A multi-gigabyte binary file with an incorrect content-type of text/plain can completely overload the fulltext indexing subsystem.
It's now possible to limit the size of the fulltext extracted from a document using the fieldSizeLimit attribute on the <fulltext> element of the <repository> configuration:
<extension target="org.nuxeo.ecm.core.storage.sql.RepositoryService" point="repository"> <repository ...> <indexing> <fulltext fieldSizeLimit="1000" ...> ... </fulltext> </indexing> </repository> </extension>
Or for DBS (where <fulltext> has no enclosing <indexing> element):
<extension target="org.nuxeo.ecm.core.storage.mongodb.MongoDBRepositoryService" point="repository"> <repository ...> <fulltext fieldSizeLimit="1000" ...> ... </fulltext> </repository> </extension>