On Mar 30, 2016, at 12:00 PM, Jay Shaughnessy <jshaughn@redhat.com> wrote:



On 3/30/2016 10:50 AM, John Sanda wrote:

On Mar 30, 2016, at 10:23 AM, Matt Wringe <mwringe@redhat.com> wrote:

----- Original Message -----
From: "Jay Shaughnessy" <jshaughn@redhat.com>
To: hawkular-dev@lists.jboss.org
Sent: Monday, March 28, 2016 10:59:05 AM
Subject: Re: [Hawkular-dev] using hawkular wildfly agent as a custom java agent


I think we should limit ourselves to the Wildfly/EAP agent that we're already
working on, as EAP hosted app monitoring/mgmt is our bread and butter. Past

Except this assumption is not necessarily correct. Hawkular, at least Hawkular-Metrics, is not only being used just for Wildfly/EAP monitoring.

We are using Hawkular-Metrics in OpenShift (https://github.com/openshift/origin-metrics) and there are other potential integration points that are being looking into. If we only want to handle the Wildfly/EAP case, then we really need to take a good look and determine if these other integrations make sense or not. Otherwise we should really start to look beyond just Wildfly/EAP so that these integration can be handled better.

We are actively working on other integration efforts, so we definitely need to look beyond WildFly/EAP.

What I'm cautioning against is trying to build all feeds ourselves.  In RHQ we used a lot of manpower to build and maintain a lot of agent plugins.  There is goodness in providing interfaces and infrastructure that make integrating as easy as possible, and then trying to avoid doing all of the integration work as well.  This is the model that is working for miq. 

We have two things going on here, one is the team charter to provide middleware mgmt via manageiq, using hawkular as the provider.  We'll need agents/feeds to hook up to hawkular to report inventory and metrics which then make their way to miq.  We've got the WFly/EAP agent and that seems like the likely place to invest the most for MW mgmt.   Then we've got H Metrics, which to some degree  has a life of its own.  Certainly the integrations with metrics are important and as I said before, I think we're already doing the right things, and just like hawkular overall, making integration easy is our biggest win because it will allow others to build their own feeds.
_______________________________________________
hawkular-dev mailing list
hawkular-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/hawkular-dev

I agree that we are doing the right thing by providing APIs to make integration easier; however it common, reoccurring question is about clients that build on top of those APIs. Considering that these questions keep coming up, I think we need to try to provide a better answer beyond simply providing APIs. I do not think this necessarily means we need to invest a lot in building client libraries. There are plenty of client libraries/tools we can look at, and things like ptrans really help with the integration.