New DS replica

Hi, If I have say 3 DS instances serving as cts stores in the repl topology. No separate repl server. Each DS is in its own ec2 and each ec2 has its own ebs volume. If I want to create a 4th DS instance in the running topology. is it ok to create snapshot of one of the running DS instances and use that to create the 4th instance. Do you see any issue with this approach…


There are many ways to accomplish your goal.
May I propose that the snapshot approach would not be best for a number of reasons… all of which can be corrected of course but it is more of an administrative effort.
The simplest and most straight forward approach is the proper installation of the new DS server using the appropriate profile, and adding the server to the topology at install time. And this way you will not have to enjoy the administrative burdens.

1 Like

Thanks, Guy. however what about the initialisation of the new instance with the existing data in the topology if the data is huge ?


Define “huge”.
But truly, there are only 2 options:
Off line or online initialization of the new instance.

it’s about 500k long-lived tokens…

500,000 objects is “small”.
In fact, so small, I would suggest an online initialization.


Thanks. So if let’s say 500 million entries then in that case initilise from the backup is better than using snaphopt ?

Now we seem to be going down the theoretical path.

I am not a proponent of the snapshot method in practice. The reason is that the data tier is stateful.
Moreover, there is administrative operations to be completed prior to the enablement of the service or you significantly risk data corruption and service interruption due to conflicts.

For reference, do see:

Please note, I state above “in practice”. Far too often I have witnessed issues due to snapshots as a necessary step was missed. The “initialization” path is fail proof and more expedient than you may think.