Our chunk store never recorded source domain, it got fl...

When source domain information is lost at the chunk level, it is often still available one level up in the document metadata. By leveraging the existing doc_id foreign key, you can perform a batch join to recover the missing domain information and update the index via a metadata-only upsert. This process is efficient and can be completed in minutes, whereas re-crawling or re-embedding the entire corpus is an unnecessary and costly operation.