Unfortunately fluentd only seems to offer up metrics based on its output buffer statuses. So that's what we've got.
usage: fluentd-metrics --statsd=STATSD --cluster=CLUSTER [<flags>]
Flags:
--help Show context-sensitive help (also try --help-long and
--help-man).
--statsd=STATSD Host:Port of Datadog Statsd agent
--cluster=CLUSTER Name of kubernetes cluster
--fluent=http://127.0.0.1:24220
Fluentd HTTP API endpoint
--interval=10s Gap between metric probes