MongoDB Database Backup and Restore

MongoDB stores all its data in the data directory mentioned by --dbpath during startup. Taking backup of MongoDB is as simple as taking a copy of all data files in the data directory.

Creating a copy of datafile is not recommended while MongoDB is running as it can create a corrupt or useless backup. It is always good to take a copy of datafile after safely shutting down the MongoDB server, but this will surely require downtime. In order to avoid downtime, we can use the below methods to take backup.

MONGODUMP

One of the methods of taking MongoDB backing while the database is running is to use the mongodump utility. mongodump generally works on query mechanism, it will cause slight performance degradation on the database during the backup time.

It can be run against a live instance, even one handling other requests and performing writes.

We can see the options with mongodump using mongodump –help

C:\Program Files\MongoDB\Server\3.6\bin>mongodump --help

Usage:mongodump

/help           print usage

/v, /verbose:   more detailed log output (include multiple                           times for more verbosity, e.g. -        vvvvv, or                      specify a numeric value, e.g. --verbose=N)

/h, /host:      mongodb host to connect to                                   (setname/host1,host- 2 for replica sets)

/u, /username:     username for authentication

/p, /password:     password for authentication

/d, /db:           database to use

/c, /collection:   collection to use
/uri:mongodb-uri   mongodb uri connection string
/q, /query:         query filter, as a JSON string,                               e.g., '{x:{$gt:1}}'
/o, /out:          output directory,

gzip              compress archive our collection output                               with Gzip

Below is an example for taking the backup of database "techon_db" in directory D:\mongodb\backup_dump

mongodump -d techon_db -o D:\mongodb\backup_dump

MONGORESTORE

Along with mongodump, there comes mongorestore for importing data in MongoDB which was taken using mongodump, backup taken using mongodump can be restored in different or new database as well.

Example

mongorestore -d techon1_db  D:\mongodb\backup_dump\techon_db

In the above example, we have restored backup taken using mongodump for database techon_db into techon1_db. all steps are clearly displayed in the screenshot.

all options can be displayed using mongorestore --help

fsync and Lock

As we already discussed, the above methods allow us to perform its operations while the database is running, but here we tend to lose some ongoing activity backup as we will miss on point in time recovery backups. MongoDB's fsync command allows us to copy the data directory of the running MongoDB server while the database is running.

fsync command forces to flush all pending writes to disk. It also, optionally holds a lock preventing any further writes to the database until it is unlocked.

Example of fsync

>db.runCommand ({"fsync" : 1, "lock" : 1});

{

"info" : "now locked against writes, use db.$cmd.sys.unlock.findOne() to unlock",

"ok" : 1

}

This will allow us to take consistent backup. Once the backup complete, we need to unlock the database

>db.$cmd.sys.unlock.findOne();

{"ok" : 1, "info" : "unlock requested"}

>db.current0p ();

{"inprog" : []}

The meaning of db.current0p is confirmation of the lock has been released.

Pros for fsync are only putting a lock on reads or write on the database. The only solution for this is Slave Backup

Slave Backup

All methods discussed above provide enough flexible backups, but the most flexible backup is taking backup from a slave server. When running MongoDB with replication, above-all methods can be applied to take a backup from the slave server instead of the primary one

Slave backup of the recommended way to backup MongoDB database