Adding a New Set Member

Adding a new member to an existing replica set is easy. The new member should either have an empty data directory or a recent copy of the data from another set member. When starting mongod on the new server, provide the replica set name:

$ mongod --replSet foo

After bringing up the new server (we'll call it broadway:27017) we need to add it to the set – we connect to our primary server using the shell:

$ mongo --host our_primary_host
MongoDB shell version: ...
connecting to: test
PRIMARY> rs.add("broadway:27017");
{ "ok" : 1 }

After adding the node it will synchronize (doing a full initial sync if starting empty, see below) and then come online as a secondary.

You can also specify any member configuration options using rs.add(). Some examples:

> // add an arbiter
> rs.add({_id: 3, host: "broadway:27017", arbiterOnly: true})
>
> // add a hidden member
> rs.add({_id: 3, host: "broadway:27017", priority: 0, hidden: true})
>
> // add a member with tags
> rs.add({_id: 3, host: "broadway:27017", tags: {dc : "nyc", rack: "rack1"}})

Adding a former member

A member can be removed from a set and re-added later. If the removed member's data is still relatively fresh, it can recover and catch up from its old data set. See the rs.add() and rs.remove() helpers.

If there is any trouble re-adding the member, restart its mongod process. You may also need to add it back with its former _id (although probably not if mongod were restarted).

Starting with an existing copy of the dataset

If you have a backup or snapshot of an existing member, you can move the data files to a new machine and use them to quickly add a new member. These files must be:

• clean: the existing dataset must be from a consistent snapshot / backup of the database from a member of the same replica set. See the backups page for more information on copying and snapshotting databases.
• recent: the snapshot/backup must have been taken more recently than the oldest operation in the primary's oplog (so that it can catch up by just copying from the oplog)

You do not have to make any modifications to the files before starting up a member.

In versions before v2.0, you must start the new member before running rs.add(). In v2.0+ you can do it in either order, although a majority of the new set must be up for the reconfig to work. Two examples:

• If you have two members and you're adding a third, you can add it to the set before it's live and then bring up the third member. Or you can bring up the third member and then add it to the set. Either way will work.
• If you have a one-member set and you're adding a second member, you must have the second member up before you add it to the set. Otherwise, after the reconfig you'll have one out of two members up (not a majority) and the primary will have to step down.

Additionally with versions before v2.0 you need to restart mongod on the newly added node after the rs.add().

See Also

• Adding an Arbiter