Ceph » Orchestrator

When the dashboard is deployed using cephadm, a monitoring stack (node_exporter, prometheus, alertmanager, grafana) can be applied to the cluster,
resulting in multiple components deployed according the provided spec definition.
However, even though we're able to push a specific json config through cephadm, we need to be able to specify the ip address on the network the
component belongs to, or the daemon will bind on *::<port>, which is a problem from a security point of view, preventing them to be deployed in
the OpenStack context (both in standalone and HA config).
The Ceph Dashboard component already provides (via the mgr) a way to support multiple instances on different ip addresses [1], but this is not
true for the other monitoring stack components.

This represents a gap compared to ceph-ansible [2]³⁴.

[1] https://docs.ceph.com/en/latest/mgr/dashboard/#host-name-and-port
[2] https://github.com/ceph/ceph-ansible/blob/master/roles/ceph-grafana/templates/grafana.ini.j2#L24
[3] https://github.com/ceph/ceph-ansible/blob/master/roles/ceph-prometheus/templates/prometheus.yml.j2
[4] https://github.com/ceph/ceph-ansible/blob/master/roles/ceph-prometheus/templates/alertmanager.yml.j2