Open
Description
Currently normalizer being present on a keyword field leads to the field using fallback synthetic source and therefore to a significant increase in storage space usage. The reason for that is that normalizer changes the original value of the field and resulting synthetic source will be normalizer as well. As a result it is not possible to perform a reindex using a different normalizer (it will double/triple/etc normalize).
That being said the conclusion in the team is that the price paid in storage is significant and may not always justify the potential problem. For some customers it can be completely acceptable to have normalized synthetic source and they are currently blocked.
See https://github.com/elastic/sdh-elasticsearch/issues/8807.