Friday, January 26, 2018

How to find recent down OSDs in a Ceph cluster

Why we need to know down OSD

There is no direct command to figure out which OSD went in a Ceph cluster. It is a persistent need in a large Ceph installation since OSD going down is a common event.
  • The way to find the down OSDs is by taking a look at the primary (leader) monitor log in the cluster.
  • In the log, search for 'down' keyword.

Tip

Once you have find the down OSD, the next step is to reahc the node that hosts the OSD.
$ sudo ceph osd find <osd id>