I found at least one reason (reliably causes this problem for me):
Repro steps:
- source documents are JSON in an Azure Blob Storage
- during index field configuration, as a key field, the meatadata_storage_path is suggested by default.. If you don't accept this but
- expand the "content" field (so it shows all your own JSON fields)
- select e.g. "id" from your content fields as the key field
you will get stuck in "Validating..."
Other problems with this configuration:
- Cognitive Skills will not work, because you can only set them on the "metadata_*" fields and not on your content field
Probably there are more problems when importing JSON. You must also take great care to select "JSON" as the parsing mode in advanced settings on the last page of the wizard, otherwise (Keep Default. JSON mode created other problems for me. So your documents may be too long for Free or Basic SKU (32k and 64k, respectively).
My gut feeling is that this simple indexer mostly will work for plain txt or PDF documents, and not so much for structured JSON.