When fixing vhost startup bugs in a cluster I discovered that having the vhost module configuration in local_config was a source of trouble for configuration data that really was consistent across the cluster almost all the time.
For what it’s worth, the reasons for using local_config may have been driven by requirements and use cases I have no experience with. The use cases I do have experience with are however harder to solve with the current design so I’m trying a change.
I made some changes and committed them in preparation for my next step. A common default vhost config will help with dynamic host start/stop in a cluster. There are still a number of bugs to figure out before taking on dynamic vhosts so for now I’m laying a foundation.