Zero priority replica set members

In some cases (for example, if we have multiple data centers), we will want some of the members to never be able to become a primary server.

In a scenario with multiple data center replications, we may have our primary data center with one primary and one secondary based in the UK, and a secondary server located in Russia. In this case, we don't want our Russia-based server to become primary, as it would incur latency on our application servers based in the UK. In this case, we will set up our Russia-based server with priority as 0.

Replica set members with priority as 0 also can't trigger elections. In all other aspects, they are identical to every other member in the replica set. To change the priority of a replica set member, we must first get the current replica set configuration by connecting (via the mongo shell) to the primary server:

> cfg = rs.conf()

This will provide the config document that contains the configuration for every member in our replica set. In the members sub-document, we can find the priority attribute, which we have to set to 0:

> cfg.members[2].priority = 0

Finally, we need to reconfigure the replica set with the updated configuration:

rs.reconfig(cfg)

Make sure that you have the same version of MongoDB running in every node, otherwise there may be unexpected behavior. Avoid reconfiguring the replica set cluster during high volume periods. Reconfiguring a replica set may force an election for a new primary, which will close all active connections, and may lead to a downtime of 10-30 seconds. Try to identify the lowest traffic time window to run maintenance operations like reconfiguration, and always have a recovery plan in case something goes wrong.

Table of Contents for Zero priority replica set members

Create new playlist

Sign In

Sign Up

Table of Contents for
Zero priority replica set members