如何重命名 mongodb 副本集？

Question

I started up a few MongoDB replica sets with the default configuration and added them to Cloud Manager , but since the default replica set name is "rs0", they're hard to tell apart in the UI. Is there a way to change the name of a replica set, preferably without having to drop and re-import all of the data in the cluster?

I've tried using rs.reconfig ( link ) but it doesn't allow you to change the replica set name.

Answer 1

Yes! The process is pretty simple:

1. Start all of the nodes in the non-replicated mode
   • Stop mongod on each server
   • Start mongod back up. If you use /etc/mongod.conf , remove the replication section. If you don't, omit the --replSet option to mongod

2. Flush the local database where the replication set configuration is cached

   • On each server, open a mongo shell as the admin user and run use local; db.dropDatabase() use local; db.dropDatabase() (Make sure that admin user / root user has dbAdmin role on local db)

3. Start all of the nodes again in replicated mode

   • Stop mongod on each server
   • If you use /etc/mongod.conf , add the replication section back in with the new name and start mongod . If not, start mongod with --replSet <new-name>
4. Initialize the replica set
   • Open a mongo shell as the admin user on one of the nodes. (It will become the new primary)
   • Run rs.initiate() . DO NOT pass any arguments to rs.initiate() . (It'll fail with an error) Any other config you want to set can be changed using rs.reconfig() later.
   • On the same node where your ran rs.initiate() , for each secondary, run rs.add('[secondary.host.name]') to add it to the replica set.
   • Wait for the secondaries to come in sync

This doesn't require you to dump and re-import your data, and it can be done with minimal downtime (and a period of degraded performance as the secondaries sync) if you automate it.

Here's an ansible playbook that does the whole thing (assuming you're using /etc/mongod.conf and managing mongod via System V/Upstart/things that speak service .)

Answer 2

You could remove the existing replica set by following below steps and then can start the mongods with new replica set

1. start all mongods without --replSet

   mongod --dbpath data/rs3 --port 27019 --oplogSize 64
   Note: - Above command is for windows machine

2. start mongo for all the respective mongods remove the existing replica set use local
   db.system.replset.remove({_id:'OldReplicaSetName'})

3. start each mongod with the new replica set with --replSet ex - mongod --dbpath data/rs3 --port 27019 --oplogSize 64 --replSet mySet Note: - Above command is for windows machine