ft-reaper-improvements-final small npe fix

[michaelsembwever]

Avoid NPE when jmx is unreachable, when getting metrics for host

[adejanovski]

Moving this here won't work because if we cannot connect to one host through JMX (which will happen in multi region clusters) then Reaper won't look if metrics are already there in storage, which it does here : https://github.com/thelastpickle/cassandra-reaper/blob/ft-reaper-improvements-final/src/main/java/com/spotify/reaper/service/SegmentRunner.java#L417-L436

The error are only displayed in DEBUG so it shouldn't clutter the logs.
In which conditions did you get a NPE ?

[michaelsembwever]

hostProxy can be null (either context.jmxConnectionFactory.connect(hostName) throws an exception or returns null.

before getMetricsForHost(..) would throw a NPE that wasn't caught.
now it's caught, and subsequently gotMetricsForAllHosts is correctly set to false.

[adejanovski]

@michaelsembwever : I've reworked the patch a bit, using Optional<> to avoid the NPE.
Could you give me your feedback ?

Thanks

[michaelsembwever]

LGTM.

Personally, I would have avoided changing so many lines of code on a living feature-branch. But that's me.

Presuming you'll rebase the commits to one before merging.

[tlpsonarqube]

Remove this useless assignment to local variable "hostMetrics".

[tlpsonarqube]

SonarQube analysis reported 3 issues

• 3 major

Watch the comments in this conversation to review them.

1 extra issue

Note: The following issues were found on lines that were not modified in the pull request. Because these issues can't be reported as line comments, they are summarized here:

1. SegmentRunner.java#L148: Either re-interrupt this method or rethrow the "InterruptedException".

[tlpsonarqube]

Extract this nested try block into a separate method.