Our server configuration is managed by Opscode's Chef. Our Chef setup is not redundant, it needs to move, and all of our other servers depend on it.
I planned to do a dry-run of the migration using an EBS snapshot. Unfortunately, before I could complete the dry-run, the old server died - just my luck. So I scrambled to get everything up and running.
Luckily, Simon Pither had already gone through moving a chef server to another machine. His post was fairly accurate. The only thing that was missing was the solr data that is used for things like `knife status`. To migrate that data, just add /var/cache/chef to the list of directories to copy.
With that blog post, and some quick typing, I was able to move the chef server before anything catastrophic happened.
EC2 security groups and VPC security groups are mutually exclusive. They need to be migrated as well. Because it's such a mind-numbing task to transfer them over one-by-one in the AWS console, and because the rules change during the transition, we had some downtime (<1 minute) due to incorrect security groups.
To avoid this issue, I would recommend writing a script that brings over the security groups programatically. The script may take a little longer, depending on the number of rules and groups, but it will be well worth it.
Instead of waiting for a future sprint to utilize the internal ELBs, I chose to use them right away. Already, I'm loving them. It's nice to not manage the scaling of my own haproxy servers.
If you intend to go through the same migration, I'd be happy to answer questions. Just leave a comment.