Re: Japanese mm parameter in Solr3.6.2 generated lots of results with big performance hit
Hi Jack, The requestHandler in solrconfig.xml is defined as, thanks for your response requestHandler name=search class=solr.SearchHandler default=true str name=q.opAND/str schema.xml: solrQueryParser defaultOperator=AND/ Thanks -- View this message in context: http://lucene.472066.n3.nabble.com/Japanese-mm-parameter-in-Solr3-6-2-generated-lots-of-results-with-big-performance-hit-tp4041200p4041557.html Sent from the Solr - User mailing list archive at Nabble.com.
Re: Japanese mm parameter in Solr3.6.2 generated lots of results with big performance hit
Hi Jack, Here is the query http://localhost:7983/cores/jp/select/?q=鹿児島%20若石%20龍玉女rows=22debugQuery=onechoParams=allwt=xml (using dismax parser). Where f1 is being boosted over f2 (qf=f1^10 f2^1) [f1: string f2: text_ja (copyField being copied from f1) int name=QTime259/int --- time is quite large result name=response numFound=187974 start=0 -- Is there a way to limit the number of outputs to a certain number from Solr to reduce the QTime str天龍サイト/str strお年玉/str strお年玉くじ/str strお年玉くじ当選番号/str strお年玉つき年賀はがき当選番号/str strお年玉はがき当選番号/str strお年玉プレゼント/str strお年玉付き年賀はがき/str strお年玉付き年賀はがき年賀状当選番号/str strお年玉付き年賀はがき当せん番号/str strお年玉付き年賀はがき当選番号/str strお年玉付き年賀はがき抽選結果/str strお年玉付き年賀ハガキ/str strお年玉付年賀はがき当選番号/str strお年玉年賀/str strお年玉年賀はがき/str strお年玉年賀はがき2012当せん番号/str strお年玉年賀はがき当選/str strお年玉年賀はがき当選番号/str strお年玉年賀はがき当選番号2012/str strお年玉年賀はがき抽選/str strお年玉年賀はがき発表/str str玉龍中学校/str str玉龍中合格発表/str str淡路島玉ねぎ/str strサッカー中学校/str str name=rawquerystring鹿児島若石龍玉女/str str name=querystring鹿児島若石龍玉女/str str name=parsedquery +DisjunctionMaxQuery(((f2:鹿児島 f2:若 f2:石 f2:龍 f2:玉 f2:女) | f1:鹿児島若石龍玉女^10.0)) () /str str name=parsedquery_toString +((f2:鹿児島 f2:若 f2:石 f2:龍 f2:玉 f2:女) | f1:鹿児島若石龍玉女^10.0) () /str lst name=explain str name=鹿児島玉龍サイト str name=鹿児島玉龍サイト.../str str name=鹿児島玉龍中学校.../str str name=鹿児島玉龍高校.../str str name=鹿児島市玉龍中学校.../str str name=鹿児島市立玉龍高校.../str str name=鹿児島玉龍中学制服.../str str name=鹿児島玉龍中学校倍率.../str str name=鹿児島玉龍中学校学費.../str str name=鹿児島玉龍中模擬.../str str name=鹿児島玉龍高校制服.../str str name=鹿児島玉龍高校合格.../str str name=鹿児島県立玉龍高校.../str str name=風水山水龍(玉れ龍)寿山石色置物.../str str name=s 鹿児島県立玉龍高校制服.../str str name=鹿児島市玉龍中学校制服.../str str name=鹿児島市玉龍中学校学費.../str str name=鹿児島市玉龍中学費.../str str name=鹿児島市玉龍高校周辺地図.../str str name=鹿児島市立玉龍中学校制服.../str str name=鹿児島市立玉龍中学校合格発表.../str str name=鹿児島市立玉龍中学校校長.../str str name=鹿児島市立玉龍中学過去問 Thanks. -- View this message in context: http://lucene.472066.n3.nabble.com/Japanese-mm-parameter-in-Solr3-6-2-generated-lots-of-results-with-big-performance-hit-tp4041200p4041303.html Sent from the Solr - User mailing list archive at Nabble.com.
Re: Japanese mm parameter in Solr3.6.2 generated lots of results with big performance hit
We also need to see the /select request handler from solrconfig.xml. For the text_ja field, the query is generating an OR of the individual unigrams of the query string. -- Jack Krupansky -Original Message- From: kirpakaro Sent: Tuesday, February 19, 2013 7:40 AM To: solr-user@lucene.apache.org Subject: Re: Japanese mm parameter in Solr3.6.2 generated lots of results with big performance hit Hi Jack, Here is the query http://localhost:7983/cores/jp/select/?q=鹿児島%20若石%20龍玉女rows=22debugQuery=onechoParams=allwt=xml (using dismax parser). Where f1 is being boosted over f2 (qf=f1^10 f2^1) [f1: string f2: text_ja (copyField being copied from f1) int name=QTime259/int --- time is quite large result name=response numFound=187974 start=0 -- Is there a way to limit the number of outputs to a certain number from Solr to reduce the QTime str天龍サイト/str strお年玉/str strお年玉くじ/str strお年玉くじ当選番号/str strお年玉つき年賀はがき当選番号/str strお年玉はがき当選番号/str strお年玉プレゼント/str strお年玉付き年賀はがき/str strお年玉付き年賀はがき年賀状当選番号/str strお年玉付き年賀はがき当せん番号/str strお年玉付き年賀はがき当選番号/str strお年玉付き年賀はがき抽選結果/str strお年玉付き年賀ハガキ/str strお年玉付年賀はがき当選番号/str strお年玉年賀/str strお年玉年賀はがき/str strお年玉年賀はがき2012当せん番号/str strお年玉年賀はがき当選/str strお年玉年賀はがき当選番号/str strお年玉年賀はがき当選番号2012/str strお年玉年賀はがき抽選/str strお年玉年賀はがき発表/str str玉龍中学校/str str玉龍中合格発表/str str淡路島玉ねぎ/str strサッカー中学校/str str name=rawquerystring鹿児島若石龍玉女/str str name=querystring鹿児島若石龍玉女/str str name=parsedquery +DisjunctionMaxQuery(((f2:鹿児島 f2:若 f2:石 f2:龍 f2:玉 f2:女) | f1:鹿児島若石龍玉女^10.0)) () /str str name=parsedquery_toString +((f2:鹿児島 f2:若 f2:石 f2:龍 f2:玉 f2:女) | f1:鹿児島若石龍玉女^10.0) () /str lst name=explain str name=鹿児島玉龍サイト str name=鹿児島玉龍サイト.../str str name=鹿児島玉龍中学校.../str str name=鹿児島玉龍高校.../str str name=鹿児島市玉龍中学校.../str str name=鹿児島市立玉龍高校.../str str name=鹿児島玉龍中学制服.../str str name=鹿児島玉龍中学校倍率.../str str name=鹿児島玉龍中学校学費.../str str name=鹿児島玉龍中模擬.../str str name=鹿児島玉龍高校制服.../str str name=鹿児島玉龍高校合格.../str str name=鹿児島県立玉龍高校.../str str name=風水山水龍(玉れ龍)寿山石色置物.../str str name=s 鹿児島県立玉龍高校制服.../str str name=鹿児島市玉龍中学校制服.../str str name=鹿児島市玉龍中学校学費.../str str name=鹿児島市玉龍中学費.../str str name=鹿児島市玉龍高校周辺地図.../str str name=鹿児島市立玉龍中学校制服.../str str name=鹿児島市立玉龍中学校合格発表.../str str name=鹿児島市立玉龍中学校校長.../str str name=鹿児島市立玉龍中学過去問 Thanks. -- View this message in context: http://lucene.472066.n3.nabble.com/Japanese-mm-parameter-in-Solr3-6-2-generated-lots-of-results-with-big-performance-hit-tp4041200p4041303.html Sent from the Solr - User mailing list archive at Nabble.com.
Japanese mm parameter in Solr3.6.2 generated lots of results with big performance hit
In Solr3.6.1 using text_ja field generated huge number of results, that degraded its performance significantly. The queries that were taking 15ms have gone up to 400ms and the other issue, it is not honoring rows parameter. The output results are not capped by the the number of documents requested using rows=100 parameter but lot more. Has anyone experienced this issue and what is the solution to improve the performance as putting the index into RAM and Cache did not have a significant impact on the performance. Thanks. -- View this message in context: http://lucene.472066.n3.nabble.com/Japanese-mm-parameter-in-Solr3-6-2-generated-lots-of-results-with-big-performance-hit-tp4041200.html Sent from the Solr - User mailing list archive at Nabble.com.
Re: Japanese mm parameter in Solr3.6.2 generated lots of results with big performance hit
Maybe you need to turn on autoGeneratePhraseQueries=true on your field type. And turn on debugQuery=true on your query to see what actually get generated. Show us a typical query - the rows parameter should always work, unless it's written wrong. -- Jack Krupansky -Original Message- From: kirpakaro Sent: Monday, February 18, 2013 2:39 PM To: solr-user@lucene.apache.org Subject: Japanese mm parameter in Solr3.6.2 generated lots of results with big performance hit In Solr3.6.1 using text_ja field generated huge number of results, that degraded its performance significantly. The queries that were taking 15ms have gone up to 400ms and the other issue, it is not honoring rows parameter. The output results are not capped by the the number of documents requested using rows=100 parameter but lot more. Has anyone experienced this issue and what is the solution to improve the performance as putting the index into RAM and Cache did not have a significant impact on the performance. Thanks. -- View this message in context: http://lucene.472066.n3.nabble.com/Japanese-mm-parameter-in-Solr3-6-2-generated-lots-of-results-with-big-performance-hit-tp4041200.html Sent from the Solr - User mailing list archive at Nabble.com.