OSD Scale Up without heavy re-balancing
Issue
-
Our scale up documentation suggests increasing CephStorage node count and letting director call ceph-ansible to add the OSDs. When ceph-ansible adds OSDs it waits for them to be "in" before the playbooks can complete.
-
When OSDs are marked "in" there is a data rebalance which uses Ceph cluster resources. The larger the amount of OSDs added, relative to the total number of OSDs, the more data is re-balanced.
-
How can we add new OSDs but delay the data re-balancing so that we can manually execute it in smaller increments during limited maintenance windows?
Environment
- Red Hat OpenStack Platform 13.0 (RHOSP)
- Red Hat Ceph Storage 3.y
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.
