[ https://issues.apache.org/jira/browse/SOLR-14987?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17245421#comment-17245421 ]
ASF subversion and git services commented on SOLR-14987: -------------------------------------------------------- Commit 8f769c6a82c6801a9c74b12131c1ac43ceb50827 in lucene-solr's branch refs/heads/branch_8x from Timothy Potter [ https://gitbox.apache.org/repos/asf?p=lucene-solr.git;h=8f769c6 ] SOLR-14987: Reuse HttpSolrClient per node vs. one per Solr core when using CloudSolrStream (#2128) > SolrStream ends up creating a new HttpSolrClient for every replica being > queried instead of reusing for the same node > --------------------------------------------------------------------------------------------------------------------- > > Key: SOLR-14987 > URL: https://issues.apache.org/jira/browse/SOLR-14987 > Project: Solr > Issue Type: Bug > Security Level: Public(Default Security Level. Issues are Public) > Components: streaming expressions > Reporter: Timothy Potter > Assignee: Timothy Potter > Priority: Major > Time Spent: 2h 10m > Remaining Estimate: 0h > > Looking into some streaming expression performance issues when there are many > collections with many shards being queried and I found that SolrStream's open > method creates a new \{{HttpSolrClient}} for every replica being queried. For > instance, in my test case, I have 10 collections with 100 shards each (rf=1) > and I get 1000 HttpSolrClient instances in my SolrClientCache. If I reuse > HttpSolrClient's per node hosting a replica (so 10 in my case), the query > time for my expression drops by half (not too mention the reduced allocation > load on the JVM). -- This message was sent by Atlassian Jira (v8.3.4#803005) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org