Question

I’m trying to make our application more resilient to downtime from our search server. To achieve this, I’m trying to disable the automatic indexing and removal, and manual perform them in a background process.

I’m trying

searchable :auto_index => false, :auto_remove => false do
...
end

But am still receiving a connection error, every time I try to save or destroy the record.

The backtrace shows that the error is coming from the solr searchable file, so it’s definetly sunspot_rails causing the problem. Here is a slice from the backtrace on destroy:

...
rsolr (1.0.8) lib/rsolr/client.rb:67:in `update'
rsolr (1.0.8) lib/rsolr/client.rb:121:in `delete_by_id'
sunspot (1.3.3) lib/sunspot/indexer.rb:36:in `remove'
sunspot (1.3.3) lib/sunspot/session.rb:137:in `block in remove'
sunspot (1.3.3) lib/sunspot/session.rb:136:in `each'
sunspot (1.3.3) lib/sunspot/session.rb:136:in `remove'
sunspot (1.3.3) lib/sunspot/session_proxy/abstract_session_proxy.rb:11:in `remove'
sunspot (1.3.3) lib/sunspot.rb:410:in `remove'
sunspot_rails (1.3.3) lib/sunspot/rails/searchable.rb:390:in `solr_remove_from_index'
sunspot_rails (1.3.3) lib/sunspot/rails/searchable.rb:93:in `block in searchable'
...

Any idea why the options to disable these hooks aren’t working? Any idea how to actually disable them?

Thanks

Answer 1

Found the answer to my own question:

Turns out that while automatic callbacks were disabled, I was still making it dirty in a different step. sunspot_rails has a cleanup step that runs at the end of a request and tries to commit all dirty records to the solr server. To disable it add auto_commit_after_request: false to your sunspot.yml.