3 - add network and cpu to all deployments
4 - define standard easy roll-out conf (use collectd.d for write_graphite, cephmetrics, cpu and network)
10 - add metrics starting message so you know collection is active
22 05/24 RGW: implement the latencies as different metrics to allow summarisation at the db layer
23 05/24 MON: add ceph health status (OK/WARN) to output dict
24 05/24 Dashboard: Add ceph health text
25 05/29 OSD: OSD metrics added, with dashboard updates