Bug 998645

+++ This bug was initially created as a clone of Bug #913672 +++

When https://bugzilla.redhat.com/show_bug.cgi?id=911117 is completed we should update our EAP6 plugin to take advantage

This is coming from https://issues.jboss.org/browse/PRODMGT-277

--- Additional comment from Charles Crouch on 2013-02-21 13:51:49 EST ---

Obviously this is only feasible in 3.2 if its already available in a released version of EAP6

--- Additional comment from Charles Crouch on 2013-02-21 13:55:18 EST ---

From RFE review meeting with larry and alan:

"Engineering will create a BZ to track this and potentially get it into 3.2 but we won't add this to the PRD as a requirement and the feature will be implemented if time permits (and EAP provides the capability)."

--- Additional comment from Charles Crouch on 2013-04-11 17:15:46 EDT ---

Dropping priority since based on Comment2 this is not a high priority feature. As I said there if we have time we'll look to address it.

--- Additional comment from Larry O'Leary on 2013-05-24 16:01:11 EDT ---

Looks like some of these metrics have now made it into JBoss EAP 6.1 but some data may be missing based on https://bugzilla.redhat.com/show_bug.cgi?id=911117#c8.

--- Additional comment from Lukas Krejci on 2013-06-06 07:21:35 EDT ---

EAP 6.1.0 contains the invocation statistics for individual methods on all kinds of enterprise beans.

Namely, it can report (on per-method basis):
"execution-time" - Time spend within this bean method.
"invocations" - Number of invocations processed.
"wait-time" - Time spend waiting to obtain an instance.

While investigating the metrics published by EAP, I found several that are missing from our plugin descriptor (even though they were present even in AS 7.1.1 times).

For all enterprise bean types, we do not track the following:
"peek-concurrent-invocations" - Peak concurrent invocations.
"invocations" - Number of invocations processed.
"wait-time" - Time spend waiting to obtain an instance.
"execution-time" - Time spend within a bean method.

Peek concurrent invocations seems to be like a good metric to collect (in a trendsup manner, so that we get the time derivative, which is the useful stuff). The rest of those seem to just be the aggregates of the per-method variants.

Not sure why we are not collecting these. Maybe they weren't working before, or we simply thought them not useful or just omitted them.

I will try to add them all and see what are the results with AS 7.1.1 and EAP 6.1.0.