Hi,

I'm trying to setup a clustered infinispan lucene index. My jgroupsSlaves starts up fine and puts data into the indexed cache using infinispans cache API. But when I startup the jgroupsMaster I get LockObtainedException for everything in the cache when replication starts. I don't know what I did wrong. Here's my setup.

I run 1 shared cachemanager that I registered in my Jboss JNDI tree. So that hibernate search can grab it. This way my indexed cache and lucene caches share the same jgroupschannel and cachemanager. Maybe this is wrong. But I didn't find documentation that said I couldn't do this.

I'm using infinispan 5.1.6.Final and hibernate search 4.1.1.Final

My indexed cache and hibernate search parameters on the master node: (I also tried with dist and indexLocalOnly=true)

<namedCache name="cache">
<clustering mode="repl">
<sync/>
</clustering>
<indexing enabled="true" indexLocalOnly="false">
<properties>
<property name="hibernate.search.default.​worker.backend" value="jgroupsMaster"/>
<property name="hibernate.search.default.directory_provider" value="infinispan"/>
<property name="hibernate.search.infinispan.cachemanager_jndiname" value="infinispanCache"/>
<property name="hibernate.search.lucene_version" value="LUCENE_35" />
<property name="hibernate.search.infinispan.locking_cachename" value ="LuceneIndexesLocking"/>
<property name="hibernate.search.infinispan.data_cachename" value ="LuceneIndexesData"/>
<property name="hibernate.search.infinispan.metadata_cachename" value ="LuceneIndexesMetadata"/>
</properties>
</indexing>
</namedCache>

Lucene caches:

<namedCache
name="LuceneIndexesMetadata">
<clustering
mode="replication">
<stateTransfer
fetchInMemoryState="true" />
<sync
replTimeout="25000" />
</clustering>
<invocationBatching enabled="true"/>
<transaction
transactionMode="TRANSACTIONAL"
useSynchronization="false"
syncCommitPhase="false"
syncRollbackPhase="false"/>
<locking
lockAcquisitionTimeout="20000"
writeSkewCheck="false"
concurrencyLevel="500"
useLockStriping="false" />
</namedCache>

<namedCache
name="LuceneIndexesData">
<clustering
mode="replication">
<stateTransfer
fetchInMemoryState="true" />
<sync
replTimeout="25000" />
</clustering>
<invocationBatching enabled="true"/>
<transaction
transactionMode="TRANSACTIONAL"
useSynchronization="false"
syncCommitPhase="false"
syncRollbackPhase="false"/>
<locking
lockAcquisitionTimeout="20000"
writeSkewCheck="false"
concurrencyLevel="500"
useLockStriping="false" />
</namedCache>

<namedCache
name="LuceneIndexesLocking">
<clustering
mode="replication">
<stateTransfer
fetchInMemoryState="true" />
<sync
replTimeout="25000" />
</clustering>
<invocationBatching enabled="true"/>
<transaction
transactionMode="TRANSACTIONAL"
useSynchronization="false"
syncCommitPhase="false"
syncRollbackPhase="false"/>
<locking
lockAcquisitionTimeout="20000"
writeSkewCheck="false"
concurrencyLevel="500"
useLockStriping="false" />
</namedCache>

Looking at hibernate search source it looks like the worker backend variables were changed to no longer start with hibernate.search

Environment.java

public static final String WORKER_SCOPE = "hibernate.search.worker.scope";
public static final String WORKER_PREFIX = "worker.";
public static final String WORKER_BACKEND = WORKER_PREFIX + "backend";
public static final String WORKER_EXECUTION = WORKER_PREFIX + "execution";

BackendFactory.java

String backend = properties.getProperty( Environment.WORKER_BACKEND );

I guess this means my jgroupsMaster properties are not read and my shared index is defaulting to lucene backend.

I'll try to change my properties and get back with results.