MongoDB script to backup replication set locally to a Windows Server - mongodb

Question

I would like to make a Daily Backup of my MongoDB from a replication set running from Windows 2012 servers.
End goal would be to get a daily backup and write the backup to a remote or local share - Windows.
Can I batch the mongodump command?
Any help would be greatly appreciated!!

Answer 1

Sorry, it's a bit late but the following seems to work OK for me. The script dumps the database and compresses the output using 7-Zip.
1) Create backup script (backup.bat)
#echo off
REM move into the backups directory
CD C:\database_backups
REM Create a file name for the database output which contains the date and time. Replace any characters which might cause an issue.
set filename=database %date% %time%
set filename=%filename:/=-%
set filename=%filename: =__%
set filename=%filename:.=_%
set filename=%filename::=-%
REM Export the database
echo Running backup "%filename%"
C:\mongodb\mongodump --out %filename%
REM ZIP the backup directory
echo Running backup "%filename%"
"c:\Program Files\7-Zip\7z.exe" a -tzip "%filename%.zip" "%filename%"
REM Delete the backup directory (leave the ZIP file). The /q tag makes sure we don't get prompted for questions
echo Deleting original backup directory "%filename%"
rmdir "%filename%" /s /q
echo BACKUP COMPLETE
2) Schedule the backup
Open Computer Management
Go to Task Scheduler and select Create Task.
On the General tab, enter a description and select Run whether user is logged on or not if you want the task to run at night.
On the Triggers tab, select when you would like the task to run.
On the Actions tab, create a new action which points at your batch script.

Answer 2

I'm running on linux, not Windows 2012, but here is what I do. On one of the servers in the replica set, this script gets run every night via a cron job.
#config
BACKUPNAME=[backup file name]
DATAPATH=[path to mongo data folder]
DATESTAMP=$(date +"%Y-%m-%d")
FILENAME=backup.$BACKUPNAME.$DATESTAMP.tar.gz
TARPATH=$DATAPATH/$FILENAME
echo $DATESTAMP;
/etc/init.d/mongod stop
/usr/bin/mongodump --journal --dbpath $DATAPATH --out $DATAPATH/backup
tar czvf $TARPATH $DATAPATH/backup
rm -rf $DATAPATH/backup
/usr/bin/s3cmd put $TARPATH s3://[backup s3 bucket name]/$FILENAME
rm -f $TARPATH
/etc/init.d/mongod start
/scripts/prunebackups
I'm using s3cmd to send files to an S3 bucket on Amazon AWS, but you could just as easily copy the file anywhere. prunebackups is a script that deletes old backups from S3 based on how old they are.
On Windows I'd create a batch file that does similar tasks. In essence:
Stop mongod
run mongodump to generate the data
zip up the dumped data and move it somewhere
clean up files
start mongod again
You can then use Task Scheduler to run it periodically.
If you have other mongod instances in the replica set, you shouldn't run into any issues with downtime. The backup instance in my setup is never used for reads or writes, but only for backups and in case one of the other instances goes down.
MongoDB has documentation on different backup strategies: http://docs.mongodb.org/manual/administration/backup/
We chose the mongodump approach because for us it's cheaper to store dump backups than snapshots. This is the specific strategy we used: http://docs.mongodb.org/manual/tutorial/backup-databases-with-binary-database-dumps/. Good news is, we actually had to restore data from a backup to production once and it was pretty painless.