Re: searches and updates concurrency problem

2005-06-06 Thread Daniel Naber
On Monday 06 June 2005 11:11, Stefano Buliani wrote: > My problem is that the index update procedure and the searches could run > simultaneously, and, if they do, they corrupt the index file. Search is a read-only thing, so why should it corrupt the index? Even having several writers at the same

Re: searches and updates concurrency problem

2005-06-06 Thread Aalap Parikh
Hi, As per my understanding of Lucene, I think concurrent search and update to an index should not corrupt the index, given that only a single index-modifying operation is executing at any point of time. So in short, you can have multiple search operations and not more than one index update (add a

Re: searches and updates concurrency problem

2005-06-06 Thread Thomas Plümpe
Thanks Stefano, well done. It's not for sure that the concurrently running updates and searches do corrupt the index (a full update that removes the files and rebuilds them can obviously do so), but I'm sure they'll get the point and will tell us how things like that are handled. Best, Thomas >

Re: searches and updates concurrency problem

2005-06-06 Thread Maik Schreiber
> My problem is that the index update procedure and the searches could run > simultaneously, and, if they do, they corrupt the index file. > Is there a way to let Lucene handle this concurrency automatically (like > stop the searches till the update is finished)? Lucene does not handle this by its