Replica Set Arbiter
On this page
In some circumstances (such as when you have a primary and a secondary, but cost constraints prohibit adding another secondary), you may choose to add an arbiter to your replica set. An arbiter participates in elections for primary but an arbiter does not have a copy of the data set and cannot become a primary.
An arbiter has exactly 1 election vote. By default an arbiter has
priority 0.
Important
Do not run an arbiter on systems that also host the primary or the secondary members of the replica set.
Warning
Using a primary-secondary-arbiter (PSA)
architecture for shards in a sharded cluster can cause a loss of
availability if a data-bearing secondary is unavailable. A PSA
cluster differs from a typical replica set: In a sharded cluster,
shards perform w: majority write concern operations that cannot complete if the remaining
cluster members required to confirm an operation have an arbiter.
To add an arbiter, see Add an Arbiter to a Self-Managed Replica Set.
Example
For example, in the following replica set with a 2 data bearing members (the primary and a secondary), an arbiter allows the set to have an odd number of votes to break a tie:
Performance Issues with PSA replica sets
If you are using a three-member primary-secondary-arbiter (PSA) architecture, consider the following:
The write concern
"majority"can cause performance issues if a secondary is unavailable or lagging. For advice on how to mitigate these issues, see Mitigate Performance Issues with a Self-Managed PSA Replica Set.If you are using a global default
"majority"and the write concern is less than the size of the majority, your queries may return stale (not fully replicated) data.
Shard Keys, Transactions, and Arbiters
You cannot change a shard key using a transaction if the replica set has an arbiter. Arbiters cannot participate in the data operations required for multi-shard transactions.
Transactions whose write operations span multiple shards will error and abort if any transaction operation reads from or writes to a shard that contains an arbiter.
Concerns with Multiple Arbiters
Use a single arbiter to avoid problems with data consistency. Multiple arbiters prevent the reliable use of the majority write concern.
To ensure that a write will persist after the failure of a primary node, the majority write concern requires a majority of nodes to acknowledge a write operation. Arbiters do not store any data, but they do contribute to the number of nodes in a replica set. When a replica set has multiple arbiters, it's less likely that a majority of data bearing nodes will be available after a node failure.
Warning
If a secondary node falls behind the primary, and the cluster is
reconfigured, votes from multiple arbiters
can elect the node that had fallen behind. The new primary will not
have the unreplicated writes even though the writes could have been
majority committed by the old configuration. The result is data
loss.
To avoid this scenario, use at most a single arbiter.
Security
Authentication
When running with authorization, arbiters exchange credentials with
other members of the set to authenticate. MongoDB encrypts the
authentication process, and the MongoDB authentication exchange is
cryptographically secure.
Because arbiters do not store data, they do not possess the internal table of user and role mappings used for authentication. Thus, the only way to log on to an arbiter with authorization active is to use the localhost exception.
Communication
The only communication between arbiters and other set members are: votes during elections, heartbeats, and configuration data. These exchanges are not encrypted.
However, if your MongoDB deployment uses TLS/SSL, MongoDB will encrypt
all communication between replica set members. See
Configure mongod and mongos for TLS/SSL for more information.
As with all MongoDB components, run arbiters in trusted network environments.
Example
For example, in the following replica set with 2 data-bearing members (the primary and a secondary), an arbiter allows the set to have an odd number of votes to break a tie: