Previously in this chapter, we covered what monitoring should be done around your entire Ceph infrastructure and also looked at the new, builtin Ceph dashboard. To gain further insights into the operation of your Ceph cluster and associated infrastructure, a more detailed monitoring setup is required. Although alert monitoring is out of scope of this book, we will now look at capturing the Ceph performance metrics with collectd, storing them in Graphite, and then finally creating a dashboard with graphs using Grafana. These captured metrics can then be used in the following chapter to help tune your Ceph cluster.
We will build this monitoring infrastructure on one of our monitor nodes in our test cluster. In a production cluster, it is highly recommended that it gets its own dedicated server.