writing to replicated influxdb instances

[donaldleckie]

Hi Florian,
In our configuration, our monitoring is replicated across 2 physical datacentres with replica InfluxDB instances. Our regular application nodes emit metrics to both InfluxDBs simultaneously.
I'd like to have our kafka cluster also report to both InfluxDBs as well - would you be happy to receive a pull request which implements this for everyone's benefit?

Separately, because our kafka nodes are updated/restarted via a rolling upgrade, the current implementation has each node recording data in the InfluxDB with different offsets - again, would you be OK to receive a pull request which aligns them to the reporting period? For example, reporting every 60 seconds would have reports created at the top of each minute.
One SNAFU on this topic is that the InfluxReporter's parent is an AbstractPollingReporter which unfortunately has a built-in time drift because it uses the ScheduledExecutorService.scheduleWithFixedDelay rather than scheduleWithFixedPeriod, which is probably what it meant to use (I guess this is a bug in already-old code). This should be OK, just means that the InfluxReporter has to extend AbstractReporter instead.

Appreciate your thoughts.
Donald