Best practices for disks on AWS

I try to deploy couchbase cluster (6.0.0 - community edition) on AWS and I’m wondering what is the best practice for disk configuration. According to guidelines from I should use EBS (eg. gp2) but what is the best practice if one VM crashes?
Should a new instance attach the existing EBS with data and join the couchbase cluster or should it attach newly created empty disk (the data will be replicated from other nodes)?
What are the best practices for that?

Thanks in advance