We will refer to the instances as source (4.x production) and target (DR or clone):
- Enable event base replication on all of your local repositories in the source instance to the target.
- Wait a day or trigger full replication manually, or determine the artifact count is within what you would consider acceptable limits. This will give the delta time to catch up as closely as possible before beginning the cutover.
- It's now time for the cutover. Disable "sync deletes" and "sync properties" on all replication configurations on the source. This will prevent missing files and properties on the source from influencing files on the target.
- Cut over the DNS from the source to the target. The result will be a small delta of difference, which may cause some builds to fail due to missing files, but all file uploads to the target will succeed and most operations should be fine.
- Manually trigger replication on all local repositories on the source. This will transfer any remaining files to the target while production traffic is pointed at the target. Eventually, replication on the source will stop. You can prove this by re-triggering replication and seeing something like "Completed local folder replication for repository-x/ with 0 deployed files, 0 deleted files, 0 properties change, 0 statistics change, 0 mkDirs... average events per second 0"