[pcp] PCP JMX PMDA

[Paul Smith]

Nathan has lots of good points here, but a one to expand on from my perspective:


> 
> As I'd hoped, it turned out Parfait does 99% of what we need and quite
> efficiently.  pmcd timeouts simply cannot happen, and the applications
> continue to export correct values under stop-the-world GC conditions -
> parfait-agent has alot of potential I think.
> 
> 

To be clear here, since Parfait uses the MMV model, PMCD can collect the last values that were set.  Some polling of JMX values done by Parfait would just be delayed by Java's Stop-the-World-Make-Coffee-and-perhaps-have-some-biscuits; you get values, likely 'stale'.  In the GC paused state, not much is happening in the JVM _anyway_ so whatever values are available aren't likely to change much (other than the GC values themselves..)

the main point is valid though, the PMCD is not blocked from reporting _some_ values back to the caller.

> 
> A big shout out to TallPaul Smith for all his Parfait help, without which
> parfait-agent would not have gotten this far!

Not able to help as much as I'd like!  was good to see you again, been too long.

Paul