+1 for having a platform-wide clustering guide.

Thanks,
Sameera

On Sun, May 29, 2011 at 4:42 PM, Charitha Kankanamge <[email protected]>wrote:

> Folks,
>
> While configuring and testing product clusters in 3.2.0 release cycle, we
> came across several issues because of the multiple options we have with
> clustering setup configurations. At the basic level, we can configure a
> product cluster with mounted governance and configuration registries. Then,
> we can have multiple modes of distributed caching mechanisms associated with
> cluster. eg:- distributed, replicated, invalidation, local
> We also have Axis2 clustering (tribes) which can be configured through
> axis2.xml.  Deployment synchronizer also plays a important role in a product
> cluster from 3.2.0 release onwards.
>
> Now, with all these options, it will be extremely difficult for a user to
> figure out the most useful and best clustering configuration. Therefore, we
> should prepare a document or wiki with all these information.
> We came across errors such as
> "org.infinispan.util.concurrent.TimeoutException: Unable to acquire lock
> after [10 seconds] on key
> [org.wso2.carbon.caching.core.registry.RegistryCacheKey@ce8858d] for
> requestor [Thread[OOB-2,localhost-36388,5,Thread Pools]]! Lock held by
> [(another thread)]" when starting replicated caching enabled cluster nodes
> in parallel. As discussed with Senaka, we should start all nodes
> sequentially in such a cluster setup.
> We also came across several issues when running a cluster if the caching
> mode is set to "distributed". So, the recommended approach should be
> "replicated" mode. We also noticed that the cluster becomes inconsistent if
> the P2 features are not installed correctly in each node in a cluster.  Most
> importantly, it is difficult (or impossible) to figure out the root cause if
> an error occurs in an inconsistent cluster setup.
> We should let users know which configurations are replicated among cluster
> nodes and which are not. For example, carbon datasources, message stores,
> processors etc.. are not replicated among child nodes of a cluster.
>
> We should prepare a clustering best practices and guideline doc and ship or
> host in OT with 3.2.0 release.
>
> Associated JIRA:- https://wso2.org/jira/browse/CARBON-10487
>
> Regards
> Charitha
>
> _______________________________________________
> Carbon-dev mailing list
> [email protected]
> http://mail.wso2.org/cgi-bin/mailman/listinfo/carbon-dev
>
>


-- 
Sameera Jayasoma
Technical Lead and Product Manager, WSO2 Carbon

WSO2, Inc. (http://wso2.com)
email: [email protected]
blog: http://tech.jayasoma.org

Lean . Enterprise . Middleware
_______________________________________________
Carbon-dev mailing list
[email protected]
http://mail.wso2.org/cgi-bin/mailman/listinfo/carbon-dev

Reply via email to