cluster mode local and transport protocol

[sachinsukhtankar]

The architecture we are using have a hotrod remote cluster in distribution mode with 2 copies of data and app war deployed in jboss app servers with cluster mode local, Some observations -

- Why jgroups transport protocol is running in local mode ?

- There seems to a bug in Statistics MBean evictions counter, it is always zero

- Is there any recomendations for garbage collection configuration.

- Also any plans for adding event processing and some way to get all key/keySet for remote cache

Thanks,

sachin

[vblagojevic]

Would you please elaborate your architecture setup - I can not quite follow you. You are running a cluster mode local and you are confused why transport is in local mode?

Cheers,

Vladimir

[sachinsukhtankar]

Here is the architecture setup,

-  3 node hotrod remote cache cluster

-  3 JBoss application servers where war is deployed

-  Web app wars has application cache  in cluster mode local and access to remote cache cluster. The logic is standard - Check first in application cache then remote cache if not get from source.

The question is why there is need of any transport in local cluster mode, Why jgroups at all runnning when cluster mode is local ?

Also any inputs on the other questions.

Thanks,

sachin

[galder.zamarreno]

Well, if JGroups is running, then you're probably using AS via the all configuration. You might still want http session replication which needs clustering and is provided by the app server.

For the statistics MBean eviction counter, you said that this is zero all the time. Maybe you can post your config to verify that eviction is correctly configured?