Why doesn't my Ceph cluster fully redistribute PGs after an OSD is marked down and out?
Issue
- Why doesn't my Ceph cluster fully redistribute PGs after an OSD is marked down and out?
- When an OSD goes offline or is marked down and out, it's been observed that most of the PGs/objects will redistribute appropriately to other OSDs but there are occasionally some Placement Groups (PGs) that remain "stuck" and/or degraded until the OSD is manually removed from the OSD map.
Environment
- Red Hat Ceph Storage
- Inktank Ceph Enterprise
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.