Ceph osd deletion

1. Introduction

In general, osd does not need to delete, but there is nothing absolute; in many cases, deletion is still needed (for example, if you re-adjust the scheme of a deployed storage cluster, you need to delete and re-adjust osd)

2. Out osd## will set osd out # # reweight to 0, and the cluster will migrate data # # equivalent to a "ceph osd reweight" operation # # when "reweight" is 0, the osd node is not included in the "in" count # ceph osd out {osd_id} 3, stop osd## closes the daemon of the osd node # # before shutting down the daemon Need to wait for the end of data migration after out # # the daemon of this node is not included in the "up" count # systemctl stop ceph-osd@ {osd_id} .service4, remove osd## remove osd## weight becomes 0 Then change the weight of crush algorithm # ceph osd crush remove osd. {osd_id} 5, aut del osd## aut del osd## delete osd node related permission information # ceph auth del osd. {osd_id} 6, rm osd## rm osd## delete osd node id number # ceph osd rm {osd_id}