[jira] [Created] (SOLR-11756) FilteredQuery in 5.5.4 needs scorer loop much more than 4.4.7

["Yan Zhao (JIRA)" <ji...@apache.org>]

Yan Zhao created SOLR-11756:
-------------------------------

             Summary: FilteredQuery in 5.5.4 needs scorer loop much more than 4.4.7
                 Key: SOLR-11756
                 URL: https://issues.apache.org/jira/browse/SOLR-11756
             Project: Solr
          Issue Type: Bug
          Components: search
    Affects Versions: 5.5.4
            Reporter: Yan Zhao


In Solr 5.5.4, FilteredQuery no longer has the LeapFrogScorer as in Solr 4.7. And the whole query is rewritten into a BooleanQuery.  

As for query, if main query is like "name:zoo" which hits 10k docs in index but filterquery like "location:NewYork" helped to filter the resultset to 5 docs. 

And in final, the loop name:zoo set is maximum to 5 times as the filter is there. 
But in 5.5.4, we are seeing the name:zoo needs to loop 10k docs at first then join with the filter set.

In our environment, the filter doesn't changed too much so FilterQuery cache can helped in performance. But keyword might changes, so the whole QueryResult cache is not helping in this case.

This significantly increase the CPU cost for keyword search since we ran the profiling of the same query for these two versions.

Is it possible to have the LeapFrogScorer back?



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org