On 3/8/2017 9:22 AM, Nicolas Bouillon wrote: > - I checked in the schema, all the fields that matter (Tika default > extracted fields) and my customer fields are stored=true. - I suppose > that the full-text index is not stored in a field? And
When you do a full text query, what field/fields are being searched? If these fields are a standard field (not a copyField destination) and not stored, then all that data will be lost when you do an atomic update, and then a document will no longer be found by a full text search after it is updated. The requirements for Atomic Update to avoid data loss are quite explicit. Look for the "Field Storage" section here: https://cwiki.apache.org/confluence/display/solr/Updating+Parts+of+Documents#UpdatingPartsofDocuments-AtomicUpdates Thanks, Shawn
