I know there there are different deployment scenarios for metrics, but let’s consider the single WildFly instance. Assuming the rhq-metrics service (not the whole rhq-metrics server) is deploy in that same WildFly instance, then the required dependencies for rhq-metrics would be minimal. We certainly would not need a full EE stack. I do not even see why we would need the REST endpoints.

As for domain mode, I think we need to consider whether data is being persisted to Cassandra or being stored in memory. If/when we are only storing data in memory, we might only support a subset of the full functionality. This has not been fully thought out. I just wanted to mention it. When Cassandra is being used, we could use it for coordination instead of introducing yet another external dependency.

On Jan 9, 2015, at 7:21 AM, Heiko Braun <ike.braun@googlemail.com> wrote:



well, that’s actually the whole point of this discussion: integrate all necessary rhq components into wildfly to give access to historical runtime data. 


On 09 Jan 2015, at 12:41, Heiko W.Rupp <hrupp@redhat.com> wrote:

I do not think that the whole rhq-metrics server would be deployed (incl C* dependency) just to allow to gather some
metrics and push them around.

_______________________________________________
rhq-devel mailing list
rhq-devel@lists.fedorahosted.org
https://lists.fedorahosted.org/mailman/listinfo/rhq-devel