Can you delete a field from a document in Solr index?

Question

I have a big index, and during the indexation process there was an error. So to avoid reindexing which takes several days, I want to simply delete specific field and reindex. Is there any suggestion?

Raker

Answer 1

You cannot. The solution would be to get the document, store temporarily in the memory, delete it, update the required field (remove, add) and then add the document back to the index.

Aviva

Answer 2

If you're using Solr 4, you can use AtomicUpdate http://wiki.apache.org/solr/Atomic_Updates to remove a field much more easily. For example:

curl http://localhost:8983/solr/update?commit=true -H 'Content-type:application/json' --data-binary '[{"id": "630911fa-711a-3944-b1d2-cda6857f9827", "field_to_be_removed": {"set": null}}]'

Bis

How can I remove certain field for all "id"? For e.g. this removes a certain field for id:630911fa-711a-3944-b1d2-cda6857f9827. But I want this all the id's. – Sjoberg

Answer 3

you can do this if your rest of the fields are stored i.e stored="true ".as follows by setting null value.

<add>
  <doc>
    <!-- your unique key field -->
    <field name="employeeId">05991</field>  
    <!-- what ever field you want to delete -->
    <field name="skills" update="set" null="true" /> 
  </doc>
</add>

source:https://wiki.apache.org/solr/UpdateXmlMessages

Weigel

Answer 4

You can delete the indexed document by it's id. If you want to change the schema by removing a field, then yes, you would have to reindex.

Lorient

+1 or just avoid returning the field in the output format via fl parameter ;) – Hellion

Answer 5

You can delete all the index using the command delete and a query like this:

java -Ddata=args -Dcommit=yes -jar post.jar "<delete><query>*:*</query></delete>"

Using the argument -Dcommit force to update the index, so, be careful not deleting all documents when you dont want.

Lateshalatest

In the documentation of solr link explain how to delete one documents, but in the FAQ of solr explain deleting by query link – Lateshalatest

Answer 6

With solr 8.11.2,

curl 'http://localhost:8983/solr/solr/schema' -d '{"delete-field":{"name":"${field_name}"}}'

Ophthalmic

Answer 7

You can delete indexed value w.r.t. a field in Solr but not the field.

If you really want to delete a specific field while indexing then you should configure the field in schema.xml file before indexing document.

Kailyard