0

I have different datasources that uploads different documents to Solr Sink. Now if two datasources sends a same name field with different data types (say integer & double) then indexing of second field fails because data type of first field is already added in managed-schema.

All I need is that both fields get indexed properly as they used to work in Solr 4.x versions .

Since field names come at runtime,please suggest a solution that would work for me. I suppose it needs a change in solrconfig.xml but couldnot find the required.

100rabh
  • 6,156
  • 5
  • 27
  • 41

1 Answers1

0

How was your Solr configured to work in 4.x? You can still do it exactly the same way in Solr 6.

On the other hand, schemaless feature will define the type mapping on the first time it sees the field. It has no way to know what will come in the future. That's also why all auto-definitions are multivalued.

However, if you want to deal with specific mapping of integer being too narrow, you can change the definition of the UpdateRequestProcessor chain that is actually doing the mapping. Just merge the mapping of integer/long/number into one final tdoubles type.

Alexandre Rafalovitch
  • 9,709
  • 1
  • 24
  • 27