[infinispan-issues] [JBoss JIRA] (ISPN-2871) All nodes are not replicated when eviction is enabled

[Chris Beer (JIRA)]

    [ https://issues.jboss.org/browse/ISPN-2871?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12757291#comment-12757291 ] 

Chris Beer commented on ISPN-2871:
----------------------------------

Mircea Markus
Sorry, I was using the terms from the clustered-cache infinispan-quickstart I modified to demonstrate the problem. In the clustered-cache replication quickstart, there are two "Nodes", Node0 and Node1. [1] As modified by me:

Node0 is a member of the cluster and spits out the cache.size() at 5 second intervals
Node1 is a member of the cluster and creates 100 cache entries, with keys named "key0" to "key99".

Both nodes load the same replication configuration [2].


I compile the clustered-cache project using:

$  mvn clean compile dependency:copy-dependencies -DstripVersion

I start Node0 with:

$ java -Dorg.infinispan.CacheDirPath=target/storage -cp target/classes:target/dependency/* org.infinispan.quickstart.clusteredcache.replication.Node0

and Node1 as:

$ java -Dorg.infinispan.CacheDirPath=target/storage2 -cp target/classes:target/dependency/* org.infinispan.quickstart.clusteredcache.replication.Node1

When eviction is disabled (by commenting it out),  in the Node0 console output (at logging level INFO), I see:

Feb 27, 2013 9:43:18 AM org.infinispan.remoting.transport.jgroups.JGroupsTransport start
INFO: ISPN000078: Starting JGroups Channel
Feb 27, 2013 9:43:21 AM org.infinispan.remoting.transport.jgroups.JGroupsTransport viewAccepted
INFO: ISPN000094: Received new cluster view: [localhost-31095|0] [localhost-31095]
Feb 27, 2013 9:43:21 AM org.infinispan.remoting.transport.jgroups.JGroupsTransport startJGroupsChannelIfNeeded
INFO: ISPN000079: Cache local address is localhost-31095, physical addresses are [127.0.0.1:7800]
Feb 27, 2013 9:43:21 AM org.infinispan.factories.GlobalComponentRegistry start
INFO: ISPN000128: Infinispan version: Infinispan 'Delirium' 5.2.1.Final
Feb 27, 2013 9:43:21 AM org.infinispan.transaction.lookup.GenericTransactionManagerLookup useDummyTM
WARN: ISPN000104: Falling back to DummyTransactionManager from Infinispan
Feb 27, 2013 9:43:21 AM org.infinispan.jmx.CacheJmxRegistration start
INFO: ISPN000031: MBeans were successfully registered to the platform MBean server.
Feb 27, 2013 9:43:21 AM org.infinispan.jmx.CacheJmxRegistration start
INFO: ISPN000031: MBeans were successfully registered to the platform MBean server.
Feb 27, 2013 9:43:24 AM org.infinispan.remoting.transport.jgroups.JGroupsTransport viewAccepted
INFO: ISPN000094: Received new cluster view: [localhost-31095|1] [localhost-31095, localhost-56291]
Feb 27, 2013 9:43:25 AM org.infinispan.quickstart.clusteredcache.util.ClusterValidation checkReplicationSeveralTimes
INFO: Cluster formed successfully!
Feb 27, 2013 9:43:25 AM org.infinispan.quickstart.clusteredcache.replication.Node0 run
INFO: Cache size: 0
Feb 27, 2013 9:43:26 AM org.infinispan.quickstart.clusteredcache.util.LoggingListener observeAdd
INFO: Cache entry with key key0 added in cache Cache 'Demo'@localhost-31095
[ SNIP key1 -> key98 ]
Feb 27, 2013 9:43:27 AM org.infinispan.quickstart.clusteredcache.util.LoggingListener observeAdd
INFO: Cache entry with key key99 added in cache Cache 'Demo'@localhost-31095
Feb 27, 2013 9:43:30 AM org.infinispan.quickstart.clusteredcache.replication.Node0 run
INFO: Cache size: 100


But when I enabled eviction in the infinispan configuration, I see:

Feb 27, 2013 9:40:41 AM org.infinispan.remoting.transport.jgroups.JGroupsTransport start
INFO: ISPN000078: Starting JGroups Channel
Feb 27, 2013 9:40:44 AM org.infinispan.remoting.transport.jgroups.JGroupsTransport viewAccepted
INFO: ISPN000094: Received new cluster view: [localhost-31370|0] [localhost-31370]
Feb 27, 2013 9:40:44 AM org.infinispan.remoting.transport.jgroups.JGroupsTransport startJGroupsChannelIfNeeded
INFO: ISPN000079: Cache local address is localhost-31370, physical addresses are [127.0.0.1:7800]
Feb 27, 2013 9:40:44 AM org.infinispan.factories.GlobalComponentRegistry start
INFO: ISPN000128: Infinispan version: Infinispan 'Delirium' 5.2.1.Final
Feb 27, 2013 9:40:44 AM org.infinispan.transaction.lookup.GenericTransactionManagerLookup useDummyTM
WARN: ISPN000104: Falling back to DummyTransactionManager from Infinispan
Feb 27, 2013 9:40:44 AM org.infinispan.jmx.CacheJmxRegistration start
INFO: ISPN000031: MBeans were successfully registered to the platform MBean server.
Feb 27, 2013 9:40:45 AM org.infinispan.jmx.CacheJmxRegistration start
INFO: ISPN000031: MBeans were successfully registered to the platform MBean server.
Feb 27, 2013 9:40:52 AM org.infinispan.remoting.transport.jgroups.JGroupsTransport viewAccepted
INFO: ISPN000094: Received new cluster view: [localhost-31370|1] [localhost-31370, localhost-24294]
Feb 27, 2013 9:40:55 AM org.infinispan.quickstart.clusteredcache.util.ClusterValidation checkReplicationSeveralTimes
INFO: Cluster formed successfully!
Feb 27, 2013 9:40:55 AM org.infinispan.quickstart.clusteredcache.replication.Node0 run
INFO: Cache size: 0
Feb 27, 2013 9:40:55 AM org.infinispan.quickstart.clusteredcache.util.LoggingListener observeAdd
INFO: Cache entry with key key0 added in cache Cache 'Demo'@localhost-31370
Feb 27, 2013 9:40:55 AM org.infinispan.quickstart.clusteredcache.util.LoggingListener observeAdd
[ SNIP key1 -> key98 ]
Feb 27, 2013 9:40:55 AM org.infinispan.quickstart.clusteredcache.util.LoggingListener observeAdd
INFO: Cache entry with key key99 added in cache Cache 'Demo'@localhost-31370
Feb 27, 2013 9:41:00 AM org.infinispan.quickstart.clusteredcache.replication.Node0 run
INFO: Cache size: 78





[1] https://github.com/cbeer/infinispan-quickstart/tree/replication-eviction-problem/clustered-cache/src/main/java/org/infinispan/quickstart/clusteredcache/replication

[2] https://github.com/cbeer/infinispan-quickstart/blob/replication-eviction-problem/clustered-cache/src/main/resources/infinispan-replication.xml
                
> All nodes are not replicated when eviction is enabled
> -----------------------------------------------------
>
>                 Key: ISPN-2871
>                 URL: https://issues.jboss.org/browse/ISPN-2871
>             Project: Infinispan
>          Issue Type: Bug
>          Components: Eviction
>    Affects Versions: 5.2.1.Final
>            Reporter: Chris Beer
>            Assignee: Mircea Markus
>            Priority: Blocker
>             Fix For: 5.3.0.Final
>
>
> When I enable replication and eviction, it appear that not all nodes are replicated to all hosts. This problem was discovered when clustering modeshape with eviction, and critical nodes were not being properly replicated.
> I've modified the clustered-cache quick-start to (hopefully) demonstrate this problem: 
> https://github.com/cbeer/infinispan-quickstart/tree/replication-eviction-problem/clustered-cache
> Node1 creates 100 cache entries (key0 -> key99). When eviction is disabled, the final cache size on Node0 is 100. When eviction is enabled, the final cache size is 78.
> This seems suspiciously similar to ISPN-2712.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira