Last couple of months I was wandering through the world of MongoDb, both as a developer and as a sysadmin. I won’t say how happy I’m with it or how much I got disappointed , I will just note that you don’t know how deep or serious you dived into a software product till the moment when you feel the need for a backup. And that day came, and I did something about it.
I’m not so much into installing backup software that will just do the things for me, I always start with the idea that I can do it with my own simple implementation that will fill my needs for saving data or doing something else. Mongo has become a huge software/database monster and there are a lot of different approaches regarding this question. Off course the right way for you depends of several factors. Some of them are: the infrastructure of the Mongo deployment, the importance of the data, the quantity of the data, the performance factor etc.
For my needs that are associated with a single cluster instance of Mongo database without replication or sharding I realized that using the “classical” dumping method will fulfil .
One of the things I like about Mongo is its nicely done documentation. A lot of information about backuping MongoDb can be found on its official site:
http://docs.mongodb.org/manual/core/backups/
The documentation is like a guide showing you the different ways of backuping. Generally there are 3 different approaches: doing file system snaphosts, using mongodump command and using MongoDB Management Service.
For my dumping approach I wrote simple and primitive bash shell script that you can use for local backups or as a push towards the idea how to backup mongo data. It is oriented around dumping databases. Mongo data can be dumped as whole instance, whole collection or part collection and as a database. Here is the scirpt, the product of it are BSON and metadata JSON files at the designated directory.
#!/bin/bash
#
# GNRP LICENSE: Script licensed by GOD
#
# @author: Igor Ivanovski <igor at genrepsoft.com>
#
# March, 2015
#
#
# Filesystem directory variables
#
# Now in format: YYYY-MM
DATEYYMM=`date +"%Y-%m"`
# Now in format: DD
DATEDD=`date +%d`
# Backup directory
BACKUPDIR="/opt/backup/mongodumps"
# Daily backup directory
BACKUPDIRDAILY="$BACKUPDIR/$DATEYYMM/$DATEDD/"
#
# List of databases to backup
#
DBs="
admin
someDb
";
#
### Mongo Server Setup ###
#
# Don't forget to add adequate roles if you are using authenticatio
#use someDb db.createUser({user:"backup",pwd:"pwd",roles:["readWrite"]})
#use admin db.createUser({user:"backup",pwd:"pwd",roles:["backup"]})
# Mongo backup username
MUSER="backup"
# Mongo backup password
MPASS="pwd"
# Mongo HOST name
MHOST="localhost"
# Mongo PORT number
MPORT="27017"
# Mongo dump binary
# Check if mongodump is installed
STATUS=0
[ "Y`which mongodump`" != "Y" ] && STATUS=$? || STATUS=$?
[ $STATUS == 1 ] && echo "No mongodump found. Exiting"; exit 1
MONGODUMP=`which mongodump`;
# Starting to dump databases one by one
if [ "X$DBs" != "X" ]; then
for db in $DBs
do
echo "Backing up database $db"
$MONGODUMP --host $MHOST --port $MPORT --username $MUSER --password $MPASS --out $BACKUPDIRDAILY --db $db
done
else
echo "All listed mongo databases dumped. Bye"
fi
Thank you for reading me, here is a bonus from me for swinging 😉
Groggy Man Reads The Man 