MLT Performance Degraded Between 4.6.1 and 5.5.2 Solr

2016-10-31 Thread Ivan Provalov
I noticed a 3X performance degradation for MoreLikeThis between 4.6.1 and 
5.5.2.  Our configuration: 
   
where text_en is a pretty standard text fieldType.
Any pointers?
Thanks,
Ivan Provalov



Re: MLT Performance Degraded Between 4.6.1 and 5.5.2 Solr

2017-01-06 Thread Ivan Provalov
After some more digging, I narrowed it down to filtering.  Without any filters, 
the MLT is back to it's normal performance (8ms average response time for our 
case).  The issue goes away with 6.0 upgrade. 
The hot method is Lucene's DisiPriorityQueue downHeap(), which takes 5X more 
calls in 5.5.2 compared to 6.0.  I am guessing that some of the Solr filters 
refactoring fixed it for 6.0 release.  I am not sure which.  
As a work-around, for now I just refactored the custom MLT handler to convert 
the filters into boolean clauses, which takes care of the issue.  
Any insights into why this is happening in Solr 5.5.2?
 Our configuration:
1. mlt.maxqt=1002. There is an additional filter passed as a parameter3. 4. text_en is a pretty standard text 
fieldType.
Thanks,
Ivan
 

On Monday, October 31, 2016 5:10 PM, Ivan Provalov  
wrote:
 

 I noticed a 3X performance degradation for MoreLikeThis between 4.6.1 and 
5.5.2.  Our configuration: 
   
where text_en is a pretty standard text fieldType.
Any pointers?
Thanks,
Ivan Provalov