Scalix

Starting from this version Lucene library decides write document into a file or keep it in memory. It’s better because in most cases it will write into file small portions of data and merge documents which gives lower IO usage and of cause guarantee that all documents will be stored in file. Before that it waited for a 200 documents and then tried to add for each operations called merge document which is not good. Because if I have 100 docs to insert into index files and restarts server ~90% of documents will be lost because after 20 sec tomcat will be forced killed and of cause SIS could not have time to flush everything. Call merge and optimize too often cause IO load. Lucene high load says dont call `optimize’ I will do it if I needed. Merge should be based not on document amount it should be based on memory size .