https://training-course-material.com/index.php?action=history&feed=atom&title=Backup_and_Restore_in_MongoDBBackup and Restore in MongoDB - Revision history2026-06-02T21:25:49ZRevision history for this page on the wikiMediaWiki 1.45.1https://training-course-material.com/index.php?title=Backup_and_Restore_in_MongoDB&diff=71388&oldid=prevKbaran: /* Backup of Replica Set */2019-04-28T11:51:09Z<p><span class="autocomment">Backup of Replica Set</span></p>
<p><b>New page</b></p><div>[[Category:MongoDB]]<br />
<br />
== Backup and Restore ==<br />
* backups generally should be done on secondaries<br />
* or on standalone servers at an off time<br />
<br /><br />
<br />
=== Method #1: Filesystem Snapshot ===<br />
* the simplest way to make backup<br />
* filesystem must support snapshotting<br />
* mongod must run with journaling enabled<br />
* stop mongod before restoring<br />
<br /><br />
<br />
<syntaxhighlight lang="bash"><br />
# Create the snapshot volume<br />
lvcreate -L128M -s -n dbbackup /dev/ops/databases<br />
<br />
# Mount the snapshot volume<br />
mkdir /mnt/ops/dbbackup<br />
mount /dev/ops/dbbackup /mnt/ops/dbbackup<br />
<br />
# Do the backup<br />
tar -cf /dev/rmt0 /mnt/ops/dbbackup<br />
<br />
# Remove the snapshot<br />
umount /mnt/ops/dbbackup<br />
lvremove /dev/ops/dbbackup <br />
</syntaxhighlight><br />
<br /><br />
<br />
=== Method #2: Copying Data Files ===<br />
* use fsyncLock() before copying files<br />
** fsycnLock() prevents all databases against any further write<br />
** saves all dirty data to disk<br />
** queues all write operations<br />
* now files are consistent and copying of all files is possible<br />
* fsyncUnlock() releases the lock and brings back database to normal operations<br />
* if using authentication do not log out from the console between fsyncLock() and fsyncUnlock()<br />
* stop mongod before restoring<br />
** do not restore single database if crash or hard shutdown occurs<br />
<br />
<br /><br />
<syntaxhighlight lang="javascript"><br />
> db.fsyncLock()<br />
> db.fsyncUnlock()<br />
</syntaxhighlight><br />
<br /><br />
<br />
=== Method #3: Using mongodump and mongorestore ===<br />
* mongorestore is slower than previous methods and has other downsides<br />
* good way to backup individual databases or collections<br />
* it will create a dump directory (with subdirectories for each database) in current directory<br />
* data is stored in .bson files<br />
* mongodump can be used when mongod is not running<br />
** do not use --dbpath when mongod is running<br />
* when mongodump is running writes are allowed so already backuped data may change before mongodump will finish<br />
** do not use fsyncLock() when using mongodump<br />
** use --oplog if you are running mongod with --replSet<br />
* mongodump will choose secondary when connected to replica set<br />
* use mongodump and mongorestore in the same version<br />
* try to avoid this type of backup if you have unique indexes other than _id<br />
<br /><br />
<br />
<syntaxhighlight lang="bash"><br />
mongodump --help<br />
mongodump --port 27017<br />
mongodump --dbpath /var/lib/mongodb<br />
mongorestore --port 27017 --drop<br />
mongodump --port 27017 --oplog<br />
mongorestore --port 27017 --oplogReplay dump/<br />
mongorestore -d dstDB -c dstCollection dump/srcDB/scrColl.bson<br />
</syntaxhighlight><br />
<br /><br />
<br />
=== Backup of Replica Set ===<br />
* all previous methods are OK but 1st and 2nd are recommended (without any modifications)<br />
* when using mongodump use --oplog<br />
* when restoring with mongorestore:<br />
*# start a server as a standalone and<br />
*# restore the data with --oplogReplay and<br />
*# restore the oplog collection<br />
*# restart the server as a member of replica<br />
<br /><br />
<br />
<syntaxhighlight lang="javascript"><br />
> use local<br />
> db.createCollection("oplog.rs", {"capped" : true, "size" : 1000000})<br />
$ mongorestore -d local -c oplog.rs dump/oplog.bson<br />
</syntaxhighlight><br />
<br /><br />
<br />
=== Backup of Sharded Cluster ===<br />
* usually it can not be done perfectly<br />
* instead of backing up everything at once, backup servers separately<br />
* turn off balancer before making backup<br />
* run mongodump through mongos to backup entire cluster<br />
* problems with restoring single shard<br />
<br /></div>Kbaran