Starting from the Luminous version, related functions based on measurement and statistical properties have been separated, which ceph-mgr is responsible for. Simply put, versions after Luminous need to install ceph-mgr and enable it in order to execute ceph df, ceph pg dump and other instructions related to measurement and statistics in CEPH.
Ceph-mgr also provides the rest API, which supports prometheus and zabbix. The former is able to receive performance statistics, the latter can get events to issue warnings or tips. Another change is that the ceph-mgr function contains a Dashboard, which is not set to start as default. You will need to add the parameters of the dashboard to the configuration file to enable it.
The following is a brief introduction to the types of information currently available on Dashboard:
Cluster health: The health status of the cluster, including the current usage, the total number of objects, the number and status of Mon, OSD, and Mgr, the status usage, and simple I/O information of all pools, logs of the cluster, etc.
Cluster: It is divided into two sub-items of the servers and the osd. The servers display the name of the host in the clusters a unit, what services are listed above, and the ceph version on the server. The Osd shows the location of the osd, the status, the number of pages on the osd, usage, and the I/O information related to reading and writing.
Cluster: Provides information related to RBD, such as the status of the current mirror and sync when RBD mirroring is established. It also shows which pools have rbd images created below and provides a list of information about the rbd images.
Filesystems: Provides information about cephfs, including the name of the pools as cephfs, usage status, current inodes and requests.
It should be noted that if ceph-mgr daemon stops executing, it does not affect the file transfer, but related instructions cannot be executed normally, and will be stuck at the time of execution, so this should be considered when deploying. After, go in and deploy your ceph-mgrs.