validate_backup()
{
# To validate the success of the backup

notify -l "Validating rman backup job ..."

RMN_STATUS=`grep "Recovery Manager complete" $RMN_LOG_FILE | tail -1 | nawk '{print $3 }'`;

if [ $DEBUG = "ON" ]; then
echo RMN_STATUS : $RMN_STATUS
fi

if [ "${RMN_STATUS}" = "complete." ];
then
notify -l "RMAN BACKUP JOB `basename ${RMN_CMD_FILE}` SUCCESSFUL !!!"
else
notify -blm "RMAN BACKUP JOB `basename ${RMN_CMD_FILE}` FAILED !!!"
exit ${EXIT3};
fi

return;
}

where ${RMN_CMD_FILE} is what you see below in the log file:


[07/05/05-01:05:02] Starting /u01/app/oracle/product/9.2.0/bin/rman /u01/app/oracle/admin/cbep2/rman/full_hbk_disk.rcv ...
[07/05/05-01:06:12] Validating rman backup job ...
[07/05/05-01:06:12] RMAN BACKUP JOB archive_all_disk.rcv SUCCESSFUL !!!
[07/05/05-01:06:12] mv /u01/app/oracle/admin/cbep2/rman/logs/full_hbk_disk.rcv.log /u01/app/oracle/admin/cbep2/rman/logs/full_hbk
_disk.rcv.logO
[07/05/05-01:06:13] Starting /u01/app/oracle/product/9.2.0/bin/rman /u01/app/oracle/admin/cbep2/rman/full_hbk_disk.rcv ...
[07/05/05-06:02:09] Validating rman backup job ...
[07/05/05-06:02:10] RMAN BACKUP JOB full_hbk_disk.rcv FAILED !!!
[07/05/05-06:16:42] Validating rman backup job ...
[07/05/05-06:16:42] RMAN BACKUP JOB full_hbk_disk.rcv SUCCESSFUL !!!
[07/05/05-06:16:42] Completed onlinerman V.1.0

We have several Oracle instances that we back up using RMAN. We have cronjobs for those RMAN jobs to run. The script that is run call the above function to validate and send page depending on the status of the RMAN. For only two of the instances we get paged that "RMAN BACKUP JOB blablabla FAILED !!!", but in fact it did not fail. As we can see in the log , after about 14 minutes, that it was SUCCESSFUL.
When we check backup logs everything looks all right. I do not see any problem with the script as well. Any idea why do we get paged?

Any input will be appreciated!
Kal

Could you please post also the file on which you apply grep ?

$ tail full_hbk_disk.rcv.log

released channel: CBE_CHNL_2

released channel: CBE_CHNL_3

released channel: CBE_CHNL_4

host command complete

Recovery Manager complete.

Now a quesition about the script. The command:

RMN_STATUS=`grep "Recovery Manager complete" $RMN_LOG_FILE | tail -1 | nawk '{print $3 }'`;

will always put in RMN_STATUS complete, unless the file is $RMN_LOG_FILE is overwritten. Is this the case ?

Why don't you set DEBUG to on ? (like DEBUG=ON)

And I think the if should be:

if [ "$DEBUG" = "ON" ]; then
echo RMN_STATUS : $RMN_STATUS
fi

Yes I have that as well. The entire script is too large.


create_log_file_name()
{
# Retrieve the rman cmdfile and construct the logfile name.

RMN_CMD_FILE=$1

base_name=`basename ${RMN_CMD_FILE} | sed -e 's/\.cmd//'`
RMN_LOG_FILE=${RMN_LOG_DIR}/${base_name}.log

if [ $DEBUG = "ON" ]; then
echo RMN_LOG_FILE : $RMN_LOG_FILE
fi

if [ -z "${RMN_LOG_FILE}" ];
then
notify -blm "** ERROR: Unable to set RMN_LOG_FILE"
exit ${EXIT2};
fi

if [ -f "${RMN_LOG_FILE}" ];
then
notify -l "mv $RMN_LOG_FILE ${RMN_LOG_FILE}O"
mv $RMN_LOG_FILE ${RMN_LOG_FILE}O > /dev/null
fi

return;
}

When you run the validate..() , it's possible that the backup logwriter has not completely finished ie some output is still buffered in memory, not yet output to disk.
Try printing the value for ${RMN_STATUS} when it fails.

That sounds good, can I put some kind of delay before calling the function? If so can you please give me an example how can I fit that into the calling environment below:

#####
#
# MAIN ROUTINE
#
#####

REV="1.0"
HOST=`uname -n`
PROG=`basename $0`
ORA_HOST=`hostname`
# Set a variable that corresponds to 'whitespace'. This variable should
# contain a space and a tab. In 'vi', 'set list' will show the following:
#
# WS=" ^I"$
#
WS=" "

# Create some variables to simplify regular expression evaluations
#
# WA - Any number of white space character(s)including 0
# WN - Any character(s) including none except white space character(s)
# W1 - At least 1 white space character(s)
#
WA="[$WS]*"
WN="[^$WS]*"
W1="[$WS]+"

# Exit codes
#
# EXIT0 Backup Successful
# EXIT1 Usage error or execution termination by the user
# EXIT2 Missing executables or unset environment variables
# EXIT3 Backup Unsucessful
#
EXIT0=0
EXIT1=1
EXIT2=2
EXIT3=3

#
# Begin
#


if [ $# -eq 0 ];
then
usage;
fi

while getopts ":s:a:b:d" OPTM "$@"
do
case "${OPTM}" in
s) SID=${OPTARG};;
a) ARC_SCRIPT=${OPTARG}
S_SCRIPT=${OPTARG}
ARC_OBK='Y';;
o) ORA_SCRIPT=${OPTARG}
S_SCRIPT=${OPTARG}
ORA_OBK='Y';;
b) BPERSON=`cat /com/oracle/omon/config/on_call_dba`;;
d) DEBUG="ON";;
*) notify -p "**ERROR : Invalid option -${OPTARG}"
usage;;
esac
done

set_def_vars;
set_local_env;
notify -l "Starting onlinerman V.${REV}"

set_ora_env;
set_rman_env;
chk_plus;
chk_rman;
# chk_oracle;

if [ ${ARC_OBK} = 'Y' ];
then
chk_arc_rman_script;
fi

if [ ${ORA_OBK} = 'Y' ];
then
chk_ora_rman_script;
fi

if [ ${ARC_OBK} = 'Y' ];
then
create_log_file_name "${RMN_ARC_SCRIPT}";
run_rman "${RMN_ARC_SCRIPT}";
if [ $? -eq 0 ];
then
validate_backup;
fi
fi

if [ ${ORA_OBK} = 'Y' ];
then
create_log_file_name "${RMN_ORA_SCRIPT}";
run_rman "${RMN_ORA_SCRIPT}";
if [ $? -eq 0 ];
then
validate_backup;
fi
fi

notify -l "Completed onlinerman V.${REV}"
cp $RMN_LOG_DIR/${OJOB}.rcv.log /dbbackup/oracle/backup/$SID
exit ${EXIT0};

#----------------------------------------------------------------------------

Appreciate
Kal