A Welcoming Ceph Experience Improvements for First-Time Users or: How to Make Beginners Love Ceph
summary
Deploys and ops notes: roughly 75 small clusters totalling ~225 nodes (≈5,400 cores, 58 TB RAM, 1.8 PB NVMe, ~900 OSDs) and consistent three-node hyperconverged Proxmox+Ceph setups with four OSDs per node that hit ~7 GB/s rados bench (4 MB). Lists frequent failures: NVMe firmware bugs and a 95% OSD usage event that made a cluster unavailable, and stresses must-know items (monitors vs managers, PGs, CRUSH, safe disk replacement, size/min size) plus need for a beginner-focused visual guide and quick recovery checks.
AI-written summary · about
chapters (9)
- 0:00Introduction to Ceph improvements
- 2:24Common beginner struggles with Ceph
- 9:12Key topics for first-time users
- 13:28Ideas for better beginner guides
- 17:12Q&A on customer workloads and support
- 19:49Proxmox and Ceph hyperconverged setups
- 22:19UI deployment and hardware considerations
- 24:47Selling Ceph over ZFS replication
- 26:41Day two operations pain points
related talks
40mJuggling Petabytes: Managing Ceph at Scale with Ceph-ansible
17mKeynote: Ceph Journey - A Perspective
16mTips and Tricks for Managing and Administering Ceph Clusters
36m200 Clusters vs 1 Admin
05mHealthier Ceph Clusters with Ceph-medic
44mCRUSH-ing the OSD Variance Problem
44mContainer Management at Scale
28mCeph Performance: The Basics of Tentacle Racing