Hi, I'm planning to upgrade my system from SOLR1.2.1 to SOLR1.4.1 version.
We had done some lucene level optimizations on the SOLR slaves in the earlier system(1.2.1), like: 1. removed the synchronized block from the SegmentReader class's isDeleted() method 2. removed the synchronized block from the FSDirectory.FSIndexInput class's readInternal() method Since, in 1.4.1 we have better alternatives as with NIOFSDirectory & Read only index reader (by defauly used by SOLR1.4.1), so, we did not applied earlier changes with 1.4.1 version. Now when load testing with 1.4.1, my CPU usage goes as high as 100%. When i repeat the load test with my earlier setup(1.2.1) the CPU usage is below 50-55%. But the total throughput of the new(1.441) version is much higher than the older(1.2.1) I would need some help in minimizing the CPU load on the new system. Could possibly NIOFSDirectory attributes to high CPU? Is there a mechanism in 1.4.1 to use the SimpleFSDirectory implementation for searching(would this require full re-index)? Help will be appreciated :) Thanx Pravesh -- View this message in context: http://lucene.472066.n3.nabble.com/High-100-CPU-usage-with-SOLR-1-4-1-tp3068667p3068667.html Sent from the Solr - User mailing list archive at Nabble.com.
