Here's another agent PR that needs peer-review and merging.
PR:
https://github.com/hawkular/hawkular-agent/pull/150
JIRA:
https://issues.jboss.org/browse/HWKAGENT-57
The issue here is that when the agent is told to shutdown by the WildFly container
controller, the agent should set availability metrics to DOWN for local resources. So in
the shutdown code of the agent service, it now gets a list of all configured avail metrics
that need to be set to DOWN and sets them, making sure WF doesn't completely shutdown
until those metrics are set (needed to do some changes around async storage to the metrics
REST API - had to add a wait period just so the stop() method doesn't return until
after the metrics are set - otherwise, WF will shutdown potentially before all the REST
calls are made).
Anyway, this doesn't fully solve the issue because there are cases when the
agent's WF container dies abruptly (i.e. crashes). That's next on my todo - see
https://issues.jboss.org/browse/HAWKULAR-407