Database Replication - Addition of instances

Hey there,

I have around 100GB 3-node running as replication topology and want to add another node in the existing topology - What would be best plan for it?

Should I just add another node into cluster, will it copy all 100GB of data on this added node or need to copy initial data?

Many thnx

if your replication factor is 3, your total storage volume will be around 3 * your database size. If you just add one node, some data will be placed on it but the total storage will remain about the same.

Hi @stdranwl,

This startup and scaling training or this replication tutorial should help you understand how CockroachDB auto-replicates and auto-balances data.

Best,
Jesse

Thnx ik_zelf, I wanted to know if let say ignore the replication_factor (I want to have all data on every node), Will the process will copy initial data itself (rebalance that on 5 now) or we need to use kind of dump-restore for initial data likewise we do the initial sync on MySQL?

A very nice thing about CRDB is that it does al those things automagically. Just join the new node the cluster and it all happens for you. You probably never want to look to mysql replication again.

@stdranwl - Jesse’s tutorials above will walk you through a real life example on a local cluster. Ronald is right, that additional work is unnecessary. CRDB will manage it for you.

You might want to take a look at this thread as well: How to accelerate new nodes rebalance speed. By default, CRDB will rebalance data to a new node at 2MiB/s - you may want to test temporarily increasing that to ensure the rebalance completes at a reasonable pace.