We observed that, since May 18th, you have different instances (from different IPs) that send regularly 2 different
forestadmin-schema to your Forest Admin Production environment. It leads to a situation where you have collections destroyed and re-created a few seconds later with a cleared UI configurations.
We suspect that you have deployed your Production admin backend on several machines and an old one is still alive and continues to synchronise an old schema regularly. Would that be possible?
If we cannot find out the instance that continue to send unexpected schema, an easy option is to generate a new environment secret for you Production instances it would automatically filter out old machines alive with the old environment secret.
Let us know.