User Feedback: Shaping the Future of Ceph
summary
Interactive BoF collects concrete Ceph deployment feedback: debates over adopting Cephadm versus DIY orchestration, custom containers and secrets/Vault integration, and a claim of upgrading block clusters Pacific to Reef in under 90 minutes using in-house automation. Operational specifics: prefer ~100 PGs per OSD tradeoffs, PG split/merge can stall for days on slow HDDs, scrub metrics are missing, a proposed seph exporter / store exporter open source plan, and an upmap balancer script roadmap.
AI-written summary · about
chapters (8)
- 0:00Introduction and setup
- 1:27Cluster management pain points
- 8:23Orchestrators and automation tools
- 15:29PG splitting and performance
- 23:10OSD latency and upgrades
- 24:37Upmap balancer and scrubbing
- 30:00HDD cluster issues and PG splitting
- 33:14Next steps and developer AMA
related talks
11mKeynote: From Busy to Impactful: Maximizing Ceph’s Collective Resources
19mWrap: What To Build Next
53mPanel Discussion - Performance & Scalability
30mManaging Kubernetes: What You Need to Know About Day 2 - Craig Tracey, Heptio
01h 39mPostOps: A Non-Surgical Tale of Software, Fragility, and Reliability
55mPanel: Technical Oversight Committee/Governing Board Roundtable
10mBridget Kromhout (Pivotal)
40mPlatform Wars: The Battle Between Golden Paths and Spaghetti Pipelines