Upmap Happy Hour: Practical Tips Keeping Ceph Data In Balance

​Join us for an Ask Me Anything on Ceph balancing and upmap-remapped the tool nearly every cluster relies on to minimize client impact during rebalances.

Adding capacity or OSDs to an existing Ceph cluster can trigger a "thundering herd" effect as Ceph starts redistributing data. upmap-remapped solves this by gradually distributing data across new OSDs in small, controlled batches instead of one massive rebalance.

There's no set agenda for this one bring your cluster and bring your questions. Dan will be on hand to help work through balancing issues and answer anything related to how balancing works in Ceph.

Whether you have misplaced objects on your cluster, or you just need a second pair of eyes on your balancer config bring whatever you've got.

Who should attend

• ​Ceph operators running clusters in production

• ​Anyone planning capacity expansions or OSD additions

• ​Engineers troubleshooting PG balance, misplaced objects, or rebalance performance issues

• ​Anyone curious about how the Ceph balancer and upmap really work under the hood