Off the top of my head ... Can you tell if GC is happening more frequently than usual/expected ?
Is the index optimized - if not, how many segments ? It's possible that one of the shards is behind a flaky network connection. Is the 10s performance just for the Solr query or wallclock time at the browser ? You can monitor cache statistics from the admin console 'statistics' page Are you seeing anything untoward in the solr logs ? -Simon On Wed, Aug 10, 2011 at 1:11 PM, Charles-Andre Martin <charles-andre.mar...@sunmedia.ca> wrote: > Hi, > > > > I've noticed poor performance for my solr queries in the past few days. > > > > Queries of that type : > > > > http://server:5000/solr/select?q=story_search_field_en:(water boston) OR > story_search_field_fr:(water boston)&rows=350&start=0&sort=r_modify_date > desc&shards=shard1:5001/solr,shard2:5002/solr&fq=type:(cch_story OR > cch_published_story) > > > > Are slow (more than 10 seconds). > > > > I would like to know if someone knows how I could investigate the problem ? I > tried to specify the parameters &debugQuery=on&explainOther=on but this > doesn't help much. > > > > I also monitored the shards log. Sometimes, there is broken pipe in the > shards logs. > > > > Also, is there a way I could monitor the cache statistics ? > > > > For your information, every shards master and slaves computers have enough > RAM and disk space. > > > > > > Charles-André Martin > > > > > >
