Hello list,

We've noticed quite huge strain on the filterCache in facet queries against
trigram fields (see schema in the end of this e-mail). The typical query
contains some keywords in the q parameter and boolean filter query on other
solr fields. It is also facet query, the facet field is of
type shingle_text_trigram (see schema) and facet.limit=500000.


Questions: are there some tools (except for solrmeter) and/or approaches to
monitor / profile the load on caches, which would help to derive better
tuning parameters?

Can you recommend checking config parameters of other components but caches?

BTW, this has become much faster compared to solr 1.4 where we had to a lot
of optimizations on schema level (e.g. by making a number of stored fields
non-stored)

Here are the relevant stats from admin (SOLR 3.4):

description: Concurrent LRU Cache(maxSize=10000, initialSize=10,
minSize=9000, acceptableSize=9500, cleanupThread=false)
stats: lookups : 93
hits : 90
hitratio : 0.96
inserts : 1
evictions : 0
size : 1
warmupTime : 0
cumulative_lookups : 93
cumulative_hits : 90
cumulative_hitratio : 0.96
cumulative_inserts : 1
cumulative_evictions : 0
item_shingleContent_trigram :
{field=shingleContent_trigram,memSize=326924381,tindexSize=4765394,time=222924,phase1=221106,nTerms=14827061,bigTerms=35,termInstances=114359167,uses=91}
name: filterCache
class: org.apache.solr.search.FastLRUCache
version: 1.0
description: Concurrent LRU Cache(maxSize=512, initialSize=512,
minSize=460, acceptableSize=486, cleanupThread=false)
stats: lookups : 1003486
hits : 2809
hitratio : 0.00
inserts : 1000694
evictions : 1000221
size : 473
warmupTime : 0
cumulative_lookups : 1003486
cumulative_hits : 2809
cumulative_hitratio : 0.00
cumulative_inserts : 1000694
cumulative_evictions : 1000221


schema excerpt:

<fieldType name="shingle_text_trigram" class="solr.TextField"
positionIncrementGap="100">
   <analyzer>
      <tokenizer class="solr.StandardTokenizerFactory"/>
      <filter class="solr.LowerCaseFilterFactory"/>
      <filter class="solr.ShingleFilterFactory" maxShingleSize="3"
outputUnigrams="true"/>
   </analyzer>
</fieldType>

-- 
Regards,

Dmitry Kan

Reply via email to