Andrey Klochkov wrote:
Even if you point multiple embedded solr servers to the same index, you
should write with only one. Once you do a commit on the writer, you'd need
But it's what lucene index locking is for, isn't it? Locking should handle
that issue. Solr doesn't use index locking in a proper way?
Lucene's lock file is used more as a fail safe to keep you from shooting
yourself in the foot than a feature. Best practice is to use one
IndexWriter on an index at a time. That said, not much stopping you from
using it as a feature ;) So I am not sure why you ran into that
exception - what type of drive is z? If its NFS, I think you want to try
the native lock factory (if I remember right - NFS needs the right lock
impl, and even then its not recommended).
What Shalin said doesn't imply Solr is not using locks right - he was
just pointing out that if you add documents to one embedded Solr server,
it will reopen readers and see the changes - the other embedded Solr
server(s) will not get the message to reload. You would have to call
commit on all of them to get each to reopen a searcher.
--
- Mark
http://www.lucidimagination.com
