Apache
/
lucene
mirror of https://github.com/apache/lucene.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

SOLR-9350: JSON Facet and method=stream: cacheDf threshold now gates use of the filter cache

This commit is contained in:
David Smiley 2016-08-08 10:36:39 -04:00
parent 153c270045
commit b63bb5167a
2 changed files with 11 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
solr/CHANGES.txt
View File

@ -205,6 +205,10 @@ Optimizations
* SOLR-9335: Solr cache/search/update stats counters now use LongAdder which are supposed to have higher throughput
under high contention. (Varun Thacker)
* SOLR-9350: JSON Facets: method="stream" will no longer always uses & populates the filter cache, likely
flushing it. 'cacheDf' can be configured to set a doc frequency threshold, now defaulting to 1/16th doc count.
Using -1 Disables use of the cache. (David Smiley, yonik)
Other Changes
----------------------

10
solr/core/src/java/org/apache/solr/search/facet/FacetField.java
View File

@ -839,9 +839,13 @@ class FacetFieldProcessorStream extends FacetFieldProcessor implements Closeable
createAccs(-1, 1);
// Minimum term docFreq in order to use the filterCache for that term.
int defaultMinDf = Math.max(fcontext.searcher.maxDoc() >> 4, 3); // (minimum of 3 is for test coverage purposes)
int minDfFilterCache = freq.cacheDf == 0 ? defaultMinDf : freq.cacheDf;
if (minDfFilterCache == -1) minDfFilterCache = Integer.MAX_VALUE; // -1 means never cache
if (freq.cacheDf == -1) { // -1 means never cache
minDfFilterCache = Integer.MAX_VALUE;
} else if (freq.cacheDf == 0) { // default; compute as fraction of maxDoc
minDfFilterCache = Math.max(fcontext.searcher.maxDoc() >> 4, 3); // (minimum of 3 is for test coverage purposes)
} else {
minDfFilterCache = freq.cacheDf;
}
docs = fcontext.base;
fastForRandomSet = null;