I know this is old, but I caught it while looking for some help on
Spellcheck.
I see that you are copying a set of fields to the field named 'text' and
then coping 'text' to 'spell'. According to the documentation:
http://wiki.apache.org/solr/SchemaXml#Copy_Fields "no copy feeds into
another cop
Hi,
With the spellcheck.build=true, IMO solr will build the spellcheck
disctionnary at each request, so with the 29m documents solr can popup from
the server with some error like "I quit" :)
I would build the dictionnary once after data index creation, you may set
this option to the spell request
