It's possible to empty the solr index if there is any issue with the database

When a solr search update job is executed (like the one executed at startup) an iterator is used to iterator all the database documents. If one of the request to get the document from the database fail the iterator just log the error but don't not throw any exception. That means the job continue thinking it's done on database side and delete all the documents found on solr side to synchronize the index...

If there is any issue with DatabaseDocumentIterator the job should strop right away with an error.