Automating the UCSD internal DB Backup for a single node deployment:
The crontab entry
==============+
[root@localhost etc]# crontab -l
1 2 * * 0 /usr/local/etc/DB_BACKUP > /dev/null 2>&1
The Script on UCSD
=================
[root@localhost etc]# cat /usr/local/etc/DB_BACKUP
#!/bin/sh
# server login password localfile remote-dir
upload_script(){
echo "verbose"
echo "open $1"
sleep 2
echo "user $2 $3"
sleep 3
shift 3
echo "bin"
echo $*
sleep 10
echo quit
}
doftpput(){
upload_script $1 $2 $3 put $4 $5 | /usr/bin/ftp -i -n -p
}
/opt/infra/stopInfraAll.sh
/opt/infra/dbBackupRestore.sh backup
BKFILE=/tmp/database_backup.tar.gz
if [ ! -f $BKFILE ]
then
echo "Backup failed. "
return 1
fi
export NEWFILE="cuic_backup_`date '+%m-%d-%Y-%H-%M-%S'`.tar.gz"
export FTPSERVER=172.17.32.110
export FTPLOGIN=cisco
export FTPPASS=cisco
doftpput $FTPSERVER $FTPLOGIN $FTPPASS $BKFILE $NEWFILE
nohup /opt/infra/startInfraAll.sh &
exit 0
mkdir /usr/local/etc
vi /usr/local/etc/DB_BACKUP
cut and paste above text
chmod +x /usr/local/etc/DB_BACKUP
Update crontab
set -o vi
crontab -e
Ideas for Multi Node Backup
=======================
(1) Enable SSH on all hosts with pub key from Inventory DB.
(2) Write a script in such a way that from Inventory, it shutdown all primary, service nodes (hardcoding the IPs)
(3) Invoke a local backup command on the inventory node and copy the backup to NFS
(4) Invoke a remote backup command on the monitoring node and copy the backup to NFS
(5) Start Service nodes, Primary nodes
Put this in a cron on Inventory node.
Another contribution by:
Alejandro Madurga Ainoza
Python scripts to stop the services on a multimode (single node work as well)
In this case customer wanted to store the backup on a server filesystem (that could be NFS mounted).
Scripts attached, they call the OOTB backup scripts but avoid the FTP steps.
