FAIL}" | wc -w`
echo "SIDs SUCCEEDED: ${SIDS_SUCCEED}" >>${LOGFILE}
echo "SIDs FAILED: ${SIDS_FAIL}" >>${LOGFILE}
echo "### `date` Backup session ended rc=${RC}" >>${LOGFILE}

exit ${RC}

 Yasuhisa, I think NetRestore was looking for some option to backup all DBs togather cause he didn't want to backup DBs one-by-one. In your script you just do the same thing but in more fancy way but the script is nice and we can avoid using one script per DB.

P.S. Your script expects that  ORACLE_USER is the same for all SIDs -- it might be not true.

Yes Gavrilov you are right, but for now we are resorting to creating a policy per script and per DB instance, we then only allocate RMAN one channel and for netbackup to not exceed 2 streams per policy, lets see how this works.

Well that did not work out so well, so we resorted to builiding a ZFS storage pool and sharing a mount point via NFS to the Oracle host, we then backup to the NFS share, and then snapshot the NFS mount point on the ZFS host which we then backup the snapshot. In this case far more efficient than Netbackup for Oracle.

Quicker backup may mean longer restore time though.  Have you tried to restore and recover yet?