1 2 Previous Next 18 Replies Latest reply on Nov 10, 2012 5:11 PM by zeeman Go to original post
      • 15. Re: Infinispan with Hibernate and Hibernate Search
        mhunts46
        mhunts46 Oct 17, 2012 10:50 AM (in response to zeeman)

        Hey Zeeman,  Any luck?  I'm trying to get a Hibernate Search infinispan backed cluster up using JDBC_PING.  The config is giving me problems.  Wondering if have a working config?

          • Actions
        • 16. Re: Infinispan with Hibernate and Hibernate Search
          zeeman
          zeeman Oct 19, 2012 4:18 PM (in response to mhunts46)

          Hi Mike,

           

          No, I gave up on that. I use Infinispan in local standalone mode. Honestly, after testing it more, it's not as good as it's hyped to be.

          For hibernate search you next best option is use JMS backend. That's the only reliable one after local file system. Either way, let me know how things go.

            • Actions
          • 17. Re: Infinispan with Hibernate and Hibernate Search
            sannegrinovero
            sannegrinovero Nov 2, 2012 5:51 AM (in response to zeeman)

            Hi Zeeman,

            I would appreciate knowing what your problems are. If they are related to the CacheStore - stored indexed, I'm afraid I agree it's not good enough yet as I just had to open new issues after a new round of load testing. For while these get fixed by the Infinispan team we improved the JMS backend on Hibernate Search.

              • Actions
            • 18. Re: Infinispan with Hibernate and Hibernate Search
              zeeman
              zeeman Nov 10, 2012 5:11 PM (in response to sannegrinovero)

              Hi Sanne,

               

              Is it possible to use Hibernate Search (I'm on latest beta) with a Lucene directory that works with Amazon S3 without using Infinispan?

               

              I found https://github.com/tjake/Solandra which uses Solar + Casandra for distributed Lucene search. I have invested heavily in Hibernate search, but master/slave JMS/jGroups design does not work well in cloud and very hard to automate for deployments (scripting, and when an instance fails, it should be easy to recreate it).

               

              Which of below options do you recommend in terms of reliability, scalability, and automation?

               

              • Infinispan with cacheStore on file system + S3 jGroups Ping + Hibernate search
              • Hibernate Search + S3 directory impl,  only thing missing S3 does not offer locking, but latest wins policy is fair game in this case.
              • Solandra, not keen on it as I want to stick to HS
              A custom approach I came up with that allows each EC2 instance to have its own Lucene index, publish index updates/inserts to a DB or S3; Other instances check DB for table changes and update their own indexes. Table in DB will have inserted/updated entity Id and class, timestamp, and which node created it. With this info other nodes can update their indexes.

              What do you think?

              I could not find any examples of using HS + Infinispan  in cluster mode for the cloud.

               

               

                • Actions
              1 2 Previous Next