Folks,
Recently we are observing an issue with one of our web applications (using Lucene 1.4) maxing out on CPU under concurrent access of ~20 users doing a lucene query. At this point web application fails to serve any more pages and thread dump shows all lucene threads "waiting for monitor" access. Oddly this only happens on our production Solaris 10 boxes, with JDK 1.5. The load test passes with flying colors on a Windows XP box, even for 50+ concurrent access. We have tried many options all the way from patching Solaris to efficient disk access as well as various JVM arguments, but nothing worked for us. The issue is similar to what this thread presents. http://www.archivum.info/java-dev@lucene.apache.org/2005-07/msg00046.htm l Env. in questions is : OS - Solaris 10, dual core AMD 64 JVM - 1.5.0_06 Tomcat - 5.0.27 Lucene 1.4-final Any pointers on this would be greatly appreciated. Thanks much! Salman
