I was profiling one of the Search Application and I noticed when I re-index with option purgeallOnStart using Mass Indexer, it does delete entries from infinispan caches but sometimes it leaves more than 50% of cache entries still there (especially in data cache). For example if I have 1200 entries in data cache, it will leave 700+ there and next time when re-indexing finished I will have 2000 or so in cache.

Am I missing something....? I would expect purgeAll means no more required and simply do a fresh start whatever is being stored in backend cachestore...

Hi Hibernate Team,

Please explain this situation. I want to fully clear down indexes pro grammatically before starting rebuilding all. At the moment MassIndexer.purgeAllOnStart doesn't seem doing what I need. Many Thanks

Problem is in case of Infinispan if space is not reclaimed instantly, it means more memory used. I did profiling and 2GB indexes once needed 5GB ram once loaded (pre-loaded in infinispan) while re-indexing and it stayed 4+GB used and then I had to delete the indexes manually from cache store and redo it.

If it is all Lucene and filesystem -> SPACE belongs to disk space that you might not worry about.

I suspect infinispan cache store fro JDBC might have problems... but this is my guess.